quarto trim size: 174mm × 240mm

Man

age

me

nt D

ec

ision

Vo

lum

e 5

6 N

um

be

r 10 2

018

em

erald

pu

blish

ing

co

m

wwwemeraldinsightcomloimd

Volume 56 Number 10 2018ISSN 0025-1747

Volume 56 Number 10 2018

Management Decision

Management Decision

Quarto trim size 174mm times 240mm

Number 10

Evidence-based management for performance improvement in healthcareGuest Editors Davide Aloini Lorella Cannavacciuolo Simone Gitto Emanuele Lettieri Paolo Malighetti and Filippo Vistintin

ISBN 978-1-78973-015-9

Evidence-based management for performance

improvement in healthcare

Guest Editors Davide Aloini Lorella Cannavacciuolo Simone Gitto Emanuele Lettieri Paolo Malighetti

and Filippo Vistintin

2061 Editorial advisory board

2063 Guest editorial

2068 What evidence on evidence-based management in healthcareAfsaneh Roshanghalb Emanuele Lettieri Davide Aloini Lorella Cannavacciuolo Simone Gitto and Filippo Visintin

2085 Three perspectives on evidence-based management rank fit varietyPeter F Martelli and Tuna Cem Hayirli

2101 Conceptual modelling of the flow of frail elderly through acute-care hospitals an evidence-based management approachSilvia Bruzzi Paolo Landa Elena Tagravenfani and Angela Testi

2125 Application of evidence-based management to chronic disease healthcare a frameworkSaligrama Agnihothri and Raghav Agnihothri

2148 Configurations of factors affecting triage decision-making a fuzzy-set qualitative comparative analysisCristina Ponsiglione Adelaide Ippolito Simonetta Primario and Giuseppe Zollo

2172 Assessing the conformity to clinical guidelines in oncology an example for the multidisciplinary management of locally advanced colorectal cancer treatmentJacopo Lenkowicz Roberto Gatta Carlotta Masciocchi Calogero Casagrave Francesco Cellini Andrea Damiani Nicola Dinapoli and Vincenzo Valentini

2187 An integrated approach to evaluate the risk of adverse events in hospital sector from theory to practiceMiguel Angel Ortiz-Barrios Zulmeira Herrera-Fontalvo Javier Ruacutea-Muntildeoz Saimon Ojeda-Gutieacuterrez Fabio De Felice and Antonella Petrillo

2225 Cost drivers for managing dialysis facilities in a large chain in TaiwanChia-Ching Cho AnAn Chiu Shaio Yan Huang and Shuen-Zen Liu

2239 Measuring information exchange and brokerage capacity of healthcare teamsFrancesca Grippa John Bucuvalas Andrea Booth Evaline Alessandrini Andrea Fronzetti Colladon and Lisa M Wade

2252 Letrsquos play the patients music a new generation of performance measurement systems in healthcareSabina Nuti Guido Noto Federico Vola and Milena Vainieri

2273 Hospital unit understaffing and missed treatments primary evidenceAshley Y Metcalf Yong Wang and Marco Habermann

GUEST EDITORSDavide AloiniUniversity of Pisa ItalyLorella CannavacciuoloUniversita degli Studi di Napoli Federico II ItalySimone GittoUniversity of Udine ItalyEmanuele LettieriPolitecnico di Milano Dipartimento di Ingegneria Gestionale ItalyPaolo MalighettiUniversity of Bergamo ItalyFilippo VisintinUniversita degli Studi di Firenze ItalyEDITORSAndy AdcroftHead Surrey Business School UKE-mail aadcroftsurreyacukProfessor Patrick J MurphyDePaul University USAE-mail profpjmgmailcomASSOCIATE EDITORSK Kathy DhandaSacred Heart University USAJoao FerreiraUniversity of Beira Interior PortugalArne FlohUniversity of Surrey UKSimone GuerciniUniversity of Florence ItalyJay J JanneyUniversity of Dayton USAPawel KorzynskiHarvard University USA amp Kosminski University PolandFranz T LohrkeLouisiana State University USABrandon Randolph-SengTexas AampM University USAReza Farzipoor SaenIslamic Azad University IranSanjay Kumar SinghAbu Dhabi University United Arab EmiratesJames WilsonUniversity of Glasgow UKYenchun Jim WuNational Taiwan Normal University Taiwan

ISBN 978-1-78973-015-9ISSN 0025-1747copy 2018 Emerald Publishing Limited

Emerald Publishing LimitedHoward House Wagon Lane Bingley BD16 1WA United KingdomTel +44 (0) 1274 777700 Fax +44 (0) 1274 785201E-mail emeraldemeraldinsightcomFor more information about Emeraldrsquos regional offices please go to httpwwwemeraldgrouppublishingcomofficesCustomer helpdesk Tel +44 (0) 1274 785278 Fax +44 (0) 1274 785201E-mail supportemeraldinsightcomOrders subscription and missing claims enquiriesE-mail subscriptionsemeraldinsightcomTel +44 (0) 1274 777700 Fax +44 (0) 1274 785201

Missing issue claims will be fulfilled if claimed within six months of date of despatch Maximum of one claim per issueHard copy print backsets back volumes and back issues of volumes prior to the current and previous year can be ordered from Periodical Service Company Tel +1 518 537 4700 E-mail pscperiodicalscom For further information go to wwwperiodicalscomemeraldhtml

Reprints and permissions serviceFor reprint and permission options please see the abstract page of the specific article in question on the Emerald web site (wwwemeraldinsightcom) and then click on the ldquoReprints and permissionsrdquo link Or contactE-mail permissionsemeraldinsightcomThe Publisher and Editors cannot be held responsible for errors or any consequences arising from the use of information contained in this journal the views and opinions expressed do not necessarily reflect those of the Publisher and Editors neither does the publication of advertisements constitute any endorsement by the Publisher and Editors of the products advertised

No part of this journal may be reproduced stored in a retrieval system transmitted in any form or by any means electronic mechanical photocopying recording or otherwise without either the prior written permission of the publisher or a licence permitting restricted copying issued in the UK by The Copyright Licensing Agency and in the USA by The Copyright Clearance Center Any opinions expressed in the articles are those of the authors Whilst Emerald makes every effort to ensure the quality and accuracy of its content Emerald makes no representation implied or otherwise as to the articlesrsquo suitability and application and disclaims any warranties express or implied to their use

Emerald is a trading name of Emerald Publishing LimitedPrinted by CPI Group (UK) Ltd Croydon CR0 4YY

Certificate Number 1985ISO 14001

ISOQAR certified Management Systemawarded to Emerald for adherence to Environmental standard ISO 140012004

Management Decisionis indexed and abstracted inABI InformAcademic Research LibraryBook Review DigestBusiness International and Company Profile ASAPBusiness Periodicals IndexBusiness Source Alumni EditionCompleteGovernment EditionCorporate

Corporate PlusElitePremierCabellrsquos Directory of Publishing Opportunities in Management amp MarketingCorporate Resource NetCurrent AbstractsDIALOGDiscoveryEmerald Management ReviewsEuropean Business ASAPExpanded Academic ASAP Health Business EliteINSPECInternational Academic Research LibraryOCLCrsquos Electronic Collections OnlineProQuestPsychINFOResearch LibraryScopusSocial Science Citation Index (ISI)TOC Premier (EBSCO)

After reports about all the facts have reached their desks after all the advice has been offered all the opinions listened to after everything has been listed for the final plan the most talkative of all the experts is on the way back to the airport deciding what to tell the next client specialists have uttered their warnings researchers have thrown doubt on the accuracy of the data and the economic adviser while voicing no views about the cash flow knits his brow and purses his lips about the cash flow situation the manager alone has to do something about it all He or she is the person who has to get something doneReg Revans The ABC of Action Learning (new edition) Lemos and Crane 1998Management Decision aims to publish research and reflection on the theory practice and techniques and context of decisions taken in and about business and business research


Guidelines for authors can be found atwwwemeraldgrouppublishingcommdhtm

EDITORIAL ADVISORY BOARD

Gianpaolo AbatecolaTor Vergata University Italy

AbdullahJamia Hamdard India

Moid Uddin AhmadJaipuria Institute of Management ndash NoidaIndia

Jameela AlmahariUniversity of Bahrain Bahrain

Nezih AltayDePaul University USA

Levent AltinayOxford Brookes University UK

Helena AlvesUniversity of Beira Interior Portugal

Gilles ArnaudESCP Europe France

Kaveh AsiaeiIslamic Azad University Iran

Erin BassUniversity of Nebraska Omaha USA

Joshua BendicksonEast Carolina University USA

Jasmina Berbegal-MirabentUniversitat Internacional de CatalunyaSpain

Tejinder K BillingRohrer College of Business USA

William P BottomWashington University in St Louis USA

Rosie BoxerUniversity of Brighton UK

Maree BoyleGriffith Business School Australia

Alan BrownUniversity of Surrey UK

Richard J ButlerSUNY Empire State College USA

Rosa CaiazzaParthenope University of Naples Italy

Claus von CampenhausenCredit Agricole Germany

Almudena CanibanoESCP Europe Spain

Sean D CarrUniversity of Virginia USA

Anjali ChaudhryDominican University USA

Russell ClaytonSaint Leo University USA

Lori CookDePaul University USA

Susan CoombesVirginia Commonwealth University USA

Amy DavidKrannert School of Management PurdueUniversity USA

Jackie DeemKaplan University USA

Arman DehpanahIslamic Azad University Babol Iran

Manlio Del GiudiceUniversity of Rome (Link Campus) Italy

Emanuela DelbufaloEuropean University of Rome Italy

Policarpo DemattosNorth Carolina AampT State University USA

Rocky J DwyerCollege of Management amp TechnologyWalden University USA

Vasco EirizUniversity of Minho Portugal

B ElangoIllinois State University USA

Christina FernandezUniversity of Beira Interior Portugal

Jason FertigUniversity of Southern Indiana USA

Denise Lima FleckCoppead UFRJ Brazil

Jane GibsonNova Southeastern University USA

Stan GlaserFred Emery Institute Australia

Catherine Glee-VermandeIAE de LyonUniversity of LyonManagement School France

Monika GolonkaKozminski University Poland

Michele A GovekarOhio Northern University USA

Paul GrantBrighton Business School UK

Christian GronroosHanken School of Economics Finland

William D GuthNew York University USA

Heiko HaaseUniversity of Applied Sciences Jena Germany

Fredrik HacklinETH Zurich Switzerland

Usha CV HaleyWichita State University USA

Nell T HartleyRobert Morris University USA

Diana HechavarrıaUniversity of South Florida USA

Jay HeizerTexas Lutheran University USA

Steven HendersonSouthampton Business School UK

Andreas HinterhuberHinterhuber amp Partners Austria

Richard C HoffmanSalisbury University USA

Brian HollandNational Workforce Development AgencyCayman Islands

Vered HolzmannHIT Tel-Aviv University Israel

Kun-Huang HuarngFeng Chia University Taiwan

Richard HuntColorado School of Mines USA

Adam JablonskiUniversity of Dabrowa Gornicza Poland

Corinne JenniAlliant International University USA

Colin JonesQueensland University of Technology UK

Rosalind JonesBirmingham Business School UK

Jay KandampullyOhio State University USA

Priya Kannan-NarasimhanUniversity of San Diego USA

Mohamad Amin KavianiIslamic Azad University Iran

Mohammad Saud KhanSchool of Management Victoria Universityof Wellington New Zealand

Jithendran KokkranikalUniversity of Greenwich UK

Artem KornetskyyUniversity of Customs and Finance Ukraine

Zoltan KrajcsakBudapest Business School Hungary

Olivia KyriakidouAthens University Greece

Giacomo LaffranchiniUniversity of La Verne USA

Patricia A LanierUniversity of Louisiana at Lafayette USA

Dominika LatusekKozminski University Poland

Helen LaVanDePaul University USA

Grace LemmonDePaul University USA

Gabriella LevantiUniversity of Palermo Italy

William S LightfootInternational University of Monaco Monaco

Eric LiguoriThe University of Tampa USA

Stephan M LiozuCase Western Reserve University USA

Xianghui LiuHuaqiao University Peoplersquos Republic ofChina

Ed LockeUniversity of Maryland USA

Nnamdi O MadichieCanadian University of Dubai United ArabEmirates

Francesca MagnoUniversita Degli Studi di Bergamo Italy

Pasquale Massimo PiconeDepartment of Management Economics andQuantitative Methods University ofBergamo Italy

Ricardo Mateo-DuenasUniversity of Navarra Spain

Catherine MatthewsBrighton Business School UK

Douglas McCabeGeorgetown University USA

Moriah MeyskensUniversity of San Diego USA

Daniel MillerNorth Carolina AampT State University USA

Albert MillsSaint Maryrsquos University Canada

Jean Helms MillsSaint Maryrsquos University Canada

Management DecisionVol 56 No 10 2018

pp 2061-2062r Emerald Publishing Limited

0025-1747

2061

Editorialadvisory

board

Quarto trim size 174mm x 240mm

Ivana MilosevicUniversity of Nebraska-Lincoln USA

Anna MinaKore University of Enna CittadellaUniversitaria Italy

Debmalya MukherjeeUniversity of Akron USA

Sidharth MuralidharanSouthern Methodist University USA

Peter A MurrayUniversity of Southern Queensland Australia

Sara NadinUniversity of Sheffield UK

Brian NagyBradley University USA

Ralitza NikolaevaLisbon University Institute Portugal

Tahir M NisarUniversity of Southampton UK

Donald NordbergBournemouth University UK

Florian NoseleitUniversity of Groningen The Netherlands

Anna NosellaUniversity of Padua Italy

Bill ldquoPatchrdquo PaczkowskiPalm Beach State College USA

Raktim PalJames Madison University USA

Daniel PalaciosTechnical University Valencia Spain

Stephanie S Pane-HadenTexas AampM University USA

Hamieda ParkerUniversity of Cape Town South Africa

Simon N ParryNewcastle University Business School UK

Giuseppe PedelientoUniversity of Bergamo Italy

Lew PerrenBrighton Business School UK

Robert PerronsQUT School of Management Australia

David PlansUniversity of Surrey UK

Shameen PrashanthamNottingham University Business SchoolPeoplersquos Republic of China

Pratheepkanth PuwanenthirenUniversity of Jaffna Sri Lanka Sri Lanka

Z George QiaoUniversity of Alabama at Birmingham USA

James L RairdonTexas AampM University USA

Mario RaposoUniversity of Beira Interior Portugal

Emmanuel B RauffletHEC Montreal Canada

Maija RenkoUniversity of Illinois-Chicago USA

Jason W RidgeUniversity of Arkansas USA

Alison RiepleUniversity of Westminster UK

Michael A RobertoBryant University USA

Foster B RobertsSoutheast Missouri State University USA

David F RobinsonIndiana State University USA

Fernando RoblesSchool of Business George WashingtonUniversity USA

Richard RoccoDePaul University USA

Carlos Rodeiro-VargasInstituto de Estudios Superiores deAdministracion Venezuela

Fabrizio RossiUniversity of Cassino and Southern LazioItaly

Matteo RossiUniversity of Sannio ndash Benevento Italy

Jennifer RowleyManchester Metropolitan University UK

Vivek RoyIndian Institute of Management RaipurGEC Campus India

Pasquale RuggieroUniversity of Siena Italy amp BrightonBusiness School UK

James C RyanUnited Arab Emirates University UnitedArab Emirates

Raiswa SahaSRM University India

Jose Marıa SallanUniversitat Politecnica de Catalunya ndashBarcelonaTech Spain

Joseph SarkisWorcester Polytechnic Institute USA

CM SashiFlorida Atlantic University USA

Ann L SaurbierWalsh College USA

Francesco SchiavoneUniversity of Naples Parthenope Italy

Timothy S SchoeneckerSouthern Illinois University EdwardsvilleUSA

Chad SeifriedLouisiana State University USA

Arash ShahinUniversity of Isfahan Iran

Gregory SheaWharton School University of PennsylvaniaUSA

Yusuf SidaniAmerican University of Beirut Lebanon

Aditya SimhaUniversity of Wisconsin Whitewater USA

Amrik SohalMonash University Australia

Pedro Soto-AcostaUniversity of Murcia Spain

Chester SpellRutgers School of Business Camden RutgersUniversity USA

Mary-Beth StanekGeneral Motors Belgium

Deryk StecUniversity of New Brunswick Saint JohnCanada

Pekka StenholmTurku Institute for Advanced StudiesFinland

Laixiang SunUniversity of London UK

Daniel J SvyantekAuburn University USA

Ian TaplinWake Forest University USA

Ugur UygurLoyola University Chicago USA

Gerwin van der LaanTilburg University The Netherlands

Joseph Van MatreUniversity of Alabama at Birmingham USA

Rogerio S VicterUniversity of Connecticut at StamfordDepartment of Management USA

Jose Enrique VilaUniversity of Valencia Spain

Dan WadhwaniUniversity of the Pacific USA

Richard WhiteSheffield Hallam University UK

Timothy M WickUniversity of Alabama at Birmingham USA

Richard Wilding OBECranfield School of Management CranfieldUK

Colin C WilliamsUniversity of Sheffield UK

Changyuan YanPNC Bank USA

Mohsen ZareinejadIslamic Azad University Tehran Iran

Lu ZengHuaqiao University Peoplersquos Republic ofChina

Lida ZhangUniversity of Macau Macau

Adrian ZicariESSEC Business School France

2062

MD5610

Guest editorial

Evidence-based management for performance improvement in healthcareThis special issue collects novel and relevant contributions that advance both thetheory and practice of evidence-based management (EBMgt) for performanceimprovement in healthcare All together the selected contributions shed new light onwhat we know so far about EBMgt in healthcare and they offer original insights to furtherthe ongoing debate

Although the term ldquoevidence-based managementrdquo (Pfeffer and Sutton 2006) is relativelynew and not yet consolidated the argument of informing management practice anddecisions through the systematic use of different sources of evidence is not novel Followingthe attention and popularity that evidence-based medicine (EBM) (Sackett et al 1996) hasreceived in healthcare over the last 20 years scholars in different disciplines haveprogressively focused their research efforts to extend what has been learned from EBM tomanagement (Arndt and Bigelow 2009) This ldquogold-rushrdquo has acquired momentum as aresult of the increasing availability of very large bodies of data In the specific context ofhealthcare not only have serious concerns about the actual sustainability of the healthcaresystems of the most developed countries reinforced the enthusiasm for EBMgt but also themanifested challenge of implementing any change that ldquocomes from the outsiderdquo in such aprofessional and knowledge-intensive socio-technical context In this view scholars ofdifferent disciplines such as strategy management organization theory and designoperations and innovation management public management and operational researchhave started an intense debate about how theories and practices about performanceimprovement developed thus far in productmanufacturing companies have to be re-thoughtand extended when applied to service professional and knowledge-intensive organizationssuch as hospitals (Wright et al 2016) EBMgt has thus emerged as the preferable approachthat connects many solutions that are currently under discussion

EBMgt asserts that managers should ground their judgment and practice on rationaltransparent and rigorous evidence that could help them explore and evaluate the pros andcons of alternatives and that they should be informed by relevant robust academicresearch and literature reviews (Tranfield et al 2003) Healthcare is among the sectorsthat might benefit more from such an approach Evidence emerges in healthcare as thekeystone for informing decision-making at all levels At the micro level evidenceshould solve frequent conflicts among physiciansrsquo different experiences and opinionsabout the most cost-effective and safe therapy for a group of patients At theorganizational level hospitals managers should look at evidence as legitimization ofthe adoption of innovative health technologies that prove to be cost-effective and safe inother organizations according to the well-established health technology assessmentparadigm Finally at the macro level policy-makers should invest in administrativehealth database research to extract evidence from their extensive and longitudinaldatabases to identify those strategies and initiatives that might work better and todevelop the so-called ldquoprecision policiesrdquo

Considering these three levels of analysis this special issue focuses theresearch attention on the use of EBMgt paradigm by physicians hospital managersand policy-makers to enable change and improvements along the whole supply andvalue chain of healthcare In doing so it reports scientific evidence regarding how thevarious actors of the healthcare ecosystem could and actually do make sense ofthe difference sources of evidence (eg clinical data administrative data laboratory and


pp 2063-2068copy Emerald Publishing Limited

0025-1747DOI 101108MD-10-2018-004

2063

Guest editorial


genetic data big data etc) and to what extent they subordinate their judgmentand experience to evidence

This special issue merges conceptual and empirical studies and it is aimed at influencingthe largest audience possible The first panel of manuscripts collects contributions thatare mostly conceptual on the role of EBMgt to support effective management practices anddecision-making in healthcare In this view they offer an overview of the literature andargumentation on the building dynamics of EBMgt

The first contribution by Roshanghalb et al (2018) presents a systematic literaturereview on EBMgt in healthcare Such a review classifies past studies accordingly toan original ldquoprocessrdquo perspective anchored on the inputndashprocessndashoutcomes modelMost notably the authors argue for the need to take a step ahead within the currentdebate on EBMgt through a more pragmatic approach that connects with a ldquogolden threadrdquofour main logical blocks They are groups of decision-makers (users of evidence) types ofmanagement practices or managerial decisions (outcomes) types of analysis and tools(processes) and sources of evidence (inputs) Their original systematization of past studiessheds light on relevant gaps that should be filled in through future research Moreoverpractitioners might take advantage of the ldquoprocessrdquo framework to consolidate and sharebest practices in terms of EBMgt

The second contribution by Martelli and Hayirli (2018) challenges the current debateon EBMgt by observing that scholars are entrapped into a sterile discussion aboutwhat ldquobest available evidencerdquo actually is and as a result that they are not able toadvance their theoretical arguments The authors claim that a possible ldquoway-outrdquo isoffered by the acknowledgment that the concept of ldquobest available evidencerdquo has three keydynamics ndash namely rank fit and variety ndash that coexist to crystallize what is the ldquobestrdquo setof evidence for a specific decisionpractice The first dynamic assumes that the evidencegenerated by certain processes ranks higher than the evidence that is generated fromother processes in supporting truth claims The second dynamic instead evaluatesldquobestnessrdquo according to the exactness of fit between a situation at a point in time and theevidence compiled for that situation Finally the third dynamic which is rooted inthe cybernetic theory assumes that the ldquobest available evidencerdquo can be generated byensuring that a broad range of knowledge types is elicited from and reconciled acrossindividuals The authors speculate that given the epistemic uncertainty and turbulencecharacterizing decision-making process in healthcare the ldquobest evidencerdquo is produced byvariety and not by rank or fit

The following two contributions therefore illustrate EBMgt-based conceptual proposalsfor improving healthcare service delivery

The contribution by Bruzzi et al (2018) proposes a novel conceptual model for managingfrail elderly patients in acute-care hospitals The model redesigns the flow of these chronicpatients and puts together organizational solutions that the literature considers effective interms of outcomes and costs The model assumes a patient-centered perspective andanalyses the main problems namely admission frail patient management and delayeddischarged hampering the patientsrsquo flow

The contribution by Agnihothri and Agnihothri (2018) develops a model for applyingEBMgt-based principles to chronic diseases The authors point out that a new theoreticalframework entitled ldquoInfluence model of chronic healthcarerdquo introduces the critical areaswhere managers can identify and evaluate relevant changes for improving patientoutcomes Their model can be used by hospital managers to determine the effectiveness oftheir decisions and strategies for improving healthcare quality

The remaining contributions are predominantly empirical and they offer acomprehensive overview on the use of EBMgt within specific healthcare processes bothclinical and administrativemanagerial

2064

MD5610

The contribution by Ippolito et al (2018) investigates EBMgt in the peculiarcontext of hospital triage through qualitative comparative analysis which is a novelmethod that has attracted enthusiasm among scholars of the social sciences The authorsinvestigated the interplay between individual and organizational factors in determiningthe emergence of errors with respect to different decisional situations They argue thatindividual and organizational factors are strictly interwoven and factors thatlead to the outcomes of the decision-making process are not homogenous As resultany intervention should emerge from an in-depth understanding of the organizationalcontext and the peculiarities of different typologies of decisions Additionallyinterventions must be aimed at fine-tuning the relationships between individualscontextual resources and constraints In so doing this study proposes a newcontingency-based perspective drawing on the theory of complex adaptive systems foridentifying the patterns of factors that determine the emergence of errors in triagedecision-making

The following contribution by Lenkowicz et al (2018) proposes a conformance checkingmethodology based on process mining to evaluate the adherence and efficiency of clinicalprocesses This research interprets the EBMgt paradigm within the assessment andevaluation of actual patient clinical pathways against established clinical guidelinesFinally the study coherently presents potential improvements for the evidence that hasbeen gathered While testing the methodology on advanced colon-rectal cancer treatmentpathways the work also offers an interesting real-case application which could inspireinterested practitioners to pursue similar initiatives

The contribution by Ortiz-Barrios et al (2018) deals with EBMgt with respect topatient risk assessment and proposes an integrated framework based on threedifferent multi-criteria methods analytic hierarchical process decision-making trial andevaluation laboratory and Vikor The authors tested their suggested approach in threehospitals in Colombia where they assessed the risk of potential adverse events inhospitalized patients and they discuss the key implications for both hospital managersand professionals

The contribution by Cho et al (2018) investigates cost determinants of dialysis facilitiesin Taiwan using multiple linear regression analysis They show that the costs of dialysistreatments are influenced by several managerial factors such as capacity resourceutilization rate and geographical location Their findings stimulate providers to considernew systems to control costs by increasing the operational efficiency Their analysis canhelp regulators of health systems worldwide to design the reimbursement rates for costaccounts dealing with dialysis

Next we have a group of contributors investigating the healthcare processes and relateddecision-making dynamics from an organizational perspective investigating resources andteams the role of performance measurement and management control systems andinformation systems

The contribution by Grippa et al (2018) investigates healthcare team interactions toredesign the care delivery model within a large US childrenrsquos hospital and to increase thevalue for health actors (patients families and employees) They apply a social networkmethodology and focus on communication flow among patients family members andhealthcare staff to measure knowledge flows communication behavior and the channels usedto interact This case study describes how the visualization and measurement of relationaldata can help the interdisciplinary healthcare teams identify patterns of interactions acrosshospital units and disciplines The authors show how it is possible to identify structuralproperties of healthcare teams to promote knowledge sharing and improve team performanceIn doing so the authors offer a strong contribution for practitioners on the value of adoptingsocial network-based methodology for organizational redesign

2065

Guest editorial

The following contribution by Nuti et al (2018) proposes a new generation ofperformance measurement systems (PMS) for the healthcare industry They emphasize thatpatient care processes increasingly involve multiple organizations and consequentlytraditional PMS considering a single organization are somewhat inadequate They presenta PMS which is graphically represented by a ldquostaverdquo whose focus is on a specific carepathway (eg the treatment of breast cancer) and it considers all organizations involved inthe pathway Such a PMS has already been adopted by 13 regional health systems in Italy

Finally the contribution by Metcalf et al (2018) examines the effects of understaffingin hospital-unit respiratory care and it evaluates the impact on error rates in the USAThey also investigate the moderating effects of teamwork and integrated informationsystems A higher rate of understaffing seems to be associated with more missedtreatments and both teamwork and integrated information systems seem to havea moderating role in avoiding errors

Davide AloiniDepartment of Energy Systems Territory and Construction Engineering

University of Pisa Pisa Italy

Lorella CannavacciuoloDepartment of Industrial Engineering Universita degli Studi di Napoli Federico II

Napoli Italy

Simone GittoPolytechnic Department of Engineering and Architecture University of Udine Udine Italy

Emanuele LettieriDipartimento Ingegneria Gestionale Politecnico di Milano Dipartimento di Ingegneria

Gestionale Milano Italy

Paolo MalighettiDepartment of Management Information and Production Engineering

University of Bergamo Dalmine Italy and

Filippo VisintinDepartment of Industrial Engineering Universita degli Studi di Firenze Firenze Italy

References

Agnihothri S and Agnihothri R (2018) ldquoApplication of evidence-based management to chronic diseasehealthcare a frameworkrdquoManagement Decision Vol 56 No 10 pp 2125-2147 available at httpsdoiorg101108MD-10-2017-1010

Arndt M and Bigelow B (2009) ldquoEvidence-based management in health care organizationsa cautionary noterdquo Health Care Management Review Vol 34 No 3 pp 206-213

Bruzzi S Landa P Tagravenfani E and Testi A (2018) ldquoConceptual modelling of the flow of frail elderlythrough acute-care hospitals an evidence-based management approachrdquoManagement DecisionVol 56 No 10 pp 2101-2124 available at httpsdoiorg101108MD-10-2017-0997

Cho C-C Chiu AA Huang SY and Liu S-Z (2018) ldquoCost drivers for managing dialysis facilities ina large chain in Taiwanrdquo Management Decision Vol 56 No 10 pp 2225-2238 available athttpsdoiorg101108MD-06-2017-0550

Grippa F Bucuvalas JB Andrea A Evaline FC and Andrea Lisa MW (2018) ldquoMeasuringinformation exchange and brokerage capacity of healthcare teamsrdquo Management DecisionVol 56 No 10 pp 2239-2251 available at httpsdoiorg101108MD-10-2017-1001

Ippolito A Ponsiglione C Primario S and Zollo G (2018) ldquoConfigurations of factors affecting triagedecision-making a fuzzy-set qualitative comparative analysisrdquo Management Decision Vol 56No 10 pp 2148-2171 available at httpsdoiorg101108MD-10-2017-0999

2066

MD5610

Lenkowicz J Gatta R Masciocchi C Casagrave C Cellini F Damiani A Dinapoli N and Valentini V(2018) ldquoAssessing the conformity to clinical guidelines in oncology an example for themultidisciplinary management of locally advanced colorectal cancer treatmentrdquo ManagementDecision Vol 56 No 10 pp 2172-2186 available at httpsdoiorg101108MD-09-2017-0906

Martelli P and Hayirli T (2018) ldquoThree perspectives on evidence-based management rank fitvarietyrdquoManagement Decision Vol 56 No 10 pp 2085-2100 available at httpsdoiorg101108MD-09-2017-0920

Metcalf AY Wang Y and Habermann M (2018) ldquoHospital unit understaffing and missedtreatments primary evidencerdquoManagement Decision Vol 56 No 10 pp 2273-2286 available athttpsdoiorg101108MD-09-2017-0908

Nuti S Noto G Vola F and Vainieri M (2018) ldquoLetrsquos play the patients music a new generation ofperformance measurement systems in healthcarerdquo Management Decision Vol 56 No 10pp 2252-2272 available at httpsdoiorg101108MD-09-2017-0907

Ortiz-Barrios MA Herrera-Fontalvo Z Ruacutea-Muntildeoz J Petrillo A and De Felice F (2018) ldquoAnintegrated approach to evaluate the risk of adverse events in hospital sector from theory topracticerdquoManagement Decision Vol 56 No 10 pp 2187-2224 available at httpsdoiorg101108MD-09-2017-0917

Pfeffer J and Sutton RI (2006) ldquoEvidence-based managementrdquo Harvard Business Review Vol 84No 1 p 62

Roshanghalb A Lettieri E Aloini D Cannavacciuolo L Gitto S and Visintin F (2018) ldquoWhatevidence on evidence-based management in healthcarerdquo Management Decision Vol 56 No 10pp 2068-2084 available at httpsdoiorg101108MD-10-2017-1022

Sackett DL Rosenberg WM Gray JM Haynes RB and Richardson WS (1996) ldquoEvidence basedmedicine what it is and what it isnrsquotrdquo British Medical Journal Vol 312 No 71

Tranfield D Denyer D and Smart P (2003) ldquoTowards a methodology for developing evidence-informed management knowledge by means of systematic reviewrdquo British Journal ofManagement Vol 14 No 3 pp 207-222

Wright AL Zammuto RF Liesch PW Middleton S Hibbert P Burke J and Brazil V (2016)ldquoEvidence-based management in practice opening up the decision process decision-maker andcontextrdquo British Journal of Management Vol 27 No 1 pp 161-178

About the Guest EditorsDavide Aloini PhD is Associate Professor of Business Process Management Informatics for Logisticsand Marketing at the Department of Energy Systems Land and Constructions Engineering at theUniversity of Pisa Italy His research interests include business process management andcollaborativeadvanced ICT solutions with special interest in large-scale project healthcare systemsand innovation in high-tech firms Specifically this includes process identification modeling analysisand improvement in complex healthcare systems and networks exploitation of big data potential inoperation management with a particular interest on marketing and CRM collaborative ICT platformenhancing open innovation He has published papers in international journals such as Information ampManagement European Journal of Operation Management Business Process Management JournalProduction Planning and Control Expert Systems with Applications and International Journal ofInnovation Management

Lorella Cannavacciuolo is Assistant Professor in Management Accounting and has a PhD Degreein Economic and Managerial Engineering Lorella Cannavacciuolo carries out her research activity atthe Department of Industrial Engineering of University of Naples Federico II Her research interestsencompass innovation network systems in SMEs process mapping and redesign networkmeasurements for large collaborative platforms activity accounting models for cost performancemanagement Her research topics are carried out mainly in the healthcare sector She has publishedpaper in international journals and she serves as Reviewer for many international journals in operationand healthcare management

Simone Gitto PhD is Associate Professor of Business and Management Engineering at Universityof Udine He was Assistant Professor at University of Rome Tor Vergata His main research interests

2067

Guest editorial

include air transport regulation health efficiency and forecasting methods His works have beenpublished in international refereed journals including British Journal of Management InternationalJournal of Production Economics Expert Systems with Applications Journal of Air TransportManagement Journal of Productivity Analysis Technological Forecasting and Social ChangeTelecommunications Policy Transportation Research Part E and Transportation Research Part A

Emanuele Lettieri is Full Professor at the Department of Management Economics and IndustrialEngineering (DIG) of Politecnico di Milano He chairs the Accounting Finance amp Control (AFC) andHealthcare Management (HCM) masterrsquos courses at Politecnico di Milano He is Director of theInternational MBA Part-Time program at MIP Politecnico di Milano Graduate School of Business Hisresearch interests are at the intersection among technology management and healthcare and deal withinnovation management and performance improvement in healthcare His current research works dealwith the development of evidence-based improvement strategies in hospitals through the use ofadministrative data the diffusion of digital innovation in healthcare with a particular interest to digitalservices to citizens apps and wearables the assessment of innovations in healthcare accordingly to thehealth technology assessment discipline the implementation of value-based strategies in healthcareHis research is both qualitative and quantitative He has conducted multidisciplinary research incollaboration with Universities research centers healthcare institutions and hospitals He hasparticipated in applied research large-scale European projects Finally he is continuously involved inthe education of healthcare professionals as well as healthcare companiesrsquo personnel with the design ofad-hoc classes

Paolo Malighetti is Associate Professor at the University of Bergamo He obtained PhD Degree inldquoEconomics and Management of Technologyrdquo with a dissertation thesis ldquoPost-deregulation patternsand competition issues in European medium size airportsrdquo He spent a research visiting period atDepartment of Air Transport Management Cranfield University Since 2007 he is Research Fellow atICCSAI Since 2014 he is Director of the HTH ndash Human factor and technology in healthcare a researchcenter co-founded by the University of Bergamo and Papa Giovanni XXIII Hospital As Director ofHTH he collaborates on several projects about the use of new technology supporting healthcaresystem and more broadly fostering wellbeing for older adults and chronic disease treatment

Filippo Visintin is Associate Professor of Service Management at the Department of IndustrialEngineering of the University of Florence Italy He is Scientific Director of the IBIS Lab and Co-founderand Co-owner of Smartoperations srl He regularly advises public and private healthcare organizationsHis research interests include servitization of manufacturing and healthcare operations managementHe is Author of several research papers published in journals such as European Journal of OperationalResearch Industrial Marketing Management International Journal of Production Economics Computersin Industry Flexible Service and Manufacturing Journal Journal of Intelligent Manufacturing ProductionPlanning and Control and IMA Journal of Management Mathematics

2068

MD5610

What evidence on evidence-basedmanagement in healthcare

Afsaneh Roshanghalb and Emanuele LettieriDepartment of Management Economics and Industrial Engineering

Politecnico di Milano Milan ItalyDavide Aloini

Department of Energy Systems Land and Constructions Engineering Universitagravedegli Studi di Pisa Pisa ItalyLorella Cannavacciuolo

Department of Industrial EngineeringUniversita degli Studi di Napoli Federico II Napoli Italy

Simone GittoPolytechnic Department of Engineering and Architecture Universitagrave di Udine

Udine Italy andFilippo Visintin

Department of Industrial EngineeringUniversita degli Studi di Firenze Firenze Italy

AbstractPurpose ndash This manuscript discusses the main findings gathered through a systematic literature reviewaimed at crystallizing the state of art about evidence-based management (EBMgt) in healthcare The purposeof this paper is to narrow the main gaps in current understanding about the linkage between sources ofevidence categories of analysis and kinds of managerial decisionsmanagement practices that differentgroups of decision-makers put in place In fact although EBMgt in healthcare has emerging as a fashionableresearch topic little is still known about its actual implementationDesignmethodologyapproach ndash Using the Scopus database as main source of evidence theauthors carried out a systematic literature review on EBMgt in healthcare Inclusion and exclusion criteriahave been crystallized and applied Only empirical journal articles and past reviews have been included toconsider only well-mature and robust studies A theoretical framework based on a ldquoprocessrdquo perspectivehas been designed on these building blocks inputs (sources of evidence) processestools (analyses on thesources of evidence) outcomes (the kind of the decision) and target users (decision-makers)Findings ndash Applying inclusionexclusion criteria 30 past studies were selected Of them ten studies werepast literature reviews conducted between 2009 and 2014 Their main focus was discussing the previousdefinitions for EBMgt in healthcare the main sources of evidence and their acceptance in hospitalsThe remaining studies (nfrac14 20 67 percent) were empirical among them the largest part (nfrac14 14 70 percent)was informed by quantitative methodologies The sources of evidence for EBMgt are published studies realworld evidence and expertsrsquo opinions Evidence is analyzed through literature reviews data analysis ofempirical studies workshops with experts Main kinds of decisions are performance assessment oforganization units staff performance assessment change management organizational knowledge transferand strategic planningOriginalityvalue ndash This study offers original insights on EBMgt in healthcare by adding to what weknow from previous studies a ldquoprocessrdquo perspective that connects sources of evidence types ofanalysis kinds of decisions and groups of decision-makers The main findings are useful foracademia as they consolidate what we know about EBMgt in healthcare and pave avenues for furtherresearch to consolidate this emerging discipline They are also useful for practitioners as hospitalmanagers who might be interested to design and implement EBMgt initiatives to improvehospital performanceKeywords Decision making Management Health care Systematic literature reviewEvidence-based practice Evidence-based managementPaper type Literature review



0025-1747DOI 101108MD-10-2017-1022

Received 19 October 2017Revised 29 July 2018

Accepted 31 July 2018

The current issue and full text archive of this journal is available on Emerald Insight atwwwemeraldinsightcom0025-1747htm

2069

Evidence onEBMgt inhealthcare


BackgroundEvidence-based management (EBMgt) concerns how to translate the best available scientificevidence into organizational practices avoiding decisions based on individual experienceand preference (Rousseau 2006 Walshe and Rundall 2001) This idea is strictly connectedto evidence-based medicine (EBM) the practice of ldquointegrating individual clinicalexpertise with the best available external clinical evidence from systematic researchrdquo(Sackett et al 1996) that has received increased attention over the past 20 yearsThe principles of this approach have been widely accepted for application in public healthalso for management and policy decisions (Walshe and Rundall 2001 Oliver et al 2004)

The ongoing debate on whether and how EBMgt practices should be developed andimplemented in healthcare has been reinforced by the increasing availability of massivedata sets from very heterogeneous sources coupled with an improved capacity to analyzethem (Hopp et al 2018) Scholars of healthcare management and decision management aswell as policy-makers and health professionals are investigating to what extent theconsolidating bodies of knowledge and practices about EBMgt are better informing andsupporting how managerial decisions are taken in healthcare echoing what has beenachieved in medicine through the EBM experience (Baba and HakemZadeh 2012 Reayet al 2009 Briner et al 2009 Kovner and Rundall 2006) With this respect also a cursoryreview of the extant literature would show that past studies on EBMgt dealt with a widespectrum of ldquoevidencerdquo sources Evidence used to inform decision-making ranged fromrobust scientific evidence (eg Veillard et al 2005 Hamlin et al 2011 Grundtvig et al 2011Francis-Smythe et al 2013 HakemZadeh and Baba 2016) to healthcare managersrsquo expertise(eg Briggs and McBeath 2009 Francis-Smythe et al 2013) from peer opinions(eg Schmalenberg et al 2005 Davies and Howell 2012 Fazaeli et al 2014) to local datasources (eg Hornby and Perera 2002 Hamlin 2002 Beglinger 2006 Willmer 2007) alsoconsidering patientsrsquo preferences (eg Marschall-Kehrel and Spinks 2011 Slater et al 2012)This variety of sources well reflect the variety of decisions and judgments that healthpractitioners (policy-makers hospital managers and health professionals such asphysicians nurses therapists etc) have to make day-to-day (Briner et al 2009) thatrequire different data and level of evidence When these different sources of evidence areused inappropriately poorer decisions are taken and poorer outcomes are achieved (Kovner2014) Like as in medicine robust scientific evidence should constitute the ldquobackbonerdquo forinforming decision-making (Aron 2015) however many decisions or managerial practicesmight require other sources of evidence whose level of robustness is lower With thisrespect Jaana et al (2014) claimed in their scoping review that past studies on EBMgtfocused to health professionals (physicians and nurses) as decision-makers overlookingother relevant groups of decision-makers (eg hospital managers policy-makers etc)In particular further light is still needed to understand how different groups of decision-makers in healthcare apply EBMgt to their daily managerial practice and decision-makingwith respect to the types of decisions the sources of evidence and their investigation Thisresearch direction would provide further elements to debate what Young (2002) called as theneed to create a ldquomanagement culturerdquo that in healthcare is still a priority In fact whilephysicians are getting used to ground their clinical decisions to the best available evidencehospital managers and policy makers are still far from this culture preferring personaljudgment and insights (Pfeffer and Sutton 2006 Walshe and Rundall 2001)

Against this background ndash and coherently to the research need pointed out above ndash thisstudy aims at shedding light on the state of art of EBMgt in healthcare from an originalangle Respect to past literature reviews on EBMgt in healthcare (eg Young 2002Jaana et al 2014 HakemZadeh and Baba 2016) this study will focus on the overlookedrelationship between managerial decisions and sources of evidence with specific referenceto different groups of decision-makers In this view this study will adopt a process

2070

MD5610

perspective that has been incorporated into a novel theoretical framework based on theinput-process-output (I-P-O) model (McGrath 1964) The I-P-O framework has been recentlytaken as theoretical anchor for other studies in the field of management (eg Simsek 2009Ghezzi et al 2017) because it can help to distinguish the main antecedents mechanisms andoutcomes of the process under investigation By taking this perspective we aim atshedding novel light on what is already known from past reviews A theoretical frameworkthat will connect groups of decision-makers with types of managerial decisions and withdifferent analyses to extract insights from source of evidence will be outlined as referencemap to understand what evidence we have so far about EBMgt in healthcare In this viewthis study aims at paving avenues for further research and thus focusing the attention ofscholars of healthcare management and decision management to areas of research that havenot been sufficiently investigated yet Additionally health professionals and managers willgather a comprehensive view of EBMgt in healthcare and a reference framework that mighthelp them designing and implementing evidence-based managerial practices

MethodsPast studies on EBMgt in healthcare have been identified and selected through a systematicapproach following the best practice of systematic literature reviews (Tranfield et al 2003)In the followings the search strategies that have been implemented how past contributionshave been selected and what data have been extracted to inform the literature review will bedetailed briefly

Search strategies and contributions identificationThe literature review was performed referring to Scopus as main source of past studiesThis database covers extensively social sciences journals and is commonly used asreference source for systematic literature reviews (eg Spender et al 2017 Ghezzi et al2017) To limit the potential risk of overlooking relevant contributions the same query hasbeen run on ISI Web of Knowledge and Pubmed without founding additional contributionsrespect to those already identified through Scopus To increase the likelihood of acomprehensive exploration of past contributions dealing with ldquoEBMrdquo in healthcare thequery strategy has been left significantly open thus searching for ldquoEBMrdquo OR ldquoEBMgtrdquo intitles abstracts and key words A time limitation has not been implemented and datacollection has been run in February 2018 in this regards all articles collected in Scopus tillFebruary 2018 have been searched through the queries that have been pointed out aboveWith respect to the ldquotyperdquo of contribution the searched has been restricted to ldquoArticlerdquo andldquoReviewrdquo because of the very large number of past contributions about EBMgt (cf in thefollowings) No ldquodomainrdquo limitation has been applied accepting contributions ranging frommedicine to management from engineering to economics etc Only studies published inEnglish have been selected

As result of this search strategy 1253 contributions have been identified for screening

Study selectionPast studies identified through queries have been screened to select those in scope withthis literature review The high number of studies ndash even if larger than other studies ndash hasbeen considered coherent to the purpose of the study ndash ie delineating the state of art withrespect to how different groups of decision-makers in healthcare implement EBMgtpractices and inform decision-making ndash and co-authorsrsquo screening capacity Inclusion andexclusion criteria have been agreed Contributions were included when dealing withsources of evidence for EBMgt with types of decisions and analysis and groups ofdecision-makers Contributions were excluded when neither empirical nor focused to

2071


healthcare Screening has been carried out by two co-authors for each contribution tolimit the risk of excluding relevant past studies or including studies that were out ofscope in case of opposite judgment the two co-authors discussed their opinions to gatheran agreed evaluation when the co-authors remained on their previous opinions and anagreement could not be achieved a third co-author reviewed the contribution todecide whether include or exclude it The first round of screening ndash coherently to the largenumber of contributions identified through the query strategies ndash dealt with titlesand key words Since titles could not provide the readers with enough confidence with theactual contribution of the article co-authors agreed to be prudent at this stage of thescreening process and to exclude only those studies that were evaluated as surely out ofscope and to leave the final decision to the next stage based on abstract and summary firstand full text then

The first screening based on title and keywords reduced the included contributions from1253 to 164 with the exclusion of 1089 studies that have judged as out of scope from tworeviewers The remained records (nfrac14 164) were screened by at least two co-authors on thebasis of their abstract and summary At this stage the exclusion criterion about the focusand the relevance for the healthcare context has been applied

Other 95 contributions have been excluded because they did not deal with EBMgt inhealthcare (eg Rudasill and Dole 2017) The remaining 69 contributions have been screenedon the full text After this stage 39 studies have been excluded either because their findingsand conclusions were not based on empirical data or the full text was not retrievable(eg Borba and Kliemann Neto 2008)

After three rounds of screening 30 past contributions have been selected and included inthis literature review

The results at the different stages have been synthetized in the PRISMA chart(Hutton et al 2015) in Figure 1

Although the included criteria concern empirical papers focused on healthcare we alsohave considered the literature reviews in order to detect further studies to be included in theanalysis through a snowball approach

Data extractionAs result of the screening 30 contributions have been selected for grounding this literaturereview Of them 20 contributions are empirical studies nine are past reviews and onesystematic review Selected contributions are listed in Table I

The authors have read the selected papers and evidences from them have beenextracted after having agreed a data extract form Articles management has beensupported through the use of the Mendeley software (version1161) Data extraction hasbeen informed by the design of a theoretical framework based on an I-P-O approachwhose building blocks are inputs (sources of evidence) processestools (types of analysisof sources of evidence) outcomes (types of managerial decisions or management practices)and target users (decision-makers) Such framework allows to identify the state of artabout EBMgt according to a ldquoprocessrdquo perspective The framework provides at least twomain insights on what we know so far about EBMgt in healthcare First reading theframework as columns four domains of analysis are pointed out the groups ofdecision-makers with respect to EBMgt in healthcare the types of decisions that are takenwithin the EBMgt domain the kinds of analysis that are run on the available evidenceand the sources of evidence Second reading the framework as rows (as shown by theexample in Figure 2) the four domains are connected in logical chains that starting fromthe main groups of decision-makers crystallize which decisions or management practicesrefer to them based on which methods of analysis of the available evidence and on whichare the sources of this evidence

2072

MD5610

FindingsAs result of our screening ten past reviews published in the timespan 2002ndash2014 have beenidentified

Their main focus was discussing previous definitions of EBMgt in healthcare the sourcesof evidence and the acceptance of EBMgt practices in hospitals Although the undoubtablerelevance of these topics they do not provide a ldquoprocessrdquo view of what we know about EBMgtin healthcare In this view the studies included in these literature reviews have been screenedthrough the inclusion and exclusion criteria applied to the Scopus database After suchprocess no additional empirical studies on EBMgt in healthcare have been included in thisreview respect to those already identified through the search within the Scopus databaseThis result confirmed the relevance of these studies for grounding this literature reviewIn this regards Table II offers a comprehensive overview about the information that is storedin the 20 papers on sources of evidence (inputs) analyses and tools (processes) managerialpractices (outcomes) and groups of decision-makers

In a nutshell this picture emerges The sources of evidence for EBMgt are publishedstudies real world evidence and expertsrsquo opinion Evidence is analyzed through literaturereviews data analysis of empirical studies and workshops with experts Decisions dealwith performance assessment of organization units staff performance assessment changemanagement organizational knowledge transfer and strategic planning Organizationalknowledge transfer concerns the transfer of knowledge created by a set of researchers toexperts intending to implement it (Graham et al 2006)

Records identified from database (Scopus)searchingN=1253

Iden

tific

atio

nSc

reen

ing

Elig

ibili

tyIn

clud

ed

Records screened on title andkeywordsN=1253

Records screened on Abstract andSummaryN=164

Studies excludedN=1089

Reason out of scope


Reason not in healthcare


Reason beingtheoreticalconceptual with

no empirical findingFinal studies included

(empirical (n=20) systematic review (n=1) andreviews (n=9))

N=30

Full-text studies assessed foreligibility

N=69

Figure 1PRISMA chart based

on the inclusionexclusion process

from Scopus database

2073


No Type Author(s) Title Journal Year

1 Review Young SAMK Evidence-based management aliterature review

Journal of NursingManagement

2002

2 Review Scott IA Determinants of Quality of In-Hospital Care for Patients withAcute Coronary Syndromes

DiseaseManagement andHealth Outcomes

2003

3 Review Arndt M and Bigelow B Evidence-based management inhealth care organizations acautionary note

Health caremanagement review

2009

4 Review DelliFraine JLLangabeer JR 2nd andNembhard IM

Assessing the evidence of SixSigma and Lean in the health careindustry

Qualitymanagement inhealth care

2010

5 Review Marschall-Kehrel D andSpinks J

The Patient-Centric ApproachThe Importance of SettingRealistic Treatment Goals

European UrologySupplements

2011

6 Review Hakemzadeh F andBaba VV

Toward a theory of evidence baseddecision making

ManagementDecision

2012

7 Review DelliFraine JL Wang ZMcCaughey DLangabeer JR 2nd andErwin CO

The use of six sigma in health caremanagement are we using it to itsfull potential


2013

8 Review Rangachari P RissingP and Rethemeyer K

Awareness of evidence-basedpractices alone does not translateto implementation


2013

9 Review Jaana M Vartak S andWard MM

Evidence-Based Health CareManagement What Is theResearch Evidence Available forHealth Care Managers

Health ServicesResearch andPractice

2014

10 Systematicreview

Nicolay CRPurkayastha SGreenhalgh A et al

Systematic review of theapplication of qualityimprovement methodologies fromthe manufacturing industry tosurgical healthcare

British Journal ofSurgery

2012

11 Empiricalarticle

Veillard J ChampagneF Klazinga N et al

A performance assessmentframework for hospitals TheWHO regional office for EuropePATH project

InternationalJournal for Qualityin Health Care

2005

12 Empiricalarticle

Willmer M How nursing leadership andmanagement interventions couldfacilitate the effective use of ICTby student nurses


2007

13 Empiricalarticle

Pritchard RD HarrellMM DiazGranados Dand Guzman MJ

The Productivity Measurementand Enhancement System AMeta-Analysis

Journal of AppliedPsychology

2008

14 Empiricalarticle

McAlearney ASGarman AN Song PHet al

High-performance work systemsin health care management Part 2Qualitative evidence from five casestudies

Health CareManagementReview

2011

15 Empiricalarticle

Grundtvig M GullestadL Hole T et al

Characteristics implementation ofevidence-based management andoutcome in patients with chronicheart failure Results from theNorwegian heart failure registry

European Journal ofCardiovascularNursing

2011

16 Empiricalarticle

Slater H Davies SJParsons R et al

A policy-into-practice interventionto increase the uptake of evidence-

PLoS One 2012

(continued )

Table IList of selectedcontributionsto inform theliterature review

2074

MD5610

Going more in-depth two main groups of decision-makers are targeted by articles aboutEBMgt in healthcare They are health professionals (mainly physicians and nurses) (nfrac14 840 percent) and hospital managers (nfrac14 10 50 percent) Other groups of decision-makerssuch as policy-makers and researchers have been targeted by just one study respectively


based management of low backpain in primary care Aprospective cohort study

17 Empiricalarticle

Davies C and Howell D A qualitative study Clinicaldecision making in low back pain

PhysiotherapyTheory and Practice

2012

18 Empiricalarticle

Booker LD Bontis Nand Serenko A

Evidence-Based Management andAcademic Research Relevance

Knowledge andProcessManagement

2012

19 Empiricalarticle

FrAtildecedillich A Identifying organizationalprinciples and managementpractices important to the qualityof health care services for chronicconditions

Danish MedicalJournal

2012

20 Empiricalarticle

Song PH Robbins JGarman AN andMcAlearney AS

High-performance work systemsin health care Part 3 The role ofthe business case


2012

21 Empiricalarticle

Kramer M Brewer BBHalfer D et al

Changing our lens Seeing thechaos of professional practice ascomplexity


2013

22 Empiricalarticle

Francis-Smythe JRobinson L and Ross C

The role of evidence in generalmanagersrsquo decision-making

Journal of GeneralManagement

2013

23 Empiricalarticle

Rangachari P MadaioM Rethemeyer RK et al

Role of communication contentand frequency in enablingevidence-based practices

QualityManagement inHealth Care

2014

24 Empiricalarticle

Jaana M Teitelbaum Mand Roffey T

It strategic planning in hospitalsFrom theory to practice

InternationalJournal ofTechnologyAssessment inHealth Care

2014

25 Empiricalarticle

Fazaeli S Ahmadi MRashidian A andSadoughi F

A framework of a health systemresponsiveness assessmentinformation system for Iran

Iranian RedCrescent MedicalJournal

2014

26 Empiricalarticle

McAlearney AS HefnerJL Sieck C et al

Evidence-based management ofambulatory electronic healthrecord system implementation Anassessment of conceptual supportand qualitative evidence

InternationalJournal of MedicalInformatics

2014

27 Empiricalarticle

Alavi SH Marzban SGholami S et al

Howmuch is managersrsquo awarenessof evidence based decisionmaking

Biomedical andPharmacologyJournal

2015

28 Empiricalarticle

Nelson KE and Pilon B Managing organizationaltransitions The chief nurseperspective

Nurse Leader 2015

29 Empiricalarticle

Bai Y Gu C Chen QXiao J Liu D andTang S

The challenges that head nursesconfront on financial managementtoday A qualitative study

InternationalJournal of NursingSciences

2017

30 Empiricalarticle

Guo R Berkshire SDFulton LV et al

Use of evidence-basedmanagement in healthcareadministration decision-making

Leadership in HealthServices

2017

Table I

2075


With respect to health professionals management practices that should be evidence-baseddeal mainly with change management initiatives (nfrac14 3 38 percent) and the assessment ofeither individual (ie of health professionals) or organizational performance (within audit orbenchmarking programs) In both cases expert or peer opinion is the most used source ofevidence to inform decision-making Evidence extracted from electronic medical records orlocal databases lack far behind Literature reviews and evidence extracted from journalarticles is cited in a limited number of studies This finding shows that while physicians andnurses are used to refer to this source of evidence ndash according to the well-established EBMdiscipline ndash for health-related issues and decision-making they refer to evidence with lowerrobustness ndash ie expert opinions ndash when dealing with managerial practices Being thesource of evidence mainly qualitative the types of analysis or tools used to extract ldquovaluerdquofrom the sources of evidence are those that are typically utilized for qualitative data such asinterviews focus groups and meetings With respect to hospital managers the picture hasboth differences and similarities Management practices that should inform by evidence dealmainly with organizational knowledge translation (nfrac14 5 50 percent) performanceassessment of organizational units (nfrac14 3 30 percent) and organizational strategic planning(nfrac14 3 30 percent) As for health professionals the most used source of evidence refers toexpertsrsquo opinion (nfrac14 7 70 percent) Data from electronic medical records and hospitaldatabases (nfrac14 2 20 percent) and articles from the extant literature (nfrac14 1 10 percent) areused in a limited number of cases In particular databases are used mainly with respect tothe assessment of organizational units Again the methods used to extract evidence fromthese sources are mainly qualitative and grounded on interviews and interactions with peersand experts Summarizing in a nutshell what has emerged from the literature is synthetizedin Figure 2 that shows the ldquoprocessrdquo view of the state of art about EBMgt in healthcarebased on an input-process-outcome framework In particular the arrows that connect thebuilding blocks of the framework show two examples of the investigated logical connectionsamong groups of decision-makers (managers in the specific example) types of managerialdecisionspractices types of analysis and tools used to extract value from the sources ofevidence and sources of data

Inputs(Sources of Evidence)

ProcessesTools(Analyses on the Sources of

Evidence)

Outputs(The Kind of the Decision)

Target Users(Decision Makers)

The scientific literatureEmpirical and

theoretical findings fromacademic journals

The organizationLocal population based

data sources (egAdministrative data

EHRs secondary data)

Practitioners1 PersonalExperts Experiences

2 Experts Preferences

3 Peerrsquos Perspective

Literature search Organizationalperformance Assessment

Health professionals

Managers

Policy-makers

Researchers

ManagementStaffperformance Assessment

Data Analyses

Conducting a prospectivestudy

Conducting organizationalqualitative analyses

Testing an evidence-basedmanagement practice in an

organization

Running expertise workshops

Change managementImplementation

Organizational knowledgetranslation

Organizational strategicplanning

Note The blue arrows show an example of the logical connections among the building blocks ofthe framework

Figure 2The ldquoprocessrdquo view ofEBMgt in healthcarebased on an input-process-outcomeframework

2076

MD5610

NoStudy titlecountry year Authors

Inputs(sources ofevidence)

ProcessesTools(analyses on thesources ofevidence)

Outputs (the kindof the decision)

Target users(decisionmakers)

1 A performanceassessmentframework forhospitals TheWHO regionaloffice for EuropePATH projectEurope 2005

Veillard JChampagn EF Klazinga Net al

PersonalExpertsexperiences

Literature searchConducting aSurvey with keyinformants

OrganizationalperformanceassessmentIdentification ofdimensions

Policy-makers

2 How nursingleadership andmanagementinterventionscould facilitate theeffective use ofICT by studentnurses UK 2007

Willmer M PersonalExpertsexperiences

Conductinginterviews withnurses mentorsmanagers

ChangemanagementimplementationDevelopment ofinformation andcommunicationstechnology skills

Healthprofessionalsstudent nurse

3 The productivitymeasurement andenhancementsystem a meta-analysis USA2008

Pritchard RDHarrell MMDiazgranadosD andGuzman MJ

Peer opinion Gathering internalgroup feedbackreports

Staff performanceassessmentReducing roleambiguity androle conflict

Researchers

4 High-performancework systems inhealth caremanagement Part2 Qualitativeevidence from fivecase studies USA2011

McalearneyAS GarmanAN Song PH et al

Peer opinion Literature searchConducting aseries ofinterviews withkey informants

OrganizationalperformanceassessmentIdentification oflinks betweenHPWPs andemployeeoutcomes tosystem andorganization-leveloutcomes

Managers

5 Characteristicsimplementation ofevidence-basedmanagement andoutcome inpatients withchronic heartfailure Resultsfrom theNorwegian heartfailure registryNorway 2011

Grundtvig MGullestad LHole T et al

Localpopulationbased datasources

Analyzing patientdata

Staff performanceassessmentMeasuringhospitalizationmorbidity andmortality rates

Healthprofessionals

6 A policy-into-practiceintervention toincrease the uptakeof evidence-basedmanagement oflow back pain inprimary care Aprospective cohortstudy WesternAustralia 2012

Slater HDavies SJParsons Ret al

PersonalExpertsexperiencesPeer opinion

Measuring self-report measuresrecords forconducting aninterdisciplinaryevidence-basedframework

Staff performanceassessmentSelf-managementstrategies wererecommendedmore frequentlypost-intervention

Healthprofessionals(primary carephysicians-(PCPs))

7 A qualitativestudy Clinicaldecision making

Davies C andHowell D

PersonalExpertsrsquoexperiences

Investigating thedecision-makingprocess PTs use

Identification ofbest practicesPreferred

Healthprofessionals(physical

(continued )

Table IIInformation stored inthe empirical papers(nfrac14 20) included inthe literature review

2077







in low back painUSA 2012

Expertspreferences

when managingpatients with LBPby conductinginterviews

classificationsystems wereidentified

therapists(PT))

8 Evidence-basedmanagement andacademic researchrelevance Canada2012

Booker LDBontis N andSerenko A

Expertspreferences

Investigating thedistribution ofknowledge aboutadvances inintervieweesrsquo fieldof expertise

OrganizationalknowledgetranslationHaving efficientmarketintermediaries inthe form ofknowledgetranslationmechanisms

Managers

9 Identifyingorganizationalprinciples andmanagementpracticesimportant to thequality of healthcare services forchronic conditionsUSA 2012

Fratildecedillich A Localpopulationbased datasources


OrganizationalperformanceassessmentPromotingcontinuity of careand quality ofhealth careservices

Managers

10 High-performancework systems inhealth care Part 3the role of thebusiness caseUSA 2012

Song PHRobbins JGarman ANandMcalearneyAS

PersonalExpertsexperiencesExpertspreferences

Investigating thebusiness case forHPWPs in UShealth careorganizations byconductinginterviews

Organizationalstrategic planningShapeunderstandingaboutorganizationsrsquoperspectives of thebusiness case forHPWP investment

Managers

11 Changing our lensSeeing the chaos ofprofessionalpractice ascomplexity USA2013

Kramer MBrewer BBHalfer D et al


Testing anevidence-basedmanagementpractice in anorganization

OrganizationalperformanceassessmentManagingmultiple patientswith simultaneouscomplex needs

Healthprofessionals

12 The role ofevidence ingeneral managersrsquodecision-makingUK 2013

Francis-Smythe JRobinson Land Ross C

PersonalExpertsrsquoexperiencesPeeropinions


OrganizationalknowledgetranslationManagers get ableto enhance theirbusiness practiceby utilizing moresources of evidence

Managers

13 Role ofcommunicationcontent andfrequency inenabling evidence-based practicesUSA 2014

RangachariP Madaio MRethemeyerRK et al


Conducting aprospective study

OrganizationalknowledgetranslationProvidingcommunicationcontent andfrequencyassociated withcollective learningand culture change

Healthprofessionalsmanagers

(continued )Table II

2078

MD5610






14 IT strategicplanning inhospitals Fromtheory to practiceCanada 2014

Jaana MTeitelbaumM and RoffeyT

ScientificliteraturePersonalExpertsexperiences

Running expertiseworkshops andconductingqualitativeanalyses

OrganizationalstrategicplanningIT strategicplanning formobile andremote access topatientsrsquoinformation andimplementation ofan integratedEMR

IT leadersManagers

15 A framework of ahealth systemresponsivenessassessmentinformationsystem for IranIran 2014

Fazaeli SAhmadi MRashidian AandSadoughi F

PersonalExpertsexperiencesExpertisepreferences

Conductingqualitativeanalyses

OrganizationalperformanceassessmentProvidingrecommendationsand developing aframework

Managers

16 Evidence-basedmanagement ofambulatoryelectronic healthrecord systemimplementationan assessment ofconceptualsupport andqualitativeevidence USA2014

McalearneyAS HefnerJL Sieck Cet al


Synthesizing bestpractices formanagingambulatory EHRsystemimplementation inhealthcareorganizations byconductinginterviews

Organizationalstrategic planningimplementingPlan-Do-Study-Act (PDSA)qualityimprovement (QI)mode

Managers

17 How much ismanagersrsquoawareness ofevidence baseddecision makingIran 2015

Alavi SHMarzban SGholami Set al

PersonalExpertsexperiencesScientificliterature

Determining thelevel of managerrsquosawareness ofevidence baseddecision makingby implementinga cross-sectionalstudy

OrganizationalknowledgetranslationRaising theefficiency ofmanagement inhealthcareorganizations

Managers

18 Managingorganizationaltransitions Thechief nurseperspective USA2015

Nelson KENS Pilon B

ScientificliteraturePersonalExpertsexperiencesPeer opinion

Implementing aproposedorganizationaltransitionframework

ChangemanagementimplementationThe organizationaltransitionframework wassuccessfulalthough thedifferent hospitaland leaderscharacteristics

Healthprofessionals(nurseleaders)

19 The challengesthat head nursesconfront onfinancialmanagementtoday a

Bai Y Gu CChen Q XiaoJ Liu D andTang S

Peer opinionPersonalExpertsexperiences

Identifying thefinancialmanagementpracticechallenges in theorganization by

ChangemanagementimplementationThe decision onimplementing acooperativemanagement

Healthprofessionals(head nursenursemanagers)

(continued ) Table II

2079


Discussion and conclusionsThis study aimed at crystallizing the state of art of EBMgt in healthcare through the novelangle of a ldquoprocessrdquo view Past reviews focused mainly to the comparison of differentdefinitions and scopes of EBMgt in healthcare pointing out the need of better formalizationof this research field Despite the undoubted value of this debate this study takes a stepahead by systematizing the main findings from past researches within an inputs-processes-outcomes framework that allows to materialize the logical connections among variousgroups of decision-makers types of managerial decisionspractices types of analysis andtools to extract value from different sources of evidence and the available sources ofevidence (Figure 2)

In the light of the results emerged from the literature review three main issues are worthof discussion First EBMgt deals mainly with two groups of decision-makers hospitalmanagers and health professionals On the one hand this result clarifies that EBMgt shouldnot be limited to managers but should include all professionals that in healthcare are incharge of taking managerial decisions and execute practices of management Headphysicians combine professional and managerial responsibilities and because of that theyshould translate those they have learned about EBM to tasks and issues that deal withmanagement On the other hand other relevant groups of decision-makers have beenlargely overlooked This is the case of policy-makers Even if the last years have seen thediffusion of narratives about evidence-based policy-making this is not what emerged fromthis study This difference might be due to the choice of including in this literature reviewonly studies with an empirical grounding Evidence-based policy-making is still far fromconsolidated practices and tools that have been investigated through quantitative analysesWhat we know and what is expected for the next years are mainly based on expert opinionsand positioning papers In this view more efforts should be paid by scholars of decisionmaking and healthcare management to pave quantitatively the avenue of evidence-baseddecision-making

Second the most investigated sources of evidence are opinions of experts and peersThis result is in contrast with the emphasis paid to electronic medical records and






qualitative studyChina 2017

conducting groupinterviews

model evidence-based managementtraining and data-driven tools toimproving thefinancialmanagementcapacity of nursemanagers

20 Use of evidence-basedmanagement inhealthcareadministrationdecision-makingUSA 2017

Guo RBerkshire SD Fulton LV et al

Peer opinion Conducting across-sectionalstudy to collectthe opinion ofmanagers

OrganizationalknowledgetranslationThe decision onmanagers prioritysetting of usingevidence sourcesfor consultingdaily and weeklyfor decision-making

Managers

Table II

2080

MD5610

administrative databases in the last decade On the one hand these sources of evidencecollect data that are not salient for management-related decisions For instance the actualcapability to explain the performance variance for a sample of hospitals in terms ofdifferent management practices is very limited through administrative health dataThese data sets do not collect exhaustive information about the organizationaldeterminants of hospital performance and thus hospital managers are forced to exploreother sources of evidence such as opinions of experts and peers or qualitative surveysOn the other hand hospital managers might not have enough confidence and skills tomake sense of quantitative sources of evidence such as administrative data Results fromthis systematic literature review show that hospital managers and health professionalshave similar behaviors in term of sources of evidence for management-related decisionsalthough physicians are used to ground clinical decisions on sources with a higher degreeof robustness and generalizability In this view further research should be carried out toinvestigate the attitude of different groups of decision-makers to ground theirmanagement practice to innovative sources of evidence

Third the development of a theoretical framework anchored in an inputs-processes-outcomes model has shown that current research on EBMgt in healthcare needs a differentangle to take a step ahead and overcome the impasse that has characterized the lastdecade The authors argue that the debate about what ldquoevidencerdquo is or should be inhealthcare is sterile where not connected with the specific group of decision-makers thespecific group of management practices or managerial decisions the specific group ofanalytic techniques and the specific sources of evidence In this view Figure 2 offersinteresting insights to both academicians and practitioners Researchers should payadditional efforts to complete such picture In fact the picture is the result of what hasbeen found so far in past studies and is not the result of theoretical arguments Forinstance other groups of decision-makers might be included (eg patients and advocacygroups) as well as other sources of evidence (eg real world data and social media)Additionally the logical connections among the building blocks should be discussedin-depth and crystallized Practitioners vice versa might benefit from this picture interms of improved awareness of the scope and complexity of EBMgt in healthcare andimproved capability to develop best practices that connects sources of evidence withanalytic techniques and with groups of management practices By leveraging on suchframework the set-up of bench-learning initiatives would be easier and more focused

References

Aron DC (2015) ldquoFrom evidence-based medicine to evidence-based management (and policy)rdquoMedical Care Vol 53 No 6 pp 477-479

Baba VV and HakemZadeh F (2012) ldquoToward a theory of evidence based decision makingrdquoManagement Decision Vol 50 No 5 pp 832-867

Beglinger JE (2006) ldquoQuantifying patient care intensity an evidence-based approach to determiningstaffing requirementsrdquo Nursing Administration Quarterly Vol 30 No 3 pp 193-202

Borba GSD and Kliemann Neto FJ (2008) ldquoGestatildeo Hospitalar identificaccedilatildeo das praacuteticas deaprendizagem existentes em hospitaisrdquo Sauacutede e Sociedade Vol 17 No 1 pp 44-60 available athttpsdxdoiorg101590S0104-12902008000100005

Briggs HE andMcBeath B (2009) ldquoEvidence-basedmanagement origins challenges and implications forsocial service administrationrdquo Administration in Social Work Vol 33 No 3 pp 242-261

Briner RB Denyer D and Rousseau DM (2009) ldquoEvidence-based management concept cleanuptimerdquo Academy of Management Perspectives Vol 23 No 4 pp 19-32

Davies C and Howell D (2012) ldquoA qualitative study clinical decision making in low back painrdquoPhysiotherapy Theory and Practice Vol 28 No 2 pp 95-107

2081


Fazaeli S Ahmadi M Rashidian A and Sadoughi F (2014) ldquoA Framework of a health systemresponsiveness assessment information system for Iranrdquo Iranian Red Crescent Medical JournalVol 16 No 6 p e17820

Francis-Smythe J Robinson L and Ross C (2013) ldquoThe role of evidence in general managersrsquodecision-makingrdquo Journal of General Management Vol 38 No 4 pp 3-22

Ghezzi A Martini A and Natalicchio A (2017) ldquoCrowdsourcing a review and suggestions for futureresearchrdquo International Journal of Management Reviews Vol 20 No 2 pp 343-363

Graham ID Logan J Harrison MB Straus SE Tetroe J Caswell W and Robinson N (2006)ldquoLost in knowledge translation time for a maprdquo Journal of Continuing Education in the HealthProfessions Vol 26 No 1 pp 13-24

Grundtvig M Gullestad L Hole T Floslashnaeligs B and Westheim A (2011) ldquoCharacteristicsimplementation of evidence-based management and outcome in patients with chronic heartfailure results from the Norwegian heart failure registryrdquo European Journal of CardiovascularNursing Vol 10 No 1 pp 44-49

HakemZadeh F and Baba VV (2016) ldquoMeasuring the actionability of evidence for evidence-basedmanagementrdquo Management Decision Vol 54 pp 1183-1204

Hamlin B (2002) ldquoTowards evidence-based management and research-informed HRD practice anempirical studyrdquo International Journal of Human Resources Development and ManagementVol 2 Nos 1-2 pp 160-169

Hamlin RG Ruiz CE and Wang J (2011) ldquoPerceived managerial and leadership effectiveness withinMexican and British public sector hospitals a cross-nation comparative analysisrdquo HumanResource Development Quarterly Vol 22 No 4 pp 491-517

Hopp WJ et al (2018) ldquoBig data and the precision medicine revolutionrdquo Production and OperationsManagement available at httpsdoiorg101111poms12891

Hornby P and Perera HSR (2002) ldquoA development framework for promoting evidence-based policyaction drawing on experiences in Sri Lankardquo International Journal of Health Planning andManagement Vol 17 No 2 pp 165-183

Hutton B Salanti G Caldwell DM Chaimani A Schmid CH Cameron C et al (2015) ldquoThePRISMA extension statement for reporting of systematic reviews incorporating network meta-analyses of health care interventions checklist and explanationsrdquo Annals of Internal MedicineVol 162 pp 777-784 doi 107326M14-2385

Jaana M Vartak S and Ward MM (2014) ldquoEvidence-based health care management what is theresearch evidence available for health care managersrdquo Health Services Research and PracticeVol 37 No 3 pp 314-334

Kovner AR (2014) ldquoEvidence-based management implications for nonprofit organizationsrdquoNonprofitManagement amp Leadership Vol 24 No 3 pp 417-424

Kovner AR and Rundall TG (2006) ldquoEvidence-based management reconsideredrdquo Frontiers ofHealth Services Management Vol 22 No 3 pp 3-22

McGrath JE (1964) Social Psychology A Brief Introduction Rinehart and Winston Holt New York NY

Marschall-Kehrel D and Spinks J (2011) ldquoThe patient-centric approach the importance of settingrealistic treatment goalsrdquo European Urology Supplements Vol 10 No 1 pp 23-27

Oliver A Mossialos E and Robinson R (2004) ldquoHealth technology assessment and its influence onhealth care priority settingrdquo International Journal of Technology Assessment in Health CareVol 20 No 1 pp 1-10

Pfeffer J and Sutton RI (2006) Hard Facts Dangerous Half-Truths and Total Nonsense Profitingfrom Evidence-Based Management Harvard Business Press Boston MA

Reay T Berta W and Kohn MK (2009) ldquoWhatrsquos the evidence on evidence-based managementrdquoAcademy of Management Perspectives Vol 23 No 4 pp 5-18

Rousseau DM (2006) ldquoIs there such a thing as lsquoevidence-based managementrsquordquo Academy ofManagement Review Vol 31 No 2 pp 256-269

2082

MD5610

Rudasill LM and Dole WV (2017) ldquoA tale of two outliers evidence-based management in non-ARLresearch libraries and pre-NARA presidential librariesrdquo Journal of Library AdministrationVol 57 No 8 pp 922-932

Sackett DL Rosenberg WMC Gray JAM Haynes RB and Richardson WS et al (1996)ldquoEvidence based medicine what it is and what it isnrsquotrdquo British Medical Journal Vol 312No 7023 pp 71-72

Schmalenberg C Kramer M King CR Krugman M Lund C Poduska D and Rapp D (2005)ldquoExcellence through evidence securing collegialcollaborative nurse-physician relationshipspart 1rdquo Journal of Nursing Administration Vol 35 No 10 pp 450-458

Simsek Z (2009) ldquoOrganizational ambidexterity towards a multilevel understandingrdquo Journal ofManagement Studies Vol 46 pp 597-624 doi 101111j1467-6486200900828x

Slater H Davies SJ Parsons R Quintner JL and Schug SA (2012) ldquoA policy-into-practiceintervention to increase the uptake of evidence-based management of low back pain in primarycare a prospective cohort studyrdquo PLoS One Vol 7 No 5 pp e38037 1-13 available at httpsdoiorg101371journalpone0038037

Spender JC Corvello V Grimaldi M and Rippa P (2017) ldquoStartups and open innovation a review ofthe literaturerdquo European Journal of Innovation Management Vol 20 No 1 pp 4-30

Tranfield D Denyer D and Smart P (2003) ldquoTowards a methodology for developing evidence-informed management knowledge by means of systematic reviewrdquo British Journal ofManagement Vol 14 pp 207-222 doi 1011111467-855100375

Veillard J Champagne F Klazinga N Kazandjian V Arah OA and Guisset AL (2005)ldquoA performance assessment framework for hospitals the WHO regional office forEurope PATH projectrdquo International Journal for Quality in Health Care Vol 17 No 6pp 487-496

Walshe K and Rundall TG (2001) ldquoEvidence‐based management from theory to practice in healthcarerdquo The Milbank Quarterly Vol 79 No 3 pp 429-457

Willmer M (2007) ldquoHow nursing leadership and management interventions could facilitate theeffective use of ICT by student nursesrdquo Journal of Nursing Management Vol 15 No 2pp 207-213

Young SK (2002) ldquoEvidence-based management a literature reviewrdquo Journal of NursingManagement Vol 10 pp 145-151

About the authorsAfsaneh Roshanghalb is PhD Student at the Politecnico di Milano Department of ManagementEconomics and Industrial Engineering She holds a Master of Science in Public Administration fromTarbiat Modares University Her research is focused on The Power of the Big Data for Evidence-basedManagement in Healthcare the case of the health administrative databases Afsaneh Roshanghalb isthe corresponding author and can be contacted at afsanehroshanghalbpolimiit

Emanuele Lettieri is Full Professor at the Department of Management Economics and IndustrialEngineering (DIG) of Politecnico di Milano He chairs the Accounting Finance amp Control (AFC) andHealthcare Management (HCM) master courses at Politecnico di Milano He is Director of theInternational MBA Part-Time program at MIP Politecnico di Milano Graduate School of Business Hisresearch interests are at the intersection among technology management and healthcare and deal withinnovation management and performance improvement in healthcare His current researches deal withthe development of evidence-based improvement strategies in hospitals through the use ofadministrative data the diffusion of digital innovation in healthcare with particular interest to digitalservices to citizens Apps and wearables the assessment of innovations in healthcare accordingly tothe Health Technology Assessment discipline and the implementation of value-based strategies inhealthcare His research is both qualitative and quantitative He has conducted multidisciplinaryresearch in collaboration with Universities research centres healthcare institutions and hospitals Hehas participated in applied research large-scale European projects Finally he is continuously involvedin the education of healthcare professionals as well as healthcare companiesrsquo personnel with the designof ad-hoc classes

2083


Davide Aloini PhD is Associate Professor of Business Process Management at the Department ofEnergy Systems Land and Constructions Engineering at the University of Pisa Italy His researchinterests include operation and information system management More recent studies have focused onBusiness Process Management and CollaborativeAdvanced ICT solutions with special interest inlarge-scale project healthcare systems and innovation in high tech firms He has published papers ininternational journals such as InformationampManagement European Journal of Operation ManagementProduction Planning and Control Expert Systems with Applications and Technology Forecasting andSocial Change In 2008 he was rewarded with a Citation of Excellence Award by Emerald

Lorella Cannavacciuolo Assistant Professor in Management Accounting and PhD inEconomic and Managerial Engineering carries out her research activity at the Department ofIndustrial Engineering of University of Naples Federico II Her research interests encompassinnovation network systems in SMSe process mapping and redesign network measurements forlarge collaborative platforms activity accounting models for cost performance managementHer research interests are mainly in the area of healthcare management She has published papers ininternational journals and she serves as reviewer for many international journals in operations andhealthcare management

Simone Gitto PhD is Associate Professor of Business and Management Engineering at theUniversity of Udine Italy He teachesed Engineering Economics Microeconomics and MarketingHe is Deputy Director of Master of Arts in ldquoHuman Resource Managementrdquo at the University of RomeTor Vergata He was Assistant Professor at the University of Rome Tor Vergata He was ResearchScholar at the John E Walker Department of Economics Clemson University SC in 2008 His mainresearch interests include air transport regulation health efficiency and forecasting methods andproductivity and economic growth He has edited special issues for research in transportationeconomics He has been principal investigator or member of research projects His works have beenpublished in international refereed journals including British Journal of Management InternationalJournal of Production Economics Journal of Air Transport Management Journal of ProductivityAnalysis Technological Forecasting and Social Change Transportation Research Part A andTransportation Research Part E

Filippo Visintin is Associate Professor of Service Management at the Department of IndustrialEngineering of the University of Florence Italy He is Scientific Director of the IBIS Lab and co-founderand co-owner of Smartoperations srl He regularly advices public and private healthcare organisationsHe was visiting research scholar at the School of Management Binghamton University NY in 2006His research interests include servitization of manufacturing and healthcare operations managementHe is author of several research papers published in journals such as European Journal ofOperational Research Industrial Marketing Management International Journal of ProductionEconomics Computers in Industry Flexible Service and Manufacturing Journal Journal of IntelligentManufacturing Production Planning and Control and IMA Journal of Management Mathematics

For instructions on how to order reprints of this article please visit our websitewwwemeraldgrouppublishingcomlicensingreprintshtmOr contact us for further details permissionsemeraldinsightcom

2084

MD5610

Three perspectives onevidence-based management

rank fit varietyPeter F Martelli

Sawyer Business School Suffolk University Boston Massachusetts USA andTuna Cem Hayirli

Harvard Medical School Boston Massachusetts USA

AbstractPurpose ndash The debate on evidence-based management (EBMgt) has reached an impasse The persistence ofmeaningful critiques highlights challenges embedded in the current frameworks The field needs to consider newconceptual paths that appreciate these critiques but move beyond them The paper aims to discuss this issueDesignmethodologyapproach ndash This paper unpacks the concept of finding the ldquobest available evidencerdquowhich remains a central notion across definitions of EBMgt For each element it considers relevant theoryand offers recommendations concluding with a discussion of ldquobestnessrdquo as interpreted across three keydynamics ndash rank fit and varietyFindings ndash The paper reinforces that EBMgt is a social technology and draws on cybernetic theory to arguethat the ldquobestrdquo evidence is produced not by rank or fit but by variety Through variety EBMgt more readilycaptures the contextual political and relational aspects embedded in management decision makingResearch limitationsimplications ndashWhile systematic reviews and empirical barriers remain importantmore rigorous research evidence and larger catalogues of contingency factors are themselves insufficient tosolve underlying sociopolitical concerns Likewise current critiques could benefit from theoretical bridgesthat not only reinforce learning and sensemaking in real organizations but also build on the spirit of theproject and progress made towards better managerial decision makingOriginalityvalue ndash The distinctive contribution of this paper is to offer a new lens on EBMgt drawing fromcybernetic theory and science and technology studies By proposing the theoretical frame of variety it offerspotential to resolve the impasse between those for and against EBMgtKeywords Management theory Knowledge management Implementation Evidence-based managementManagement strategy Theory of evidencePaper type General review

1 IntroductionOver the past decade the evidence-based management (EBMgt) debate has arrived at animpasse with two strands of scholarship developing in tandem yet in relative isolation Despitea few attempts at comprehensive theory building (Baba and HakemZadeh 2012 Mankelwiczand Kitahara 2008) the field remains perilously undertheorized A manager newly venturinginto this literature could easily develop some confusion about EBMgt and its practice

On the one hand arguments for EBMgt have largely built upon and refined earlydefinitions in a realist orientation (Martelli 2012) For those adherent EBMgt has beendefined as the ldquosystematic application of the best available evidence to the evaluation ofmanagerial strategies for improving [organizational] performancerdquo (Kovner and Rundall2006) Over time this definition has been refined into one or another version of ldquomakingdecisions through the conscientious explicit and judicious use of the best available evidencefrom multiple sources by asking acquiring appraising aggregating applying andassessing to increase the likelihood of a favorable outcomerdquo (Barends et al 2014)

On the other hand arguments against EBMgt have typically taken a social constructivistorientation (Martelli 2012) and have eschewed existing definitions on theoretical andpractical grounds Authors from this position write that ldquodespite claims to be scientific andimpartial EBMgt is managerialist ie it is for management not about managementrdquo



0025-1747DOI 101108MD-09-2017-0920

Received 30 September 2017Revised 6 March 2018

3 May 2018Accepted 24 June 2018


2085

Evidence-based

management


(Morrell and Learmonth 2015 see also Arndt and Bigelow 2009 Mowles 2011)In particular a consistent rebuttal is that EBMgt minimizes the range evidence can take bymarginalizing other forms besides research evidence In this it ldquodevalues stories ornarrative forms of knowledge Yet [hellip] is itself a story about relations between research andpractice one of many possible storiesrdquo (Morrell and Learmonth 2015)

Recent reviews have called for a pause in theory building and an increase in ldquotheproduction of high-quality empirical studies in EBMgtrdquo (Rynes and Bartunek 2017) Howeverit is difficult to advance the field without implementing EBMgt practices built upon a strongtheoretical foundation such that practices are comparable and replicable It is important tonote that EBMgt has a dual nature both as a suggested method of improving socio-behavioraltechnologies in the organization as well as a socio-behavioral technology in itself

If we are to consider EBMgt ldquoa simple idea [hellip] [that] means finding the best evidencethat you can facing those facts and acting on those factsrdquo (Pfeffer and Sutton 2006) then itis important to consider within the management context what counts as evidence what itspurpose is and how it fits into the decision-making process

The aim of this paper is to reimagine EBMgt in a way that is sensitive to both theaspirations and limitations of the project In Section 2 it reviews the similarities anddifferences between Evidence-based Medicine and EBMgt highlighting the unique featuresof healthcare organizationsrsquo contexts In challenging the realist core of the ldquobest availableevidencerdquo Section 31 stresses the social aspects of evidence describing EBMgt as a socialtechnology Section 32 discusses availability in terms of literal availability of sources andcognitive availability ldquoBestnessrdquo is then operationalized as ldquohierarchical rankingrdquo and ldquofitbetween situation and evidencerdquo in Section 4 with both operationalizations falling shortwhen uncertainty abounds The third operationalization in Section 5 suggests thatemploying a variety of knowledge types is a preferable approach in healthcare because itincreases organizational regulation states shapes interpersonal knowledge structures anddirects organizational attention

This general review paper derives from a multi-stage and multidisciplinary literaturereview conducted over several research projects including the authorsrsquo dissertation andthesis work and associated research studies funded by Agency for Healthcare Research andQuality the Gordon and Betty Moore Foundation and the National Science FoundationThus the literature presented here comes not from a single review methodology but from aseries of reviews over a decade feedback in multiple professional venues and conversationswith prominent scholars in the field

2 EBMgt for performance improvement in healthcareThe earliest formulations of EBMgt were based on the design of its forerunner conceptevidence-based medicine These models favored the increased use of research literature as themain function of the process arguing not only that the evidence being used is sub-optimal butalso implicitly that much of it is simply not evidence However just as healthcare managementis not the provision of healthcare EBMgt in healthcare is not evidence-based medicineAs Walshe and Rundall (2001) note

Overall the tightly defined well-organized highly quantitative and relatively generalizable researchbase for many clinical professions provides a strong and secure foundation for evidence-basedpractice and lends itself to a systematic process of review and synthesis and to the production ofguidelines and protocols In contrast the loosely defined methodologically heterogeneous widelydistributed and hard-to-generalize research base for healthcare management is muchmore difficult touse in the same way

On one hand Pfeffer and Sutton (2006) argue that ldquomanagers (like doctors) can practicetheir craft more effectively if they are routinely guided by the best logic and evidencerdquo

2086

MD5610

On the other hand Learmonth and Harding (2006) argue nevertheless ldquothe basic doctrine ofEBMgt remains one appropriated from evidence-based healthcare that a consideration ofevidence will increase the rationality and thus the effectiveness of managersrsquo decisionsrdquo

The pursuit of improvement in healthcare provides a perfect setting to explore theconcerns above First there is ldquoplenty of evidence that a research practice gap also exists inhealthcare policy and managementrdquo (Walshe and Rundall 2001) Second healthcarerepresents a form of complex service organization in which uncertainty is present (Plsek andGreenhalgh 2001) and failure is never desired though highly likely (Edmondson 2010)Third health services and hospitals compose a knowledge-intensive knowledge-centeredindustry in which speed of change and expertise play critical roles (Brint 2001) Fourth inthe delivery of healthcare ldquocomplexity is reflected in the number variety andfragmentation of producers involvedrdquo including mutually interactive dynamic andnon-linear relationships between system parts (Begun et al 2003) Moreover decisionmaking in this domain is ldquoquasi-scientific in a particular sense competent decision makingrequires scientific knowledge but scientific knowledge is not sufficient to make decisionsrdquo(Turner 2004) Finally while medicine operates in an ldquoenvironment of fairly high validityrdquowhere validity refers to the stability of relationships between ldquoobjectively identifiable cuesand subsequent events or between cues and the outcomes of possible actionsrdquo (Kahnemanand Klein 2009) the management of healthcare like management in general is more likelyoperating in a low validity environment

The discussion below presents an argument generic in nature though particularlyamenable to strategic improvement initiatives As such the target audience is healthcareadministrators responsible for strategic or high-level operational decisions related to therestructuring positioning prioritizing and financing of care delivery Improvement inhealthcare requires contending with highly differentiated yet highly reciprocal tasks in asetting where ldquophysicians align with technical expertise nurses with reliability and safetyand health administrators with efficiencyrdquo and ldquowhile health administrators may advocatefor organizational change they typically do not have real administrative authority overhealth professionalsrdquo (Garman et al 2006)

With these factors in mind this paper elaborates on the nature of EBMgt as a socialtechnology and offers three perspectives on its operationalization

3 What is the ldquobest available evidencerdquoEmbedded in the definition of EBMgt is the implication that the ldquobest available evidencerdquoshould be marshaled in management decision making Table I presents several accepteddefinitions that highlight the importance of this concept Though the breadth of applicationchanges over time the underlying intention of ldquobestnessrdquo remains For this reason it isuseful to briefly overview what is meant by each of these three terms and the consequencesof framing decision making accordingly

31 Evidence is social EBMgt is a social technologyEvidence is ldquoground for belief testimony or facts tending to prove or disprove anyconclusionrdquo (Oxford English Dictionary 2nd ed 1989) That observation is theory-laden issufficient to show that individual knowledge is distinct from objectively true facts orinformation about entities in the world (Kuhn 1962) This distinction magnifies in a socialcontext where the shared perspectives standards and goals of a community influence thestatus of knowledge claims Evidence is context specific and relational tied to a particularstance perspective or intention and is compiled in support of a particular end Whereasknowledge can exist free-form evidence can only exist as a package of knowledge directedtowards a goal For organizations this means that evidence is wrapped up in contextshared meaning and interpersonal goal reconciliation

2087

Evidence-based

management

Kuhn (1962) underscored the importance of shared meaning by proposing the commonvalues (ie empirical accuracy consistency broad scope simplicity and fruitfulness) bywhich individuals can discuss and reconcile different scientific paradigms Referringespecially to evidence-based practice Donaldson (2009) proposes relevance coherenceverisimilitude justifiability and contextuality as the common values which govern the useof evidence in organizations Likewise Baba and HakemZadeh (2012) propose that ldquothe bestevidence needs to be evaluated against methodological fit contextualization transparencyreplicability and consensusrdquo Like most social propositions the dimensions of value inevidence are often in tension - for example Keller (2009) suggests that features of saliencecredibility and legitimacy are interconnected such that procedures developing one tend toundermine another In sum rhetoric plays a large role in persuading individuals to switchgestalts between positions using an evidence-based process

This paper suggests that EBMgt is not merely a tool or process but a social technologyinextricably embedded in personal and organizational values and culture As such EBMgtis not a value neutral tool to be used by technocratic managers but is ldquosituated in cultureand embedded in historyrdquo ( Jasanoff 2012) with actors making decisions in social contextsinvolving power dynamics For instance Arndt and Bigelow (2009) elaborate on theconsideration of evidence in healthcare contexts by noting that ldquolsquoBest evidencersquo in turn isan artifact of the social processes that lead to its creation reflecting researchersrsquo ororganizationsrsquo interests in the selection of topics what questions to ask and what sources ofinformation to legitimaterdquo Regulation of epistemic uncertainty in an organizationalmanagement context depends on social perception and complex environments alter thestructure of decision making since ldquothe environment in which decisions are made is key notsimply [hellip] as a setting but as an embedded entity which forms both lsquosubstancersquo and lsquoarenarsquofor the strategic actorsrdquo (Gore et al 2006) In socio-cultural systems mental models areformed interpersonally and form the regulatory mechanisms by which organizationsdiscriminate act upon and respond to uncertainty in the environment

Barends et al (2014) propose that evidence-based practitioners ask acquire appraiseaggregate apply and assess four unique sources of evidence scientific organizationalexperiential and stakeholder In that same order such sources deal with published researchfindings data from the organization tacit knowledge from professional experience and the

Source Definition

Kovner et al (2000) [T]he conscientious explicit and judicious use of current best reasoning and experiencein making decisions about strategic interventions

Kovner and Rundall(2006)

The systematic application of the best available evidence to the evaluation ofmanagerial strategies for improving the performance of organizations

Rousseau (2006) [EBMgt] means translating research principles based on best evidence intoorganizational practice

Pfeffer and Sutton(2006)

[EBMgt] is a commitment to finding and using the best theory and data available at thetime to make decisions

Briner et al (2009) EBMgt is about making decisions through the conscientious explicit and judicioususe of four sources of information practitioner expertise and judgment evidence fromthe local context a critical evaluation of the best available research evidence and theperspectives of those people who might be affected by the decision

Rynes et al (2014) [EBMgt] is about making decisions through the conscientious explicit and judicioususe of the best available evidence from multiple sources to help managers chooseeffective ways to manage people and structure organizations

Barends et al (2014) Evidence-based practice in management is about making decisions through theconscientious explicit and judicious use of the best available evidence from multiplesources by asking acquiring appraising aggregating applying and assessing

Table ICommon definitionsof EBMgt

2088

MD5610

values and concerns of stakeholders ldquowho may be affected by an organizationrsquos decisionsand their consequencesrdquo This model is concerned with how stakeholders ldquotend to react tothe possible consequences of the organizationrsquos decisionsrdquo imagined as a tool that providesa ldquoframe of referencerdquo An appreciation of EBMgt as a social technology however demandsthat one envision factors like culture and values as inextricable parts of the social contextenveloping how decisions are formulated acted upon and received Such factors should notbe divorced from other sources of evidence and should be interpreted reflexively Managersin healthcare should realize the variance of ldquoideas and experiences and engage in dialoguethat is critical open and questioningrdquo (Cunliffe and Jun 2005) within their social realitiesbeing careful to not ldquoignore the situated nature of that experience and the cultural historicaland linguistic traditions that permeate [their] workrdquo (Cunliffe 2003) Just as ldquothe skilledclinician does not first collect and deploy evidence and then soften it up with narrativerdquo(Charon and Wyer 2008) so should managers in healthcare vigilantly remain reflexive tothe conditions surrounding a decision and their own role in specifying them

To that end this paper argues a decision-making approach more in the tradition of therational decision logic of appropriateness which is concerned with ambiguity and attentionthan the rational decision logic of consequences which privileges intentionality andbounded rationality (Frederickson and Smith 2003) The logic of appropriatenessemphasizes that ldquobehavior in a specific situation is said to follow from the rules that governthe appropriate course of action for a given role or identityrdquo (Balsiger 2016) In healthcareparticularly shared values and norms within professions play a compelling role inestablishing and maintaining the assumptions underlying otherwise rational justificationsKeeping in mind this complex climate of healthcare and the social nature of evidenceembedded in it it is important to discuss how the availability of such evidence is imaginedwith respect to decision making

32 Availability takes two formsUsing the best evidence implies that it is available to the decision-maker at the time of thedecision Available can be interpreted in two ways Evidence is transmitted throughsources yet sometimes these sources are literally unavailable to them in time for a decisionImplementation research has documented various common technical barriers andfacilitators to compiling evidence such as the cost of journals and difficult technologicalinterfaces (Rundall et al 2009) These are important but comparatively simple issues toaddress Available can also refer to what can be comprehended by the decision-maker ororganization ndash a sort of cognitive availability Individuals modeling their worlds undercertain assumptions may not be able to conceive of competing knowledge claims and mayreject evidence as rhetorically unpersuasive Models of decision choice under uncertaintyare subject to the incompleteness hypothesis which asserts that ldquobecause [a decision] modelfails to capture all relevant aspects of the problem it will yield inaccurate estimates of theexpected benefits of any given course of actionrdquo (Quiggin 2004)

Likewise organizations have limited attention available to search and process evidencewhere attention is defined as the ldquonoticing encoding interpreting and focusing of time andeffort by organizational decision makers on both (a) issues [hellip] and (b) answersrdquo (Ocasio1997) Firms faced with ldquotoo much data and not enough informationrdquo compel organizationaldecision makers to ldquooversimplify to deal with overloadrdquo (Matheson and Matheson 1998)The focus of attention is important for discovery innovation and strategic action Forinstance both the total number of sources and the number of sources across severalknowledge types used exhibit an inverted-U shaped relationship with corporate innovation(Laursen and Salter 2006) ndash search breadth alone itself doesnrsquot yield more robust attentionOrganizations may also have influential individuals or sub-systems that attend to certaintypes of evidence more than others leading the organization through socio-behavioral

2089

Evidence-based

management

drivers to privilege that evidence in rhetorical justification to the exclusion of others In thiscase the evidence similarly becomes unavailable to decision makers

Uncertainty and how an individual or a community of individuals comes to know theunknown remain the motivating issues On this Rousseau (2016) commented in an onlinegroup discussion on EBMgt ldquoI would bet (really) that [EBMgt] practice will lead to greaterdiversity of decision processes as practitioners come to recognize the degree ofuncertainty that actually exists in management decisions Thus I would expect differencesin processes used to deal with low uncertainty decisions vs high uncertainty decisions andwhatever is in betweenrdquo Understanding how such processes vary depends on howldquobestnessrdquo in addition to availability is interpreted and how the dynamics of eachconceptualization affects practice

4 Best as rank or fitAs a thought experiment assume that a ldquobestrdquo set of evidence for a decision existed Howwould you know what is was How would you compile it

Two immediate interpretations come to mind First consider an interpretation whichevaluates ldquobestnessrdquo according to a hierarchy of evidence This ranking perspective wouldimply that a certain type of evidence or perhaps evidence generated by certain processeswill rank higher or lower in its capacity to support truth claims

Best has traditionally been established with an underlying assumption of logos (ie anappeal to the strength and consistency in logical argument) with the ldquobestrdquo evidencemeeting the epidemiological standard of the randomized controlled trial (RCT) Howeverwhere evidence is better it is also worse In evidence-based medicine virtually allinstitutional reviewers of evidence (ie USPSTF ICSI SORT GRADE Oxford Center) gradeexpert assessment as the lowest strength of evidence The problem with thischaracterization in socio-behavioral settings is twofold

First consider the example of a ldquoparachute approach to evidence-based medicinerdquo(Potts et al 2006) referring to an earlier tongue-in-cheek article calling for an RCT toestablish definitively whether parachute use prevents trauma due to ldquogravitationalchallengerdquo This view advocates making policy decisions on ldquogood sciencerdquo even whenRCTs are unavailable In health research circumspection about the RCT has manifested asthe ldquoreal-world evidencerdquo (RWE) movement which promotes evidence gathered ldquoin clinicalcare and home or community settings as opposed to research intensive or academicenvironmentsrdquo (Sherman et al 2016) Potential sources of data expand to claims datadisease registries and health-monitoring devices (FDA 2018) Yet using only codifiedsources of evidence assumes they can act as substitutes for non-codifiable types ofknowledge in the rhetoric of decision making Moreover the strength of evidence is one ofmany considerations including the fiscal and sociopolitical climate within whichgovernments institutions and communities operate (Tang et al 2003)

Second evidence derives its potency from the knowledge it represents and knowledge istheory-laden and embedded in the language and rhetoric of a given paradigm of inquiry(Kuhn 1962) The ranking approach privileges experimentally collected codifiable andquantifiable knowledge about causal efficacy Yet knowledge takes various forms rangingfrom the nature of relationships between variables to a pragmatic understanding aboutimplementation and can be categorized along several useful dimensions such as publicnesstacitness and codifiability Researchers have characterized a larger typology of knowledgetypes important to the EBMgt process which include knowledge about the relationshipsbetween values and policy directions (ie know why) and knowledge about how to build andengage alliances for action (ie know who) (Ekblom 2002 Nutley et al 2003 Gasson 2005)

Probably the best known of these knowledge types is the individual tacit andqualitative form of ldquoknow-howrdquo (namely expertise) which draws on Polanyirsquos (1962)

2090

MD5610

explication of tacit knowledge In the case of experts classifying their guidance as ldquolowqualityrdquo is misclassifying the role that they play in decision making Experts are oftenexpected to engage in prediction ndash yet research suggests that experts are no better thannon-experts in prediction and making judgments outside of their domain as evidenced bytheir poor long-term forecasting (Tetlock 2017) and ldquofractionated expertiserdquo (Kahneman andKlein 2009) Instead experts play a crucial role in decision making by providing ldquovaluable andreliable information on the state of the knowledge in their field how to solve problems and onthe certainty of their answersrdquo (Meyer and Booker 2001) This tacit background knowledgealso ldquoallows individuals to limit the factors which they consider to be important in a decisionrdquoto systematically structure them and to discriminate among information (Bennett 1998)Experts also use ldquofast and frugalrdquo heuristics to process information (Gigerenzer and Goldstein1996) and are able to define a problem space and focus attention to its features (Chisholm1995) reducing the parameters considered in problem formulation

Proponents of a realist EBMgt platform offer a twofold response thereby settling thedebate about ldquobestnessrdquo as rank alone First call the process not evidence-based butevidence-informed to reinforce that decision makers must incorporate judgement Secondforego a strict ranking perspective widening the notion of evidence to incorporate aportfolio For instance a given portfolio might consist of ldquofour sources of informationpractitioner expertise and judgment evidence from the local context a critical evaluation ofthe best available research evidence and the perspectives of those people who might beaffected by the decisionrdquo (Briner et al 2009)

The portfolio is an excellent insight into the problem but seems to be incomplete in termsof what is ldquobestrdquo Increasing the amount of evidence within a given type leaves ldquothedisturbing possibility that when people experience uncertainty and gather information toreduce it this often backfires and uncertainty increasesrdquo (Doumlrner 1996 quoted in Weick2001) In other words more information is not always better ndash a knowledge regulationstructure is necessary to control epistemic uncertainty

Second consider an interpretation which evaluates ldquobestnessrdquo according to the exactness offit between a situation at a point in time and the evidence compiled for that situation Thiscontingency perspective would imply that the true conditions associated with decision makingsuch as the ldquocongruence between properties of knowledge properties of units and properties ofrelationships between unitsrdquo (Argote et al 2003) are known with enough certainty

Researchers associated with the Research Unit on Research Utilization at the Universityof St Andrews have modeled the problem in such a way (see eg Nutley et al 2007) In thisframework studies of organizational implementation successes and failures are aggregatedby disciplinary application to suggest combinations of organizational individualevidentiary source and search factors that promote high performance Althoughreasonable under stable conditions this approach becomes problematic under moreturbulent conditions Consider that finding the right evidence to support actions given acontingency of multiple social factors depends on knowing what those factors are andwhether and when they are permanent or changing features When epistemic uncertainty isthe highest the organization is least likely to be able to determine and adequately manage atleast some of the necessary factors of contingency

From what is known about the role of evidence in decision making the conditions tospecify fit are extensive including at a minimum the characteristics of

bull the evidence itself including its ability to represent and control aspects of the worldand its stickinesstransferability in an organizational context

bull the evidence source with special emphasis on legitimacy status and network position

bull the organizational search routines and procedures related to evidence searchand incorporation

2091

Evidence-based

management

bull the decision at hand especially whether focused on discovery (eg strategyinnovationnon-routine) or justification (eg operationalroutine)

bull the decision makers including their professional affiliation and dispositional factors(eg integrative complexity)

bull the organizationrsquos capability to translate evidence into action such as culture formalstructure and absorptive capacity and

bull the severity of the outcome errors that might accrue after an EBMgt processparticularly the immediacy and reversibility of results and the interdependencebetween target organizational or environmental components

In short the contingency solution is likely as difficult to specify as the problem itself andthe tension between exploration and exploitation looms

When the above conditions are clear the contingency framework could be sufficient andperhaps even preferable to produce the best evidence for management However forconditions to be clear the environment of the evidence use should be relatively stable(ie low turbulence) and the attendant uncertainty surrounding the decision relatively lowYet the often relatively unstable setting of healthcare presents the need for an intricateattention-orienting mechanism that both respects the social nature of evidence and thereflexivity necessary to characterize a decision and its environment

5 Best as varietyUncertainty is a special concept which is prone to confusion in common usage and itscharacter has important consequences for the manner in which an organization registers itspotential severity and the strategies to be enacted In strict logical usage uncertainty refersto the ldquoabsence [or] insufficiency of a certain kind of knowledgerdquo and is distinct fromvagueness and inexactness (Mattesich 1978) Wallsten and Budescu (1995) note thatuncertainty takes two forms it may be ldquodue to external quantifiable sources of randomvariation (aleatory) or to internal sources such as imperfect or incomplete information(epistemic)rdquo If the uncertainties affecting organizations are aleatory then faster higherquality collection of technical data and more adept statistical analysis are the key features incharacterizing solutions However if the uncertainty is of an epistemic character then theabsence or insufficiency of particular knowledge and the nature of knowledge in formingopinion and providing foundation and value are critical features in determining how anorganization should represent and respond to environmental threats (Quiggin 1993)

Improving performance in organizations requires contending with both forms ofuncertainty The promise of the received version of EBMgt appears to largely focus on thereduction of aleatory uncertainty through the accumulation of evidence ndash an issue roughlyakin to Pfeffer and Suttonrsquos (1999) ldquoknowing-doingrdquo gap In terms of performanceimprovement the contingency framework seems most applicable when decisions arerelatively algorithmic and programmable

However when the conditions are unclear or if the decision makers are unsure whether theconditions are clear then relying on the contingency specification of EBMgt becomesproblematic The problem is not merely an issue of bounded rationality but derives from themathematics of diversity and the epistemological problem of the underspecification of theoriesby evidence To the extent that we know what drives performance ldquowe should select the bestcollection on the basis of that information [hellip] [however] if we are not sure of what wersquoredoing we should err toward greater diversityrdquo (Page 2011) particularly ldquoon complex tasksthat involve multiple dimensions or variablesrdquo (Page 2017) The challenge of identifyingwhich parameters should be incorporated in an EBMgt strategy suggests a different solutionDrawing from the cybernetic tradition this paper extends a third interpretation of ldquobestnessrdquo

2092

MD5610

51 Insights from the cybernetics movementStarting in 1942 a series of interdisciplinary meetings between anatomists psychologistsphilosophers and social scientists sought to reconcile insights on how organizations exist inrelation to and under the constraints of complex systems (Dupuy 2000) The field wasdubbed cybernetics deriving from the ancient Greek ldquoΚυβερνήτηςrdquo (helmsman) a termrelated to steering ruling and government In addressing the way in which organismsself-regulate in complex environments the cyberneticists became fascinated with the way inwhich organizations sense measure and respond to the diversity of constraints theenvironment posed Drawing on Norbert Weinerrsquos work on how living systems exhibitcontrol functions and Claude Shannonrsquos theorem on disturbance in communicationchannels W Ross Ashby (1956) proposed the law of requisite variety which posited thatonly a variety in responses can ldquodestroyrdquo the variety in disturbances His great insight wasto focus on the notion of the variety of states and its consequences to a systemrsquos regulationof diverse environmental disturbances From that insight it should follow that creating andretaining diversity in knowledge types is a key way of increasing the organizationalcapacity to recognize relevant patterns of information from the environment

In the above sections this paper suggested that making inferences is a social process andthat knowledge and not evidence or information should be the focus of EBMgt Extendingsuch arguments through a requisite variety lens evokes Buckleyrsquos (19682008) suggestion

The concept of requisite deviation needs to be proffered as a high-level principle that can lead us totheorize a requisite of socio-cultural systems is the development and maintenance of a significantlevel of non-pathological deviance manifest as a pool of alternate ideas and behaviors with respectto the traditional institutionalized ideologies and role behaviors

In socio-cultural systems Buckley (19682008) argues that an organization can controlexternal variety by acquiring regulatory features such as information that allow it todiscriminate act upon and respond to its environment The cybernetic view of anorganization interacting with an open complex environment is predicated on theconceptualization of a social system as a ldquoset of elements linked almost entirely bythe intercommunication of informationrdquo (Zaltman et al 1973) A study of general systemsby complexity suggests that social systems are distinguished by the fact thatldquosymbol-processing actors who share a common social order organize informationfrom the environment into a knowledge structurerdquo (Anderson 1999 Boulding 1956)In socio-cultural systems subjective knowledge structures are formed interpersonally andthese form the regulatory mechanisms by which organizations discriminate act upon andrespond to uncertainty in the environment EBMgt can function as that technology whichaims to reduce organizational uncertainty

The exchange of organizational knowledge requires shared mental models and theldquoability to define relevant knowledge-domains is essential for collaborative sensemakingrdquo(Gasson 2005) Mental models are collective cognitive representations that range from adistributed configuration of representations with no overlap between individuals tooverlapping representations to identical representations among individuals (Klimoskiand Mohammed 1994) Maintaining a variety of knowledge types ensures that they areavailable to decision makers as a ldquoconsensually validated grammar for reducingequivocalityrdquo where equivocality is defined as ldquothe multiplicity of meanings which can beimposed on a situationrdquo (Weick 1979) The organizational complexity retained bymaintaining a diverse set of regulatory knowledge states can be conceived of as aldquosolution for a problem yet to be describedrdquo (Ahlemeyer 2001) Cognitive diversity inparticular increases perspective taking and ldquoimproves outcomes when making predictionsand solving problemsrdquo (Page 2017) In other words the variety of knowledge governs thesense made in sensemaking

2093

Evidence-based

management

The aim of pursuing variety in EBMgt is not only to ensure that individuals share andreconcile relevant knowledge but also to prevent the circumstance where regulators (ie people)systematically notice and represent problems in the same way Compiling more evidence doesnot necessarily imply compiling a wider range of knowledge types Likewise compilingevidence across a portfolio does not necessarily imply a balanced distribution of types acrossthe decision makers in the organization Individuals specialized to focus on one knowledge typedevote their attention to perceiving one element of the uncertainty that they apprehend whichunder the logic of appropriateness creates an organizational attention issue In the context ofreducing epistemic uncertainty variety assists the organization in balancing the ldquovaluation andlegitimization of issues and answersrdquo (Ocasio 1997) across the knowledge types reducing thedanger of becoming anchored or directing too much attention to a particular framing

In the healthcare setting technical evidence (ie quantified codified) displaysextraordinary rhetorical power to frame issues and drive decision making Withoutdedicated effort the organizationrsquos attention might naturally drift toward thesejustifications To prevent this drift decision makers can ensure the incorporation of otherforms of knowledge through processes of collaborative sensemaking By enforcing thereconciliation of arguments across knowledge types management can ensure that thetechnical rhetoric doesnrsquot crowd out relevant knowledge Under highly routine decisions orgiven a stable environment expanding one type of evidence or merely accruing perspectivefrom a given stakeholder may suffice However under unclear conditions the diversitybenefits of knowledge can only accrue through argument and discussion across individuals

Table II presents an illustration of a knowledge typology as applied to a decision toimplement a given safety culture intervention in a hospital setting Note that eachknowledge type confers a different perspective on the potential intervention Consistentwith the sociotechnical embeddedness of knowledge in evidence it is insufficient to slot onesource into one type of knowledge rather each source presents every type of knowledgeand decision makers together ascertain their value

Category ofknowledge Definition Example

Incorporating andreconciling

Know aboutproblems

The nature formulation naturalhistory and interrelations of socialproblems

Definition of safety culture andthe mechanisms by which itaffects communication in groups

ConceptsResearch definitionsand mechanisms

Know why(you mightimplement achange)

Explaining the relationshipbetween values and policydirections

Symbolic emotional ethical andcultural meaning of enacting asafety culture intervention

StoriesExplanations of whyit is important tochange

Know what(has worked)

What policies strategies orspecific interventions havebrought about desired outcomesat acceptable costs and with fewenough unwanted consequences

Existing safety cultureinterventions such as trainingsessions that have produceddesired outcomes

ExemplarsThe things that haveworked elsewhere

Know how (toput a changeinto practice)

Pragmatic knowledge aboutprogram implementation

How to practically implementand evaluate an effective safetyculture-focused intervention

SkillsThe know-how tosolve problems

Know who (toinvolve)

Building alliances for action Internal and externalcollaborators to advise andsupport a given safety cultureintervention

NetworksPeople who can adviseand support

Notes Table content developed based on Ekblom (2002) Gasson (2005) Nutley et al (2007) and Martelli (2012)

Table IIKnowledge typologyillustration

2094

MD5610

6 ConclusionLack of agreement about the fundamental nature of EBMgt has led to an impasse betweenproponents who take the endeavor as an inevitable incremental and realist approach todecision making and opponents who argue from a constructivist learning and powerpoliticsperspective This impasse prevents an extension of argumentation beyond ldquouse morerdquo vsldquowatch outrdquo While systematic reviews and empirical barriers remain important morerigorous research evidence and larger catalogues of contingency factors are themselvesinsufficient to solve underlying sociopolitical concerns Likewise current critiques couldbenefit from theoretical bridges that not only reinforce learning and sensemaking in realorganizations but also build on the spirit of the project and progress made towards bettermanagerial decision making This paper proposes a pragmatic framework to move beyondthe impasse refocusing the discussion on variety of knowledge while respecting themeaningful critiques by each side

By arguing from variety this paper suggests that the ldquobest available evidencerdquo can begenerated by ensuring that a broad range of knowledge types is elicited from and reconciledacross individuals Maintaining knowledge regulation states allows the organization tomanage attention and balance the valuation and legitimization from mechanismimplementation and policy knowledge

For practitioners this paper appreciates that organizational ldquodecision-makers generallydonrsquot seek evidence they seek an answer to their questionrdquo (Martelli 2012) As a resultEBMgt can be a disappointingly loose guide for decision makers because it ldquodoesnot prescribe the kind of evidence how to obtain it or what decisions should be maderdquo(Rundall and Kovner 2009) Under the best of circumstances when parameters are knownand fixed finding and applying the ldquobestrdquo evidence is elusive However under turbulent orotherwise nebulous conditions expecting practitioners to well-specify the characteristics oftheir particular decision process is untenable Additionally it highlights the tension inherentin the role of EBMgt in the complex service organizations of healthcare where the technicaldecision processes of healthcare management are distinct from technical decision processesgoverning the delivery of the healthcare product

The benefits to decision making should accrue when a diverse team reconcilesevidence for or against a course of action across each knowledge type A simplemanagerial intervention might be to distribute a structured evidence collection formwhich would be completed by all attendees prior to an administrative meeting The formrequires each attendee to compile and arrange evidence about a given decision on theagenda within each type (eg know what know why) For example the CMO and CNO ofa hospital each presents evidence for a safety culture intervention justifying theirperspective by reconciling evidence gathered within each of the knowledge types Whereevidence is lacking in a type attendees could critically examine the reasons for thedeficiency where it is unusually abundant attendees could consider whether it isconfirmatory or deceptive

This is not duplication it is a critical way to leverage the power of diversity to reduceepistemic uncertainty by eliciting tacit information giving voice to individuals andviewpoints that are less precise technical or aligned with the powerful and preventingdrift of organizational attention away from weak signals Potentially a Chief EvidenceOfficer could be responsible for supporting the collection and reconciliation of evidence tothat end

For researchers this paper argues that EBMgt is not merely a managerial tool but rathera technology ldquosituated in culture and embedded in historyrdquo ( Jasanoff 2012) Consequent tothe relationship between uncertainty complexity and diversity the ldquobestnessrdquo of evidenceis not determined through either rank or fit but rather through variety As social systemsare open and dynamic the best evidence is likely to vary as the problems specified and

2095

Evidence-based

management

solutions desired themselves vary This analysis places EBMgt in the tradition of thecybernetic regulation of social systems and the rational decision logic of appropriateness

Further research might make better use of existing cognitive diversity measures such asinterpretive ambiguity (Kilduff et al 2000) and knowledge heterogeneity (Rodan andGalunic 2004) to examine variety in EBMgt In this way it may be possible to explore howan organizationrsquos attention is misdirected to one or another type of evidence leading topotential strategic errors One such concept is a Type III error or the probability of resolvingat the expense of solving a problem or of ldquosolving the lsquowrongrsquo problem preciselyrdquo (Mitroffand Featheringham 1976) A second is the overadoption of innovation or the assumptionthat ldquoto adopt innovations is desirable behavior and to reject innovations is less desirable[hellip] [which] may not be true Overadoption often results from insufficient knowledgeoveradopters perceive the innovation as a panaceardquo (Rogers 1962) Overadoption could stemfrom the implementation of ldquobest practicesrdquo without social and contextual knowledge ndash aprocess observed in healthcare management (Arndt and Bigelow 1992 Denis et al 2002Kaissi and Begun 2008)

A critical goal of the EBMgt movement should be to help organizations develop andmaintain a common or at least commonly understood mental model for strategic decisionmaking This is especially true with respect to strategic improvement initiatives inhealthcare where prior research has shown the significance of knowledge intermediariesparticularly consulting groups such as The Advisory Board and Sg2 in ldquocompilingevidence developing alternatives or managing implementationrdquo (Martelli 2012) Under abevy of constraints to assessing contingency factors organizations adopting thesestandardized ldquomanagement bundlesrdquo risk falling into overadoption and innovationfailures as the diffusion of surgical checklists attests (eg Dixon-Woods et al 2011)Considering the ldquobest available evidencerdquo as variety offers a promising resolution bothpractically and theoretically

The field of EBMgt has made great strides both in convincing practitioners to useevidence and in tempering that drive with warnings about potential misapplicationsResolving the impasse rather than repeating it will require developing new foundationsand strategies for the project

References

Ahlemeyer HW (2001) ldquoManagement by complexity redundancy and variety in organizationsrdquoin Geyer RF and van der Zouwen J (Eds) Sociocybernetics Greenwood Press Westport CTpp 59-71

Anderson P (1999) ldquoComplexity theory and organization sciencerdquo Organization Science Vol 10 No 3pp 216-232

Argote L McEvily B and Reagans R (2003) ldquoManaging knowledge in organizations an integrativeframework and review of emerging themesrdquo Management Science Vol 49 No 4 pp 571-582

Arndt M and Bigelow B (1992) ldquoVertical integration in hospitals a framework for analysisrdquoMedicalCare Review Vol 49 No 1 pp 93-115

Arndt M and Bigelow B (2009) ldquoEvidence-based management in health care organizations acautionary noterdquo Health Care Management Review Vol 34 No 3 pp 206-213

Ashby WR (1956) An Introduction to Cybernetics Chapman amp Hall London


Balsiger J (2016) ldquoLogic of appropriatenessrdquo Encyclopaeligdia Britannica

Barends E Rousseau DM and Briner RB (2014) ldquoEvidence-based management the basicprinciplesrdquo Center for Evidence-Based Management Amsterdam

2096

MD5610

Begun JW Zimmerman B and Dooley K (2003) ldquoHealth care organizations as complex adaptivesystemsrdquo in Mick SS and Wyttenbach ME (Eds) Advances in Health Care OrganizationTheory Jossey-Bass San Francisco CA pp 253-288

Bennett RH III (1998) ldquoThe importance of tacit knowledge in strategic deliberations and decisionsrdquoManagement Decision Vol 36 No 9 pp 589-597

Boulding KE (1956) ldquoGeneral systems theory the skeleton of sciencerdquo Management Science Vol 2No 3 pp 197-208


Brint S (2001) ldquoProfessionals and the knowledge economy rethinking the theory of postindustrialsocietyrdquo Current Sociology Vol 49 No 4 pp 101-132

Buckley W (19682008) ldquoSociety as a complex adaptive systemrdquo Emergence Complexity andOrganization Vol 10 No 3 pp 86-112

Charon R and Wyer P (2008) ldquoNarrative evidence based medicinerdquo The Lancet Vol 371 No 9609pp 296-297

Chisholm D (1995) ldquoProblem solving and institutional designrdquo Journal of Public AdministrationResearch and Theory Vol 5 No 4 pp 451-492

Cunliffe AL (2003) ldquoReflexive inquiry in organizational research questions and possibilitiesrdquoHumanRelations Vol 56 No 8 pp 983-1003

Cunliffe AL and Jun JS (2005) ldquoThe need for reflexivity in public administrationrdquo Administration ampSociety Vol 37 No 2 pp 225-242

Denis JL Hebert Y Langley A Lozeau D and Trottier LH (2002) ldquoExplaining diffusion patternsfor complex health care innovationsrdquo Health Care Management Review Vol 27 No 3 pp 60-73

Dixon-Woods M Bosk CL Aveling EL Goeschel CA and Pronovost PJ (2011) ldquoExplainingMichigan developing an ex post theory of a quality improvement programrdquoMilbank QuarterlyVol 89 No 2 pp 167-205

Donaldson SI (2009) ldquoIn search of the blueprint for an evidence-based global societyrdquo in DonaldsonSI Christie CA and Mark MM (Eds) What Counts as Credible Evidence in Applied Researchand Evaluation Practice Sage Publications Los Angeles CA pp 2-18

Doumlrner D (1996) The Logic of Failure Recognizing and Avoiding Error in Complex SituationsMetropolitan Books New York NY

Dupuy J-P (2000) The Mechanization of the Mind Princeton University Press Princeton NJ

Edmondson AC (2010) ldquoMapping the failure landscape process deviations system breakdowns andunsuccessful trials as sources of improvement problem solving and innovation in teamsrdquo paperpresented at the 3rd International HRO Conference New Orleans LA January 9ndash10

Ekblom P (2002) ldquoFrom the source to the mainstream is uphill the challenge of transferringknowledge of crime prevention through replication innovation and anticipationrdquo in Tilley N(Ed) Analysis for Crime Prevention Crime Prevention Studies Vol XIII Criminal Justice PressMonsey NY pp 131-203

Food amp Drug Administration (2018) ldquoReal world evidencerdquo available at wwwfdagovScienceResearchSpecialTopicsRealWorldEvidencedefaulthtm (accessed February 25 2018)

Frederickson HG and Smith KB (2003) The Public Administration Theory Primer Westview PressBoulder CO

Garman AN Leach DC and Spector N (2006) ldquoWorldviews in collision conflict and collaborationacross professional linesrdquo Journal of Organizational Behavior Vol 27 No 7 pp 829-849

Gasson S (2005) ldquoThe dynamics of sensemaking knowledge and expertise in collaborativeboundary-spanning designrdquo Journal of Computer-Mediated Communication Vol 10 No 4available at httpsacademicoupcomjcmcarticle104JCMC10494614479

2097

Evidence-based

management

Gigerenzer G and Goldstein DG (1996) ldquoReasoning the fast and frugal way models of boundedrationalityrdquo Psychological Review Vol 103 No 4 pp 650-669

Gore J Banks A Millward L and Kyriakidou O (2006) ldquoNaturalistic decision makingand organizations reviewing pragmatic sciencerdquo Organization Studies Vol 27 No 7pp 925-942

Jasanoff S (2012) ldquoGenealogies of STSrdquo Social Studies of Science Vol 43 No 3 pp 435-441

Kahneman D and Klein G (2009) ldquoConditions for intuitive expertise a failure to disagreerdquo AmericanPsychologist Vol 64 No 6 pp 515-526

Kaissi AA and Begun JW (2008) ldquoFads fashions and bandwagons in healthcare strategyrdquo HealthCare Management Review Vol 33 No 2 pp 94-102

Keller AC (2009) ldquoCredibility and relevance in environmental policy measuring strategies andperformance among science assessment organizationsrdquo Journal of Public AdministrationResearch and Theory Vol 20 No 2 pp 357-386

Kilduff M Angelmar R and Mehra A (2000) ldquoTop management team diversity and firmperformance examining the role of cognitionsrdquo Organization Science Vol 11 No 1 pp 21-34

Klimoski R and Mohammed S (1994) ldquoTeam mental model construct or metaphorrdquo Journal ofManagement Vol 20 No 2 pp 403-437


Kovner AR Elton JJ and Billings J (2000) ldquoEvidence-based managementrdquo Frontiers of HealthServices Management Vol 16 No 4 pp 3-24

Kuhn T (1962) The Structure of Scientific Revolutions Chicago University Press Chicago IL

Laursen K and Salter A (2006) ldquoOpen for innovation the role of openness in explaining innovationperformance among UK manufacturing firmsrdquo Strategic Management Journal Vol 27 No 2pp 131-150

Learmonth M and Harding N (2006) ldquoEvidence-based management the very ideardquo PublicAdministration Vol 84 No 2 pp 245-266

Mankelwicz J and Kitahara R (2008) ldquoPropositions to guide evidence-based decision-makingrdquoJournal of Business Economics amp Research Vol 6 No 10 pp 41-56

Martelli PF (2012) An Argument for Knowledge Variety in Evidence-Based Management Universityof California Berkeley Berkeley CA

Matheson D and Matheson J (1998) The Smart Organization HBS Press Cambridge MA

Mattesich R (1978) Instrumental Reasoning and Systems Methodology Reidel Publishing Boston MA

Meyer MA and Booker JM (2001) Eliciting and Analyzing Expert Judgment Society for IndustrialMathematics Philadelphia PA

Mitroff II and Featheringham TR (1976) ldquoTowards a behavioral theory of systemic hypothesis-testing and the error of the third kindrdquo Theory and Decision Vol 7 No 3 pp 205-220

Morrell K and Learmonth M (2015) ldquoAgainst evidence-based management for managementlearningrdquo Academy of Management Learning amp Education Vol 14 No 4 pp 520-533

Mowles C (2011) Rethinking Management Radical Insights from the Complexity SciencesGower Press Burlington VT pp 17-20

Nutley SM Walter I and Davies HTO (2003) ldquoFrom knowing to doing a framework forunderstanding the evidence-into-practice agendardquo Evaluation Vol 9 No 2 pp 125-148

Nutley SM Walter I and Davies HTO (2007) Using Evidence How Research Can Inform PublicServices The Policy Press Bristol

Ocasio W (1997) ldquoTowards an attention-based view of the firmrdquo Strategic Management JournalVol 18 No S1 pp 187-206

2098

MD5610

Oxford English Dictionary (1989) ldquoevidence nrdquo 2nd ed available at wwwoedcomoed200079136(accessed July 18 2018)

Page SE (2011) Diversity and Complexity Princeton University Press Princeton NJ

Page SE (2017) The Diversity Bonus Princeton University Press Princeton NJ

Pfeffer J and Sutton RI (1999) The Knowing-Doing Gap How Smart Companies Turn KnowledgeInto Action HBS Press Cambridge MA

Pfeffer J and Sutton RI (2006) ldquoEvidence-based managementrdquo Harvard Business Review Vol 84No 1 pp 62-74

Plsek PE and Greenhalgh T (2001) ldquoThe challenge of complexity in health carerdquo British MedicalJournal Vol 323 No 7314 pp 625-628

Polanyi M (1962) ldquoTacit knowing its bearing on some problems in philosophyrdquo Reviews of ModernPhysics Vol 34 No 4 pp 601-616

Potts M Prata N Walsh N and Grossman A (2006) ldquoParachute approach to evidence basedmedicinerdquo British Medical Journal Vol 333 No 7570 pp 701-703

Quiggin J (1993) Generalized Expected Utility Theory The rank-dependent model KluwerAcademic Publishers Boston MA

Quiggin J (2004) ldquoThe precautionary principle and the theory of choice under uncertaintyrdquo workingpaper University of Queensland Brisbane 11 January

Rodan S and Galunic DC (2004) ldquoMore than network structure how knowledge heterogeneityinfluences managerial performance and innovativenessrdquo Strategic Management Journal Vol 25No 6 pp 541-562

Rogers EM (1962) Diffusion of Innovations Free Press of Glencoe New York NY

Rousseau D (2016) ldquoEvidence-based managementrdquo available at httpsgroupsgooglecomforumtopicevidence-based-managementt1G08LUIu7Y (accessed September 10 2017)


Rundall TG and Kovner AR (2009) ldquoEvidence-based management reconsidered 18 months laterrdquoin Kovner AR Fine D and DrsquoAquila R (Eds) Evidence-based Management in HealthcareHealth Administration Press Chicago IL pp 79-82

Rundall TG Martelli PF McCurdy R Graetz I Arroyo L Neuwirth EB Curtis P Schmittdiel JGibson M and Hsu J (2009) ldquoUsing research evidence when making decisions views of healthservices managers and policymakersrdquo in Kovner A DrsquoAquila R and Fine D (Eds) Evidence-based Management in Healthcare Health Administration Press Chicago IL pp 3-16

Rynes SL and Bartunek JM (2017) ldquoEvidence-based management foundations developmentcontroversies and futurerdquo Annual Review of Organizational Psychology and OrganizationalBehavior Vol 4 No 1 pp 235-261

Rynes SL Rousseau DM and Barends E (2014) ldquoFrom the guest editors change the world teachevidence-based practicerdquoAcademy ofManagement Learning ampEducation Vol 13 No 3 pp 305-321

Sherman RE Anderson SA Dal Pan GJ Gray GW Gross T Hunter NL LaVange LMarinac-Dabic D Marks PW Robb MA and Shuren J (2016) ldquoReal-world evidence ndash what isit and what can it tell usrdquo The New England Journal of Medicine Vol 375 No 23 pp 2293-2297

Tang KC Ehsani JP and McQueen DV (2003) ldquoEvidence-based health promotion recollectionsreflections and reconsiderationsrdquo Journal of Epidemiology and Community Health Vol 57No 11 pp 841-843

Tetlock PE (2017) Expert Political Judgment Princeton University Press Princeton NJ

Turner S (2004) ldquoQuasi-science and the staterdquo in Stehr N (Ed) Governing Science in ComparativePerspective Transaction Publications New Brunswick NJ pp 241-268

Wallsten TS and Budescu DV (1995) ldquoA review of human linguistic probability processing generalprinciples and empirical evidencerdquo Knowledge Engineering Review Vol 10 No 1 pp 43-62

2099

Evidence-based

management

Walshe K and Rundall TG (2001) ldquoEvidence-based management from theory to practice in healthcarerdquo Milbank Quarterly Vol 79 No 3 pp 429-457

Weick KE (1979) The Social Psychology of Organizing 2nd ed Addison-Wesley Reading MAWeick KE (2001) ldquoGapping the relevance bridge fashions meet fundamentals in management

researchrdquo British Journal of Management Vol 12 No S1 pp S71-S75Zaltman G Duncan R and Holbek J (1973) Innovations and Organizations John Wiley amp Sons

New York NY

Corresponding authorPeter F Martelli can be contacted at pmartellisuffolkedu


2100

MD5610

Conceptual modelling of theflow of frail elderly through

acute-care hospitalsAn evidence-based management approach

Silvia BruzziDepartment of Economics and Business Studies University of Genova

Genoa ItalyPaolo Landa

Medical School University of Exeter Exeter UK andElena Tagravenfani and Angela Testi

Department of Economics and Business Studies University of GenovaGenoa Italy

AbstractPurpose ndash The ageing of the worldrsquos population is causing an increase in the number of frail patientsadmitted to hospitals In the absence of appropriate management and organisation these patients risk anexcessive length of stay and poor outcomes To deal with this problem the purpose of this paper is to proposea conceptual model to facilitate the pathway of frail elderly patients across acute care hospitals focussed onavoiding improper wait times and treatment during the processDesignmethodologyapproach ndash The conceptual model is developed to enrich the standard flowchart of aclinical pathway in the hospital The modified flowchart encompasses new organisational units and activitiescarried out by new dedicated professional roles The proposed variant aims to provide a correct assessment offrailty at the entrance a better management of the patientrsquos stay during different clinical stages and an earlydischarge sending the patient home or to other facilities avoiding a delayed discharge The model iscompleted by a set of indicators aimed at measuring performance improvements and creating a strongdatabase of evidence on the managing of frail elderlyrsquos pathways providing proper information that canvalidate the model when applied in current practiceFindings ndash The paper proposes a design of the clinical path of frail patients in acute care hospitalscombining elements that according to an evidence-based management approach have proved to be effectivein terms of outcomes costs and organisational issues The authors can therefore expect an improvement inthe treatment of frail patients in hospital avoiding their functional decline and worsening frailty conditionsas often happens in current practice following the standard path of other patientsResearch limitationsimplications ndash The framework proposed is a conceptual model to manage frailelderly patients in acute care wards The research approach lacks application to real data and proof ofeffectiveness Further work will be devoted to implementing a simulation model for a specific case study andverifying the impact of the conceptual model in real care settingsPractical implications ndash The paper includes suggestions for re-engineering the management of frailelderly patients in hospitals when a reduction of lengths of stay and the improvement of clinical outcomesis requiredOriginalityvalue ndash This paper fulfils an identified need to study and provide solutions for the managementof frail elderly patients in acute care hospitals and generally to produce value in a patient-centred modelKeywords Conceptual model Hospital management Patient flow Evidence-based managementClinical pathway Frail patientsPaper type Research paper

1 Introduction to the problem under studyDuring the last decades the demand for healthcare has faced deep changes due to severalfactors such as an ageing population The number of older persons is rapidly increasingand forms a growing share of the population all over the world people aged 60 years or over



0025-1747DOI 101108MD-10-2017-0997

Received 15 October 2017Revised 9 March 2018

13 May 201814 July 2018



2101

Flow offrail elderly


numbered 962m in 2017 (more than twice the number in 1980) and are expected to doubleagain by 2050 reaching 2bn The number of people aged 80 years or over is projected toincrease more than threefold between 2017 and 2050 rising from 137m to 425mThis growth is faster in Europe and in Northern America where in 2050 older people areexpected to account for 35 and 28 per cent of the population respectively (United NationsDepartment of Economic and Social Affairs Population Division 2017)

The increase of the older population often with chronic pathologies and multimorbiditiesproduces a frailer and more dependent population (van Eeden et al 2016) From a clinicalperspective frailty is considered the most problematic expression of population ageing(Clegg et al 2013) Even though a unanimous international definition of and consensus onhow to measure frailty does not yet exist it is recognised that frailty develops as aconsequence of the age-related reduction in physiological reserve and the ability to resistenvironmental stressors This leads to the elderly being vulnerable to relatively minor stressorevents entailing a high risk of falls disability hospitalisation and mortality (Fried et al 2001)

These risks are generally recognised to be associated with age (Song et al 2010) As aconsequence of population ageing frail patients are increasing and will continue to increase inthe future demanding new and more complex care solutions (McColl-Kennedy et al 2012)

Unlike acute patients frail patients are chronic and never exit the healthcare system oncethey start their care pathway Hence they begin a continuum of care (primary secondaryand home care) and a continuum of relationships that involve a large number of actors withdifferent skills and roles Consequently the way these relationships are organised andmanaged decisively impacts the outcome of the care solutions adopted

Under these pressures from the demand side the supply sidersquos ability to provideappropriate organisational solutions depends on the healthcare systemsrsquo ability to organisethe network of services around these patientsrsquo needs They should do so according to a newpatient-centred approach (Chewning and Sleath 1996 Mead and Bower 2000) that linksdifferent care settings (Black and Gallan 2015) In this network the design and constructionof integrated healthcare systems becomes a critical issue

The contribution of this paper is a presentation of a conceptual model for the hospitalmanagement of frail patients This conceptual model meets the specific needs of frailpatients offering them a more appropriate care including the use of different professionalroles (hospitalist case manager and bed manager) units (intermediate care area (ICA) andcentral discharge unit) and tools (comprehensive frailty assessment (CFA)) that work jointlyto improve the clinical paths of frail patients In the existing literature several authorsprovided evidence of single elements through trials or simply using observational dataThe main idea of this work is to fill the gap left by the large existing literature that discussesdifferent approaches by considering all of these elements together using a conceptualmodel to represent the flows of frail patients in acute care hospital wards The model alsoprovides an approach based on both patient and hospital processes in order to improve theoverall hospital performance and patient outcome It uses a dedicated clinical pathwayfor frail elderly patients with the introduction of facilitators tools and units that are usuallynot present in hospitalsrsquo organisation even if the need for these facilitators is rising inhospital settings

The assumption of this paper is that the acute care ward still plays a central role insuccessful integrated patient-centred solutions since it is a major crossroads of patientsand therefore must adopt management principles and tools to manage frail patientFrail patients spend some time in acute hospital wards coming from and returning to theirown residence or to less intensive-care levels (nursing homes post-acute facilities socialcare units caregivers etc) (Philp et al 2013)

In this network of services at different levels the role of the acute ward is stillcrucial since the hospital stay is often a major cause of problems The waiting and the

2102

MD5610

organisational bottlenecks cause patients and their familiesrsquo distress which risks aregression of patientsrsquo health and mental conditions Appropriately managing the flow offrail patients in acute hospital wards can be considered a prerequisite for efficientlymanaging the flows within the broader health system This management can also lead to thedecongestion of acute care hospitals with consequent positive effects in terms of careappropriateness and a reduction in healthcare costs

This study aims at contributing to this by proposing a new conceptual model fordesigning the flow of hospital care delivery to frail elderly patients in order to facilitate theirclinical pathway across acute care hospitals their discharge and if necessary theiradmission to another facilityservice (nursing homes social care units etc)

The conceptual model is expected to be able to gather evidence about its ability toprovide frail patients with appropriate and affordable acute care and thus to contribute tothe construction of a model of evidence-based practices for frail patients Indeed thecontribution of the conceptual model provides new insights into evidence-basedmanagement (EBMgt) EBMgt helps the decision-maker to identify the organisationalstrategies relative structures and change-management practices that enable healthcareprofessionals and managers to provide evidence-based care (Walshe and Rundall 2001Shortell et al 2007) In EBMgt healthcare managers make organisational decisions usinginformation provided by social science and organisational research (Lemieux-Charles andChampagne 2004 Rousseau 2005) considering the best scientific evidence available in theliterature The literature analysed shows the limited number of integrated solutionscapable to face problems deriving from hospital frail patientsrsquo admissions management anddelayed discharges

According to the principles of evidence-based practice evidence has to be taken intoaccount from four different sources the scientific literature the organisation thepractitioners and the stakeholders (Barends et al 2014) Our approach included three of thefour sources and the fourth only in an indirect way The scientific literature source consistsof evidence from empirical studies published in academic journals and in our approach isrepresented by the literature on the different tools adopted to face frailty emergencydepartment (ED) boarding complex patient management and discharge

The organisation source consists of representing the organisation using data facts andfigures gathered from it In our approach the organisation is represented by the analysis ofhospital flows and the organisation of hospital activity The practitionerrsquos componentconsists of the professional experience and judgment of the practitioner about the approachIn the analysis presented in this paper we interviewed hospital managers physicians andward staff to understand the organisation and to define the hospital flow of frail patientsand the main sources of bottlenecks in the care process

Finally the stakeholder component encompasses the values and concerns of the peopleinvolved the decision are evaluated only by a set of indicators that prove the ex post effects(Porter 2010) In this way the stakeholder principle is indirectly considered by theproposition of a set of indicators The indicators measure the outcome for the people affectedby the decision ndash in this case the patients and the hospital ndash and consider a reduction inpatient boarding and bed blockers and a better management of frail elderly patients whichreduces inappropriate discharges and repeated hospital admissions and leads to a better useof resources

The paper is organised as follows Section 2 focusses on the debate concerning thedefinition and measurement of frailty and its increasing relevance in healthcare systemswith reference to the major critical issues of frail patientsrsquo care in acute care hospitalsIn Section 3 we review some evidence-based instruments (ie organisational roles units andtools) to face the above-mentioned critical issues In Section 4 we describe the systemldquoas isrdquo and in Section 5 we develop our conceptual model with a schematic flowchart

2103


representation where roles units and changes proposed are introduced along with a set ofquality indicators aimed at evaluating the impact of our model In Section 6 someconcluding remarks for future research are discussed

2 Frail patients in acute care hospitalsThe recent rise in life expectancy and advances in medical technology are increasing thenumber of elderly hospitalised which account for more than 50 per cent of hospitaladmissions in industrialised European countries (Eurostat 2016) We expect that anumber of these older patients present some features that will worsen hospital outcomessuch as an increased length of stay functional decline iatrogenic complication cognitiveimpairment and so on They are commonly considered a subgroup frailer than otherpatients One of the first definitions of the concept of frailty dates back to about 30 yearsago when the American Medical Association reported the growth of ldquofrailrdquo vulnerable oldadults as the group of patients that presents the most complex and challenging problems(Scott et al 1990) Nowadays the current practice in health is to deal with the problem ofmeeting the needs of frail patients Frailty is a term widely used to denote amultidimensional syndrome of a loss of reserves (energy physical ability cognitionhealth) that gives rise to vulnerability This appears to be a valid construct but its exactdefinition remains unclear (Rockwood et al 2005)

Indeed frailty overlaps with other conditions in particular with ldquodisabilityrdquo andldquocomorbidityrdquo The first condition refers to a situation in which the person has difficultycarrying out activities required to live independently the so-called activities of daily living(ADL) originally proposed in the 1950s and in current use all over the world after beingrevisited by many researchers (Katz 1963) It also refers to a more complex set ofbehaviours such as telephoning shopping food preparation housekeeping doing thelaundry using transportation and using medicine the so-called instrumental activities ofdaily living proposed by Lawton and Brody (1969) Scales are used to assess an individualrsquosindependent living skills and measure the functional ability as well as deteriorations andimprovements over time

The second condition comorbidity consists of the presence of two or more chronicdiseases This condition is rather simple to measure and quantify The prevalence ofmultimorbidity is over 60 per cent worldwide and is probably greater than 80 per centamong people aged ⩾85 years (Salive 2013) These two conditions however still do notcoincide with frailty The latter refers rather to a state of high vulnerability includingdisability and comobordity but also to a risk factor due to the geriatric problems of olderage such as falls and incontinence This situation is usually not reported in administrativedata or billing systems and requires a clinical assessment or patient self-report methodsFrailty therefore is an aggregate expression of risk deriving both from age and fromthe accumulation of many problems not only clinical conditions All these dimensionsshould be seen as distinct which would help explain why some persons with frailty have noadverse outcomes some frail persons have no chronic conditions and some persons with asingle chronic condition are frail and vulnerable with poor outcomes

In order to get some insight into the complexity of estimating the prevalence of frailpatients inside a hospital we refer to Figure 1 where the results of a study are reported(Fried et al 2001) separating the three different dimensions The study identified368 patients out of 4317 as frail (85 per cent) and further identified overlaps withcomorbidities and disabilities Figure 1 also shows how only about 10 per cent of patientswith comorbidity have frailty characteristics

A more recent study provides higher values for the prevalence of frailty declaring thatapproximately 10 per cent of people aged over 65 and 25ndash50 per cent of those aged over85 are living with frailty (Lincolnshire Community Health Services 2015) This evidence is

2104

MD5610

in line with the current demographic increase of expected life duration engendering acorresponding increase in the period during which one lives in a condition of frailty We cantherefore expect that acute care hospitals will admit a greater number of frail peoplerequiring urgent organisational interventions to face their new needs What is generallylacking in our opinion is an additional assessment of socioeconomic conditions which arefurther determinants of frailty and which result in poor outcomes with few exceptions Thisis reported in a study (Rodriacuteguez-Mantildeas et al 2013) that recognises that frailty may involvenot only physical components but also social aspects

Frailty needs to be appropriately managed inside the acute care hospital by designingappropriate pathways which are expected to work together with trajectories for acute andnot-frail patients The debate concerning appropriate care for frail patients has traditionallyfocussed mainly on the development of low clinical content and low-cost intensityinterventions such as home care day care nursing homes and social care in order todecongest acute care hospitals and also on the development of geriatric units or unitsspecialised in elderly needs inside acute care hospitals (Fox et al 2013) The problem in ourview should be faced by taking into account the entire care process of the patient whateverthe stay ward is orthopaedic urology or general surgery and not only medicine wards

In order to contribute to and enrich the debate our paper adopts a process-based viewaimed at optimising frail elderly patient flows inside acute care hospitals in order to reducetheir admission time and length of stay better coordinate multidisciplinary interventionsencourage speed discharging and if necessary admission to other long-term facilities andeventually reduce the risk of adverse events Hospitalised frail patients in particular are at ahigher risk of adverse events which when they occur complicate patientsrsquo health status andlead to functional impairment or death (Brennan et al 1991 Leape et al 1991 Madeira et al2007 Szlejf et al 2012) Therefore it is critical to minimise the length of time that suchpatients spend in acute care hospitals When designing solutions for new care settings andclinical pathways able to improve these patient flows we focussed on the three most criticalmoments during frail patientsrsquo acute care hospital stay which concern the admission thehospital stay and the discharge Frail patients are often already under the care of otherfacilities (community hospital nursing home domiciliary care) where they come from when

Disability1 ADL(n=67)

(n=79)

(n=21)

(n=98)

(n=170)

(n=196)(n=2131)

Comorbidity

215

57 462

266

Frailty

Source Fried et al (2001)

Figure 1Venn diagram

showing the overlapbetween frailtydisability and

comorbity conditions

2105


admitted and where they need to go back to when discharged For this reason well-designedflows inspired by the transitional care approach are very important Transitional care aimsin fact at promoting a safe and timely passage of patients between levels of healthcare andacross care settings The American Geriatric Society defines transitional care as ldquoa set ofactions designed to ensure the coordination and continuity of healthcare as patients transferbetween different locations or different levels of care within the same locationrdquo (Colemanand Boult 2003) This is particularly important for frail elderly patients that need to movefrequently within different health care settings for their health status (Coleman Boult 2003Naylor et al 2004)

For frail patients who cannot be transferred home for any reason discharge from anacute care hospital can be very complex and difficult thus resulting in inappropriatehospital stays and increasing the phenomenon of bed blockers ie elderly patients whocannot go back home for any reason and must remain in hospital until a bed in anotherinstitution ( facility) is available (Benson et al 2006 Manzano-Santaella 2010) or delayeddischarges (Bryan et al 2006) Delayed discharges are in fact one of the most critical issuesconcerning frail patients in acute care hospitals Naylor and Keating (2008) report at thisregard that many factors contribute to gaps in care during critical transitions among thempoor communication incomplete transfer of information and the absence of a single personto ensure continuity of care

The flows should be improved in order to reduce older patientsrsquo stay in the hospitaladmitting only those older patients who really need hospital treatment minimising delaysfor those who are admitted and discharging them from hospitals as soon as possibleie when patients are clinically stabilized to be discharged Different solutions(organisational units professional roles and tools) have been discussed by the literatureand introduced in practice to reduce hospital admissions or length of stay of frail elderlypatients In the following section the most important and evidence-based organisationalinterventions are described

3 Evidence-based tools a literature reviewIn recent years alternative organisational changes have been proposed in many countries inorder to facilitate the clinical pathways of patients inside acute care hospitals Thesechanges have paid attention to the transition of care towards other healthcare facilities thusdeveloping or improving existing integrated care models (World Health Organization 2016)

In this section the changes that are most suitable to facilitate the path of frail patients aredescribed in detail We attempted to find evidence for their effectiveness in the literaturealthough unfortunately proof is often neglected in the case of organisational toolsWe choose the following organisational interventions addressed to frailty assessment theintroduction of new professional roles (case manager hospitalist and bed manager) and neworganisational units (an ICA and a central discharge gateway (CDG)) Based on an analysisof the literature these interventions seem able to reduce emergent patientsrsquo admission timeand length of stay speed up the discharging process and if necessary the patientrsquosadmission to other long-term facilities Each intervention is briefly explained after whichthe relevant literature is discussed paying particular attention to main findings in terms ofproof of impact

31 Frailty assessment and comprehensive frailty assessment (CFA)Once the frail elderly patient enters the acute care hospital (both as elective or emergent) afrailty assessment must be carried out by an specially designed elderly care assessment unitor commission in order to determine hisher medical psychological and functionalcapabilities (Ellis et al 2011) When compiling the assessment the patient is assigned a codethrough which respecting hisher privacy heshe is placed in an tailored path where a

2106

MD5610

specific professional figure ( front-end staff ) is in charge of himher A continuous flow ofinformation monitoring the patientrsquos activity is ensured (back office) The tracking andtracing system of the patient informs any actor or part of the system in advance about thepresence (or arrival) of a patient who needs specific care

The assessment can be done by means of different tools a card an electronic device(eg RFID) etc As different definitions of frailty are provided so different algorithms areutilised (Woo et al 2015)

Each algorithm and each scale is assessed through consultation with clinicians andhospital managers considering different risk factors such as comorbidities and geriatricconditions The assessment has to be done as soon as the patient enters the hospital in orderto have the information on hisher clinical and frailty condition available so as to activate theservices dedicated to patient care sooner

The frailty first aid (FFA) should be present in the emergency room 24 hours a dayThe FFA immediately alerts a commission called the CFA The CFA conducts amultidimensional medical functional psycho-social and environmental evaluation of theolder personrsquos problems and resources in order to develop a personalised path inside thehospital assigning a case manager a hospitalist a bed manager and all the other functionscharged with following the frail patient Most hospitals have some form of initial frailtyassessment in place although these are rarely integrated with other hospital processes andcarry many different denominations (Stuck et al 1993)

Frailty assessment has always proved to be effective One of the first studies dates backto about 20 years ago (Stuck 1997) A randomised controlled study in unselected olderpatients admitted to an acute care hospital found that thanks to the assessment patientsrsquofunction at hospital discharge was improved and the risk of nursing home admissionsdecreased in patients receiving integrated geriatric care as compared to patients receivingthe usual acute hospital care Another trial found a statistically significant reduction ofhospital readmissions and cost savings in the intervention group compared with controls(Stuck 1997)

The most recent and convincing results are reported in a systematic review (Ellis et al2011) where 22 trials evaluating 10315 participants in six countries were identifiedPatients who underwent a specific frailty assessment were more likely to be alive and intheir own homes after up to six months and at the end of a scheduled follow-up (median12 months) when compared to those who received general medical care

This systematic review was recently updated and completed (Ellis et al 2017) in order toalso estimate the cost-effectiveness of frailty assessment While CFA may lead to a smallincrease in costs evidence of cost-effectiveness is uncertain due to imprecision andinconsistency in the studies

In conclusion the CFA proposed herein is a multidimensional early assessment toolcrucial to guiding frail people towards the proper diagnostic and therapeutic process insidethe hospital CFA results in a coordinated and integrated treatment plan until discharge thesubsequent follow-up and the transitional step towards other care settings (home nursinghomes and so on) The frailty assessment is effective and is the first step of a care approachfor detecting frailty in the community allowing targeted intervention to potentially delaydecline and future disability This means that like other suggested tools in the paperCFA should be integrated coordinated and guided by a unique frailty team that supportsthe work of central health management

32 Case managerOf the professional roles introduced in the healthcare delivery practice and studied by theliterature the case manager and the hospitalist seem to best facilitate the clinical trajectoriesof frail patients

2107


In our opinion both figures should be activated at the beginning of the care process andassigned to the patientrsquos care one nurse (the case manager) mostly dedicated to theassistance aspects of the care and one physician (the hospitalist) mostly dedicated tothe clinical aspects Both originated in a US context and aim at meeting the needs of serviceintegration They also offer cost control and over-performance deterrence and help ensurethe continuity of care (Haggerty et al 2003) There is no unique definition of case managersbut they are primarily focussed on achieving quality while controlling costs throughcoordination and the management of care

The primary tasks of a case manager are therefore to assess the patientrsquos and carerrsquosneeds develop tailored care plans organise and adjust care processes accordingly monitorthe quality of care and maintain contact with the patient and carer (Singh and Ham 2006)

Case management developed in Europe ( first in the UK) when the management and careof patients with long-term conditions increasingly deinstitutionalised became a priority inthe financially restricted European public health systems In those systems casemanagement is considered a solution for the care of the elderly and dependent population inorder to reduce emergency and acute hospital bed use (Reilly et al 2010)

While case management is mostly developed in acute care settings it is primarily aresponse to those patients who need coordinated actions taken by a professional Thisprofessional mostly has a background in nursing or social works (White and Hall 2006) andtakes action according to a patient-centred logic of integrating healthcare and social servicesprovided by different players

Evidence shows that case management decreases the number of hospital (re)admissionsand improves patient satisfaction while evidence on the cost-effectiveness of casemanagement remains controversial (Curry and Ham 2010) Indeed case managementinterventions reduced hospital admissions and the length of stay in hospitals withcorresponding savings in total healthcare costs (Leung et al 2004)

33 HospitalistThe hospitalist is another professional role coming from the organisational healthcarelandscape of the USA introduced in 1996 with the aim of creating a generalist within thehospital responsible for managing the care of hospitalised patients The hospitalist assumesthe role of a general practitioner (GP) within the hospital (Wachter and Goldman 1996)Unlike the case manager who is born out of the need to cope with the progressivedeinstitutionalisation of patients and hence is mostly a nurse the hospitalist is a physicianspecialised in supervising a patientrsquos care during a hospital stay This person receivespatients from the GP is their personal medical advisor and manager of their health for theduration of their hospital stay and then returns the patients to the GP after discharge(Cammarata 2005)

After only five years since its introduction the hospitalist has been shown to beassociated with significant reductions in costs (134 per cent) and hospitalisation (166 per cent)(Wachter 2002 Wachter and Goldman 2002)

Subsequently this figure of the generalist has spread very quickly and 20 years laterhospitalists are present in 75 per cent of US hospitals (Wachter and Goldman 2016)

Nowadays the hospitalist is common in many US hospitals where they play a key roleand collaborate with other medical specialists and the administration increasinglytaking on a leading role in quality improvement programs (Yousefi and Wilton 2011)The hospitalist model of care delivery inside the hospital became a point of reference forCanada as well (Yousefi and Wilton 2011) and then for other countries such as Singapore(Hock Lee et al 2011) and Brasil (Schnekenberg 2011) Especially at the beginningsome criticism was raised because hospitals created a discontinuity of care between thehospitalist and the figure of the GP in the US-managed care system (Goldmann 1999)

2108

MD5610

More recently other criticisms were formulated with regard to costs the hospitalistallows for a decrease in the duration of hospital stays and therefore costs of the hospitalbut shifts these costs to post-hospital care and increases the probability of readmission(Kuo and Goodwin 2011) However opposite results come from other studies where it isshown that hospitalists significantly reduce hospital stays without increasing costs(Rachoin et al 2012)

What is certain is that most trials and tests prove that a hospitalist can decrease the lengthof stay thus reducing hospitalisation risks for frail patients There still is little proof howeverwith a few exceptions that the quality of care improves (Yousefi and Wilton 2011)

34 Bed managerBed management has been introduced to face ED boarding which is a major reason for EDovercrowding and elective admission postponements (Bagust et al 1999) Emergencypatient admissions into wards and patient boarding were widely reported in the literatureduring the last decades (Bagust et al 1999 Proudlove et al 2007)

The main criticalities regard two central aspects how to guarantee the completion of acare pathway in a timely and proper manner for emergency patients that were alreadydiagnosed in ED and are waiting to be admitted into inpatient wards and how to avoid thedelay of care delivery for elective patients waiting to be admitted to the hospital to receivetheir timely and proper care

A suggested solution is the introduction of the bed manager a dedicated professionalrole that keeps a balance between a flexibility that allows for admitting emergency patientsand a high bed occupancy (Green and Armstrong 1994) Its main task is to report at giveninterval time slots during the day the volume census and occupancy rates of the availableward-stay beds in order to synchronise the expected discharges ie bed supply with theexpected admissions from ED ie bed demand (Haraden and Resar 2004)

When analysing the literature we found few published academic studies reporting onthe performance of bed management or its effectiveness in terms of patient flow andexperience In a study proposed by Howell et al (2008) a decrease of the ED throughputtimes is reported which is mainly due to a reduction of about 21 per cent (approximately onehour and half ) of the time spent inside ED by patients waiting to be admitted This effectwas still larger (28 per cent) in the case of transferring patients from ED to intensive careunits (Howell et al 2010) Again the percentage of hours during which the ED had to divertambulances due to ED crowding and a lack of intensive-care unit beds decreased by 6 and27 per cent respectively (Howell et al 2008)

35 Organisational unitsThe first organisational unit selected to deal with the problem of frail patient management isthe ICA The ICA is usually located downstream from the acute area (which is in turndivided into a medical and surgical area) and is inspired by the community or countryhospital model directed to deliver sub-acute care seeking to reduce the number ofinappropriate admissions to acute care hospitals and to facilitate the discharge of patientsfrom acute care (Pitchforth et al 2017)

Given the extent of definitions and operational experiences in the literature (Melis et al2004 Steiner 2001) it is worth referring to the British Geriatric Society which includes inintermediate care services that are limited in time (normally no longer than six weeks)involving cross-professional working and targeted at people who would otherwise faceunnecessarily prolonged hospital stays or inappropriate admission to acute inpatientlong-term residential or continuing NHS inpatient care Using the framework of the servicemodels of intermediate care fixed by the British Geriatrics Association the ICA we refer toin the following is structured as a community hospital or a nurse-led unit The ICA is mostly

2109


created through the conversion of acute beds and is designed to institutionalise frail olderpatients who can be discharged but cannot yet stay at home or in another facility until theyare not clinically stabilized to be discharged (Paton et al 2004) The ICA is actually aimed atimproving the integration of care between acute hospitals and post-acute care providers(such as nursing facilities inpatient rehabilitation hospitals long-term care hospicesresidential units home care agencies etc) bridging on two areas especially for frail elderlyandor chronic patients

Evidence for the effectiveness of intermediate care and community hospitals is relativelyscarce and evidence for many services that fall under the broad rubric of intermediate careis lacking (Pitchforth et al 2017 Steiner 2001) In one study (Swanson and Hagen 2016) theauthors found evidence of reduced service utilisation such as readmissions or communityservices use among those treated in a community hospital compared with those treated in ageneral acute hospital The authors demonstrated a correlation between the introduction ofthese beds and a small but significant reduction in acute care admissions highlightingintermediate care bedsrsquo potential to alleviate the burden on acute care hospitals In anotherstudy (Dahl et al 2015) a retrospective comparative cohort showed a reduction of thelength of hospital stays following the introduction of intermediate care beds for elderly andchronically ill patients

The second organisational unit selected is the CDG unit aimed at following andfacilitating the discharge process frail elderly in the final stage of acute hospitalisationFrom a theoretical point of view this unit belongs to the complex of actors and actions thatthe debate refers to with the wide term ldquotransitional carerdquo The American Geriatric Societydefines transitional care as ldquoa set of actions designed to ensure the coordination andcontinuity of healthcare as patients transfer between different locations or different levels ofcare within the same locationrdquo (Coleman and Boult 2003) For frail patients who cannot betransferred home for any reason discharge from an acute care hospital can be very complexand difficult thus resulting in inappropriate hospital stays and increasing the phenomenonof bed blockers (Benson et al 2006 Manzano-Santaella 2010) or delayed discharges(Bryan et al 2006) The issue needs to be addressed in terms of flows management as amajor cause of bottlenecks and criticalities in the system (Proudlove et al 2007) Theincreasing presence of frail elderly patients that are usually difficult to discharge because ofa lack of family support social care or the unavailability of post-acute facilities are in factamong the main causes of distress and delay for both patients and hospital staff

We propose that the discharge process should be led by a multidisciplinary team that isactivated at the beginning of the care process in acute care hospitals and is coordinated by aprofessional role that is in charge of the patient The team should conduct a comprehensivegeriatric assessment of discharge and then indicate the most suitable health facility for thepatient support the process of identification select the patientrsquos target structure as well astransmit all information that allows for the continuity of care and the pursuit of all activitiesthat favour the patientrsquos transfer This unit is required to develop strong relationships withall the systemrsquos players downstream and upstream (such as the GP) and to provide thepatient and caregiving relatives with all the support they need in order to take consciousdecisions It should also act as a facilitator for the transfer of patients that need to be takenover by the new structure It should therefore handle not only the patientrsquos transfer butalso the transfer of all relevant information respecting the patientrsquos privacy This unit andits introduction into the discharge process proved to be effective in terms of patient processand hospital outcomes (Mileski et al 2017 Carr 2007 Venkatasalu et al 2015)

4 A standard flowchart to describe clinical pathways across the hospitalThe conceptual model developed herein focusses mainly on a clinical governance approachin specific on clinical pathways that ldquodescribe the spatial and temporal sequences of

2110

MD5610

activities to be performed based on the scientific and technical knowledge and theorganisational professional and technological available resourcesrdquo (De Blaser et al 2006)

The methodrsquos approach starts by a simplified representation of standard clinicalpathways that is able to mimic the flows of all patients both emergent and elective insideacute care hospitals In the first flowchart developed in Figure 2 only the organisationalaspects common to all hospitals all countries and all disease conditions are represented Ina second step the standard pathway representation is enriched with the specificorganisational tools for frail patients analysed in Section 3 and a set of performanceindicators aimed at evaluating the impact and effectiveness of the organisational changes

To represent the standard clinical pathways we use a flowchart map where rectanglesrepresent macro activities (ie groups of services delivered such as stay interventionsdiagnoses etc) the rhombus are decision nodes and the queues generated when a resourceblockage occurs in the patient flow are represented as triangles The flowchart is shownin Figure 2

Patients can enter the hospital system as elective or emergent and they move across asequence of activities that constitute the care process inside the hospital until they exit

ELECTIVEEMERGENT

MEDICAL AREA

WARD INPATIENTSTAY BEDS

OPERATINGTHEATRE

NOAT HOME

YES TO OTHER FACILITIES

YESHOSPITAL

ADMISSION

NOAT HOME

GPsMEDICAL

PRACTICE

SURGICALINTERVENTION

YES

NO

EMERGENCYDEPARTMENT

HEALTHAND SOCIALFACILITIES

SURGICAL AREA


READYTO BE

DISCHARGED

NEEDFURTHER

ASSISTANCE

Figure 2Flowchart

representation ofstandard clinical

pathways across thehospital

2111


returning to their home or to other health and social facilities such as nursing homesor rehabilitation centres Elective patients enter the system after an outpatient visit(not present in the flowchart) when a clinician evaluates the patient defines the diagnosisand the possible surgical intervention required Depending on the diagnosis patients areincluded in the elective waiting list of a given specialty before being admitted to hospitalTwo different waiting lists (queues) and stay areas are modelled ie the medical and thesurgical area of treatment

Elective admissions are constrained by the availability of free beds The number of freebeds available on each day is determined by considering the patients who already occupiedinpatient beds assigned to the specialty as well as the expected number of patient dischargesalso considering uncertain emergency patient arrivals In the surgical area if the patient needsan intervention heshe is admitted while also considering the availability of operating roomsrsquoslot times Once admitted the patient is included in the elective surgical waiting list

Emergency patients are directly admitted from the ED if a free bed in the medical orsurgical area is available More particularly after the clinical evaluation by clinicians in EDa decision to admit can be generated The decision of patient admission includes theassigned inpatient ward where the patient must hospitalised If no beds are available thepatient must stay in the ED and wait for a free bed

Once admitted in the assigned inpatient ward both elective and emergent patientsoccupy the bed for a given amount of time (length of stay) before being considered ldquoready tobe dischargedrdquo

If further assistance is needed or the patient cannot go back home for any reason(eg lack of caregivers at home) then heshe must wait until a bed becomes available in oneof the health or social facilities dedicated to taking care of the patientrsquos pathology after theacute care in hospital such as nursing homes rehabilitation centres hospices long-termcare centres etc

The great challenge in hospital management is to provide to patients an appropriateclinical pathway reducing the presence of resource blockage (represented in Figure 2 astriangles) Concerns about blockages have increased in recent years and this paper focusseson these problems as they affect elder patients The main source of these problems is theorganisation of hospital management but also structural problems can be related to thewhole health delivery systems What is crucial is however to face the problem in a holisticmanner mapping the care process as in Figure 2 to ensure coordination among the differentsolutions tools

Some resource blockages seem to be ascribed to bed shortage This is the case of theboarding problem given by the increase of patients arriving from the ED with respect to theelective patients In Shi et al (2016) are reported the average waiting times for patients in EDwaiting to be admitted for a set of specialties (surgery cardiology general medicineorthopaedics gastroenterology oncology neurology kidney unit respiratory) of a majorpublic hospitals in Singapore The authors show that the average waiting time is 282 (with aSD of 001) hours and the percentage of patients that have to wait for more than 6 h variesbetween 479 (with a SD of 047) for general medicine unit to 116 (with a SD of 131) for kidneyunit One possible solution consists in a flexible organisation of the hospital resources thatconsiders seasonal peaks of service demand An increase of the overall number of hospitalbeds will not solve the problem as it will lead to an excess of supply in the periods wherepeaks are absent with indicators such as bed occupation ratio too small for the ward Anothersolution consists in the improvement of bed capacity planning and changing the rules used bythe bed manager to allocate patients into inpatient wards (Landa et al 2018)

Considering the second blockage (waiting lists) shortages are present only for electivepatients waiting for a surgical intervention as reported in the literature (Siciliani et al 2014)Siciliani et al (2014) reported the measuring and comparing of waiting time for 12 OECD

2112

MD5610

countries for a set of the most common elective procedure hip replacement kneereplacement cataract hysterectomy prostatectomy cholecystectomy hernia coronaryartery bypass graft percutaneous transluminal coronary angioplasty In spite ofimprovement of waiting times in recent years the trend has reversed and the meanwaiting times are increasing Even if there is a high variability hip replacement and kneereplacement have a high mean value for waiting time with a minimum of 39 days forDenmark to 495 days for Slovenia Cataract has a minimum of 46 days in Canada and111 and 113 days in Finland and Ireland respectively This shortage is also linked to theback-door entry for elective patients that try the emergency patient path (Lane et al 2000)In this case the solution is related to hospital organisation The solution is not representedby an increase of hospital beds but should consider the admission of patients with therelative clinical priority with the constraint of the maximum waiting time (Curtis et al 2010Sanmartin and Steering Committee of the Western Canada Waiting List Project (2003)Noseworthy et al 2003)

The increase of hospital bed is not generally useful as the resource that creates theblockage is the operating room with respect to the beds or the poor allocation of beds amongspecialties The problem is still an issue depending on the hospital management as itconsists to ensure the optimum mix of OR availability with respect to bedsrsquo availability(Ozcan et al 2017) or the allocation of beds following the intensity of care model for wardorganisation rather than the traditional based on surgical specialty (Landa et al 2013)

Finally the third blockage that causes delays in discharge process seems out of thehospital responsibility due mainly to shortage of home care nursing home services orshortage of occupational therapists and other service staff outside the hospital In ouropinion this is only partially true because the key driver is the insufficient capacity in thehealth and social systems to effectively work together ensuring coordination Incentivestowards better coordination have been proposed for instance in Baumann et al (2007) butthe problem still exists as reported in another study (Landeiro et al 2017) where delayeddischarges of elder patients in different countries vary from 16 to 913 per cent (average of229 per cent) with a large negative impact on costs and health outcomes

5 A conceptual model for frail patientsrsquo clinical pathwaysThe specific aim of this paper is to enrich the standard clinical pathway represented abovewith new organisational units and activities (developed by new dedicated professional roles)aimed at optimising the path of frail patients inside acute care hospitals

From a managerial point of view this means that we introduce

bull a frailty assessment for patients that are admitted in hospital (Section 31)

bull new professional roles ie case manager hospitalist and bed manager in charge offrail elderly patients from admission to discharge (Sections 32 33 and 34) and

bull two new organisational units ie ICA and CDG that are assumed to improve the flowsof frail elderly patients towards discharge and new facilities (Sections 35 and 36)

In the conceptual model we assume that for each emergent and elective patient entering thesystem an evaluation process is performed by a commission of clinicians a CFA to verifywhether there is any frailty condition

Once frail elderly patients are admitted to the wards (medical or surgical) to receive acutecare they follow the same clinical pathway of other patients with the exception that theycontinue to be followed by the hospitalist and the case manager who coordinate thepatientrsquos interventions with the ward staff If the patient is frail then heshe falls under theresponsibility of a hospitalist and a case manager that are responsible for specific aspects ofthe care process The hospitalist supports the patientrsquos clinical pathway with respect to all

2113


needs in terms of healthcare and frail conditions and will supervise any phase ofthe process intervening if and when necessary The case manager will be in charge of theday-by-day management of the patient

The flowchart representation is customised to frail patientsrsquo needs when the patient isready to be discharged from acute wards It considers different hypotheses the first one isthat patients can be discharged to their home only if they have appropriate family orcaregiversrsquo support In this case the patient goes back home and the entire pathwaydocumentation such as exams tests visits and the results is sent to the patientrsquos GP ormedical practice The second hypothesis is that patients cannot be discharged if they needfurther assistance eg patientsrsquo psychophysical conditions have not yet stabilised and theyare expected to continue to be temporarily instable In this case patients can be admitted tothe ICA where they can receive less intensive and multidisciplinary care for a limited periodof time

Since the number of patients requiring access to the ICA may vary in order to geteconomies of scale the intermediate care area can also be opened to non-frail patients In anycase frail patients should take priority and the frailty code alerts the ICA staff at anymoment about the number of frail in-patients that need to be admitted once they aredeclared dischargeable by the acute area Indeed the ICA is introduced primarily to reduceor at least shorten bed blockersrsquo inappropriate hospital stays in acute wards

The last hypothesis is that other patients once dischargeable from the acute ward(or even from ICA) need further long-term assistance and must be institutionalised in othersocial or health facilities ie nursing facilities inpatients rehabilitation hospitals long-termcare hospices or residential units It can take a long time for the ward staff (or even for theICA staff ) to find the most appropriate facility for the specific patientrsquos needs so theflowchart is enriched with a CDG The CDG is a unit in charge of contacting the differentfacilities outside the hospital in order to safely and quickly transfer the patient and allinformation about their clinical pathway to the institution that can continue the process ofcare outside the hospital CDGrsquos main goal is to facilitate the flow of frail elderly patients inorder to avoid delayed discharges and bottlenecks due to a lack of communication amongthe different actors involved in the care processes For this reason just like ICA CDG isintroduced to face critical issues linked to frail elderly patients Indeed in order to obtaineconomies of scale CDG can also support the transfer of any patient who cannot bedischarged to their home but is in need of admission in another facility after hisherdischarge for any reason

The introduction of these elements in hospitals requires a re-engineering of someprocesses with new resources and new competences of a part of hospital staff Hospitalareas already available or obtained from space optimisation of different wards can be usedfor ICA while CDG services can be performed by an office with administrative staff thatcontact the facilities and organise the logistic aspects of patient discharge Case managerand bed manager are professional tasks that can be assigned to specialised nurse whilehospitalist has to be a physician of general medicine with both organisational and clinicalcompetences FCA requires staff already present in inpatient wards

A full representation of the tools and the professional roles integrated into the hospitalorganisation is represented in Figure 3

51 A set of quality indicators for an evidence-based model for frail patientsIn order to validate the model a set of indicators was defined to monitor the flow of patientsand evaluate the impact of the modelrsquos application on the delivery of care to frail patients inacute care hospitals Naturally this set of indicators needs to be supported by a hospitalinformation system (HIS) that is able to collect data and information concerning frailpatients In case there is no unanimously accepted medical definition of frailty or missing

2114

MD5610

updates for frail elderly conditions in the HIS the information system should focus on thepopulation aged 65 years and over in order to collect relevant data

In order to build the set of indicators we refer to Donabedianrsquos (1966 1988 2005)healthcare quality model which was introduced in the 1960s and named after the physicianand researcher who developed it This model became a milestone for quality improvementprocesses and for models of evidence-based practice in healthcare (Anderson Elverson andSamra 2012 Titler et al 2011) Donabedianrsquos model is based on the measurement of threedimensions ndash structures processes and clinical outcomes ndash that are assumed to be strictlyrelated Improvements in the structure of care should lead to improvements in clinicalprocesses which should in turn improve patient outcomes (Moore et al 2015)More specifically structure indicators are expected to measure the settings in which care isdelivered in terms of material human and organisational resources while process indicatorsassess what the provider does for the patient Finally outcome measures try to describe theeffects of care or of a change in care processes on the health status of patients (Mainz 2003)

In order to validate the model and gather some evidence about its ability to overcome themost critical issues (eg providing frail patients with appropriate and affordable care) the

EMERGENT

EMERGENCYDEPARTMENT

HOSPITALADMISSION

NOAT HOME

YESMEDICAL AREA

BEDMANAGER

SURGICAL AREA



NO

YES

OPERATINGTHEATRE

READYTO BE

DISCHARGED

NEEDFURTHER

ASSISTANCE

NOAT HOME

INTERMEDIATE CAREAREA(ICA)

CASE MANAGER ANDHOSPITALIST

GPsMEDICAL

PRACTICE


CENTRALDISCHARGE

GATEWAY (CDG)

YES TO ICA YES TO OTHER STRUCTURES


FCA and FRAILTY CARD

FRAILPATIENT

HOSPITALIST AND CASEMANAGER ASSIGNMENT

ELECTIVE

Figure 3Flowchart of

conceptual model forfrail patients

2115


set of (structure process and outcome) indicators is expected to measure if and how themodel is able to achieve the objectives it pursues ie to reduce frail patientsrsquo admission timeand length of stay to better coordinate multidisciplinary interventions to speed updischarging and if necessary admission to other long-term facilities and eventually toreduce the risk of adverse events

For each of these objectives some structure process and outcome indicators have beenchosen based on research and practice evidence about the delivery of care to frail patients inacute hospitals In Table I a general overview of the indicators is provided

511 Reducing frail patientsrsquo admission admission time and length of stay In order toassess the degree to which this objective is achieved the model proposes the use of someindicators The indicator ldquoProportion of frail elderly patients being admitted to wardsbeyond the assessmentrdquo (National Audit Office Department of Health UK 2016) is proposedin order to evaluate whether the model contributes to better managing admissionspreventing inappropriate ones Other relevant indicators are ldquobed occupancy for frail elderlypatients and average length of stay for frail elderly patientsrdquo which are expected to decreasewith the application of the model Also the ldquoreadmission rate of frail elderly patientsrdquo forthese patients appears to be an appropriate indicator since timely and appropriate care isexpected to promote a decrease in readmission after 30 days (Silvester et al 2014) Finallythe ldquofrail elderly patientshospitalist ratio and frail elderly patients case manager ratiordquo aretwo structure indicators for measuring the efficiency and effectiveness of the two humanresources we introduced in the model

512 Better coordinating multidisciplinary interventions Coordination is at the verybasis of the model The patient-centred approach improves coordination inside the hospital

Objective Indicator

Type structure(S) process (P)outcome (O)

Reducing frail elderly patientsrsquoadmission time and length of stay

Proportion of frail elderly patients being admitted towards beyond the assessment process

P

Frail elderly patients ndash hospitalist ratio SFrail elderly patients ndash case manager ratio SBed occupancy of frail elderly patients PAverage length of stay of frail elderly patients PReadmission rate of frail elderly patients O

Better coordinating multidisciplinaryinterventions

Average number of frail elderly patients waiting foradmission to ICA

P

Average waiting time of frail patients waiting foradmission to ICA

P

Prevalence and types of medication discrepancies OSpeeding discharges and if necessaryadmission to other long-term facilities

Average length of delayed discharges ( from the daythe patient is declared dischargeable to the day of thedischarge)

P

No of delayed discharges attributable to frail elderlypatients

P

Average length of a delayed transfer of careattributable to frail elderly patients

P

No of delayed transfers of care attributable to frailelderly patients

P

Reducing the risk of adverse events Hospital-acquired infections (HAI) of frail elderlypatients

O

In-hospital mortality of frail elderly patients ONo of geriatric syndromes O

Table ISet of qualityindicators for anevidence-based modelfor frail patients

2116

MD5610

among its units and among hospital and other actors of the healthcare system The ldquonumberof frail elderly patients waiting for admission to ICArdquo and ldquoaverage waiting time of frailpatients waiting for admission to ICArdquo are two process indicators that are meant to evaluatethe ability of the model to speed frail patientsrsquo admission to this unit the ldquoprevalence andtype of medication discrepanciesrdquo on the contrary concern coordination problems amonghospital and other actors during for example patientsrsquo transitions from community to acutecare hospitals (Villanyi et al 2011) Coordination between long-term facilities and acutehospitals is expected to improve information flows and decrease medication discrepancies

513 Speeding up discharging and if necessary admission to other long-term facilitiesWith reference to speeding up the discharging of patients that are ready to be dischargedthe most appropriate indicators appear to be the ldquonumber of delayed discharges attributableto frail elderly patientsrdquo and the ldquoaverage length of delayed discharges attributable to frailelderly patientsrdquo (National Audit Office Department of Health 2016) Similarly if admissionto other facilities is necessary the indicators to use are the ldquoaverage length of a delayedtransfer of care attributable to frail elderly patientsrdquo and the ldquonumber of delayed transfers ofcare attributable to frail elderly patientsrdquo (National Healthcare System BenchmarkingNetwork 2017)

514 Reducing the risk of adverse events Concerning the impact on the health status offrail older patients which needs more time to be evaluated the ldquoin-hospital mortality of frailelderly patientsrdquo appears to be a fundamental indicator (Silvester et al 2014) Moreoverconsidering the vulnerability of frail patients it is important to reduce high-risk eventsFor this reason the ldquonumber of hospital-acquired infections (HAI) of frail elderly patientsrdquo isconsidered with specific reference to the infections most often observed in frail patientssuch as pneumonia urinary tract and skin infections ( Jones 1990) Also the ldquonumber ofgeriatric syndromesrdquo such as delirium falls incontinence poor nutrition immobilityfunctional decline and pressure sores (George et al 2013) is considered

6 ConclusionFuture demographic trends lead us to expect a modification of the composition of peopledemanding to be admitted to acute care hospitals Nowadays more than half of patients inEuropean countries are elderly and they are increasing rapidly This causes more frailpeople to address health services because frailty depends on a set of conditions all linked toage such as comorbidity disability and geriatric disorders Over time specific healthservices for frail elderly have been developed in all countries building a network in order tofollow them continuously across different care settings For a successful integrated carepathway a central role is still played by the acute care hospital where frail patients spendsome time coming from and returning home or to less intensive care levels (nursing homespost-acute facilities social care units caregivers etc)

Compared to the growing demand for hospital services the corresponding supplyappears to be inadequate It is not a matter of resources but rather a matter of theorganisational structure of the hospital Following the evolution of medical science thisstructure has evolved according to a more and more specialist approach aimed at caring forthe single diseases of a specific organ

Frail older people on the other hand require a holistic approach that takes intoaccount all dimensions as a whole Hospitals are generally not equipped to treat complexpatients properly

This organisational gap results in unnecessary waits and increasing patient length ofstay More time spent in hospital wards means poorer outcomes because in addition to theusual iatrogenic risk for an elderly person a hospital stay means leaving hisherenvironment involving functional decline and a deterioration of their mental conditions

2117


The problem is not new and tools have been developed for years to try and avoid thesenegative consequences such as a comprehensive assessment of geriatric conditions a casemanager a low intensity ward and so on

The novelty of the paper is to propose that all positive previous experiences areincluded in the care process by developing a conceptual model designing the carepath for frail patients inside an acute care hospital The conceptual model wasdeveloped looking for the main available evidence-based instruments that have alreadybeen found to facilitate a frail elderly path The conceptual model is therefore in a certainsense already EBMgt because the standard clinical pathway of the hospital hasbeen enriched with new organisational units and activities (developed by newdedicated professional roles) aimed at optimising the path of frail patients inside acutecare hospitals

But even if different tools have been proved to be effective during years of localexperience in single countries or hospitals we maintain that further research on theevidence is necessary applied to the entire process The developed conceptual model can beconsidered a framework for finding further proof of the entire process and not only of thesingle tools as was done until now

However the overall study presents both strengths and weaknesses The strength of thisstudy lies in its contribution consisting of providing a new organisational path for frailelderly that considers a holistic view with respect to the actual literature Each elementincluded in the model derives from an efficient innovation in hospital management andorganisation but each study analysed it separately The hospital is composed of a synergyof different elements and units that interact and are integrated to provide healthcare topatients in need Focussing on and analysing only a singular problem or area within theorganisation is the wrong approach

The weakness of the framework proposed herein consists of the lack of proof for theconceptual modelrsquos effectiveness Each element of the model has proved effectiveness interms of outcome and output when implemented inside a hospital system but wecannot prove the effectiveness of joining all the elements inside a unique framework as weproposed In order to verify the real effectiveness hard work needs to be donefirst coming to an agreement with a hospital that can help with the provision ofdetailed data and second through the development of a simulation model that canrepresent the system Once the system is represented and validated a what-if and scenarioanalysis can be performed in order to verify the impact of the conceptual model and thedifferent strategies in terms of resource (quantity) and organisation Another limitation isrepresented by the adoption of only three principles of evidence-based practice as we didnot consider the stakeholder point of view directly especially patients In the developmentof this point it is necessary to provide a qualitative study based on patient andpublic involvement interviews to analyse the preferences of both National andRegional Healthcare System directors and frail patients As this element is reallyimportant this will be a supplementary study that will be developed in the future tosupport the framework

Some studies have already been proposed by some of the authors and they attempt tomodel and verify the impact of bed management in hospital organisations by using differentsimulation techniques such as discrete event simulation system dynamics and hybridsimulation approaches Future directions of research will be focused on introducing anddeveloping a hybrid simulation model able to represent the care process and verify theimpact of the organisational changes in the current practice The simulation model willrepresent reality providing a scenario analysis to evaluate the impact of the conceptualmodel on the hospitalrsquos organisation under several resource constraints and considering thevariations of service demand and supply

2118

MD5610

References

Anderson Elverson C and Samra HA (2012) ldquoOverview of structure process and outcome indicatorsof quality in neonatal carerdquo Newborn and Infant Nursing Reviews Vol 12 No 3 pp 154-161

Bagust A Place M and Posnett J (1999) ldquoDynamics of bed use in accommodating emergencyadmissions stochastic simulation modelrdquo The British Medical Journal Vol 310 No 7203pp 155-158

Barends E Rousseau DM and Briner RB (2014) ldquoEvidence-based management the basicprinciplesrdquo available at wwwcebmaorgwp-contentuploadsEvidence-Based-Practice-The-Basic-Principlespdf (accessed 5 March 2018)

Baumann M Evans S Perkins M Curtis L Netten A Fernandez JL and Huxley P (2007)ldquoOrganisation and features of hospital intermediate care and social services in English siteswith low rates of delayed dischargerdquo Health and Social Care in the Community Vol 15 No 4pp 295-305

Benson RT Drew JC and Galland RB (2006) ldquoA waiting list to go home an analysis of delayeddischarges from surgical bedsrdquo Annals of The Royal College of Surgeons of England Vol 88No 7 pp 650-652

Black HG and Gallan AS (2015) ldquoTransformative service networks cocreated value as well-beingrdquoThe Service Industries Journal Vol 35 Nos 1516 pp 826-845

Brennan TA Hebert LE Laird NM Lawthers A Thorpe KE Leape LL Localio ARLipsitz SR Newhouse JP Weiler PC and Hiatt HH (1991) ldquoHospital characteristicsassociated with adverse events and substandard carerdquo The Journal of the American MedicalAssociation Vol 265 No 24 pp 3265-3269

Bryan K Gage H and Gilbert K (2006) ldquoDelayed transfers of older people from hospital causes andpolicy implicationsrdquo Health Policy Vol 76 No 2 pp 194-201

Cammarata JF (2005) ldquoThe hospitalist creating a patient-focused paradigm for a changerdquo Journal ofthe American Medical Directors Association Vol 6 No 2 pp 162-164

Carr DD (2007) ldquoCase managers optimize patient safety by facilitating effective care transitionsrdquoProfessional Case Management Vol 12 No 2 pp 70-80

Chewning B and Sleath B (1996) ldquoMedication decision-making and management a client-centredmodelrdquo Social Science and Medicine Vol 42 No 3 pp 389-398

Clegg A Young J Iliffe S Rikkert MO and Rockwood K (2013) ldquoFrailty in elderly peoplerdquo TheLancet Vol 381 No 9868 pp 752-762

Coleman EA and Boult C (2003) ldquoImproving the quality of transitional care for persons withcomplex care needs position statement of the American Geriatrics Society Health Care SystemsCommitteerdquo Journal of American Geriatric Society Vol 51 No 4 pp 556-557

Curry N and Ham C (2010) Clinical and Service Integration The Route to Improved OutcomesThe Kingrsquos Fund London

Curtis AJ Russell COH Stoelwinder JU and McNeil JJ (2010) ldquoWaiting lists and elective surgeryordering the queuerdquo The Medical Journal of Australia Vol 192 No 4 pp 217-220

Dahl U Johnsen R Saeligtre R and Steinsbekk A (2015) ldquoThe influence of an intermediate carehospital on health care utilization among elderly patients ndash a retrospective comparative cohortstudyrdquo BMC Health Services Research Vol 15 No 48 pp 1-12

De Blaser L Depreitere R De Waele K Vanhaecht K Vlayen J and Sermeus W (2006) ldquoDefiningpathwaysrdquo Journal of Nursing Management Vol 14 No 7 pp 553-563

Donabedian A (1966) ldquoEvaluating the quality of medical carerdquo The Milbank Memorial FundQuarterly Vol 44 No 3 pp 166-506

Donabedian A (1988) ldquoThe quality of care How can it be assessedrdquo The Journal of the AmericanMedical Association Vol 260 No 12 pp 1743-1748

Donabedian A (2005) ldquoEvaluating the quality of medical carerdquo The Milbank Quarterly Vol 83 No 4pp 691-729

2119


Ellis G Whitehead MA Robinson D OrsquoNeill D and Langhorne P (2011) ldquoComprehensive geriatricassessment for older adults admitted to hospital meta-analysis of randomised controlled trialsrdquoThe British Medical Journal Vol 343 No d6553 pp 1-10 available at wwwbmjcomcontentbmj343bmjd6553fullpdf

Ellis G Gardner M Tsiachristas A Langhorne P Burke O Harwood RH Conroy SP Kircher TSomme D Saltvedt I Wald H OrsquoNeill D Robinson D and Shepperd S (2017) ldquoComprehensivegeriatric assessment for older adults admitted to hospitalrdquo Cochrane Database of SystematicReviews No 9 doi 10100214651858CD006211pub3 available at wwwcochranelibrarycomcdsrdoi10100214651858CD006211pub3epdffull

Eurostat (2016) ldquoHospital discharges and length of stay statisticrdquo available at httpeceuropaeueurostatstatistics-explainedindexphpHospital_discharges_and_length_of_stay_statistics(accessed 3 March 2018)

Fox MT Sidani S Persaud M Tregunno D Maimets I Brooks D and OrsquoBrien K (2013) ldquoAcutecare for elders components of acute geriatric unit care systematic descriptive reviewrdquo Journal ofthe American Geriatrics Society Vol 61 No 6 pp 939-946

Fried LP Tangen CM Walston J Newman AB Hirsch C Gottdiener J Seeman T Tracy RKop W J Burke G and McBurnie MA Cardiovascular Health Study Collaborative ResearchGroup (2001) ldquoFrailty in older adults evidence for a phenotyperdquo The Journals of GerontologySeries A Biological Sciences and Medical Sciences Vol 56 No 3 pp 146-156

George J Long S and Vincent C (2013) ldquoHow can we keep patients with dementia safe in our acutehospitals A review of challenges and solutionsrdquo The Journal of the Royal Society of MedicineVol 106 No 9 pp 355-361

Goldmann DR (1999) ldquoThe hospitalist movement in the United States what does it mean forinternistsrdquo Annals of Internal Medicine Vol 130 No 4 pp 326-327

Green J and Armstrong D (1994) ldquoThe views of service providersrdquo in Morrell D Green JArmstrong D Bartholomew J Gelder F Jenkins C Jankowski R Mandalia S Britten NShaw A and Savill R (Eds) Five Essays on Emergency Pathways Institute for the Kings FundCommission on the Future of Acute Services in London Kingrsquos Fund London pp 11-31

Haggerty JL Reid RJ Freeman GK Starfield BH Adair CE and McKendry R (2003) ldquoContinuity ofcare a multidisciplinary reviewrdquo The British Medical Journal Vol 22 No 327 (7425) pp 1219-1221

Haraden C and Resar R (2004) ldquoPatient flow in hospitals understanding and controlling it betterrdquoFrontiers of Health Services Management Vol 20 No 4 pp 3-15

Hock Lee KYY Song Yang K Chi Ong B and Seong Ng H (2011) ldquoBringing generalists into thehospital outcomes of a family medicine hospitalist model in Singaporerdquo Journal of HospitalMedicine Vol 6 No 3 pp 115-121

Howell E Bessman E Marshall R and Wright S (2010) ldquoHospitalist bed management effectingthroughput from the emergency department to the intensive care unitrdquo Journal of Critical CareVol 7 No 2 pp 184-189

Howell E Bessman E Kravet S Kolodner K Marshall R and Wright S (2008) ldquoActive bedmanagement by hospitalists and emergent department throughputrdquo Annals of InternalMedicine Vol 149 No 11 pp 804-810

Jones SR (1990) ldquoInfections in frail and vulnerable elderly patientsrdquo The American Journal ofMedicine Vol 88 No 3C pp 30S-33S

Katz TF (1963) ldquoADL activities of daily livingrdquo The Journal of the American Medical AssociationVol 185 pp 914-919

Kuo YF and Goodwin JS (2011) ldquoAssociation of hospitalist care with medical utilization after dischargeevidence of cost shift from a cohort studyrdquo Annals of Internal Medicine Vol 155 No 3 pp 152-159

Landa P Tagravenfani E and Testi A (2013) ldquoSimulation and optimization for bed re-organization at asurgery departmentrdquo in Kacprzyk J Leifsson L Obaidat M Koziel S and Oumlren T (Eds)Proceedings of the 3rd International Conference on Simulation and Modeling MethodologiesTechnologies and Applications (SIMULTECH) SciTEPress (Science and TechnologyPublications Lda) Reykjaviacutek pp 584-594

2120

MD5610

Landa P Sonnessa M Tagravenfani E and Testi A (2018) ldquoMultiobjective bed management consideringemergency and elective patient flowsrdquo International Transactions in Operational ResearchVol 25 No 1 pp 91-110

Landeiro F Roberts K Mcintosh Gray A and Leal J (2017) ldquoDelayed hospital discharges of olderpatients a systematic review on prevalence and costsrdquo Gerontologist gnx028 pp 1-12 availableat httpsacademicoupcomgerontologistadvance-articledoi101093gerontgnx0283850587

Lane DC Monefeldt C and Rosenhead JV (2000) ldquoLooking in the wrong place for healthcareimprovements a system dynamics study of an accident and emergency departmentrdquo Journal ofthe Operational Research Society Vol 51 No 5 pp 518-531

Lawton M and Brody E (1969) ldquoAssessment of older people self-maintaining and instrumentalactivities of daily livingrdquo Gerontologist Vol 9 No 3 pp 179-186

Leape LL Brennan TA Laird N Lawthers AG Localio R Barnes BA Hebert LNewhouse JP Weiler PC and Hiatt H (1991) ldquoThe nature of adverse events in hospitalizedpatients Results of the Harvard Medical Practice Study IIrdquo The New England Journal ofMedicine Vol 324 No 6 pp 377-384

Lemieux-Charles L and Champagne F (2004) Using Knowledge and Evidence in HealthcareMultidisciplinary Perspectives University of Toronto Press Toronto

Leung AC Liu C and Chi NW (2004) ldquoCost-benefit analysis of a case management project for thecommunitydwelling frail elderly in Hong Kongrdquo Journal of Applied Gerontology Vol 23 No 1pp 70-85

Lincolnshire Community Health Services (2015) ldquoFrailty pathway ndash a patient-centred approachguidance for cliniciansrdquo available at wwweolccoukuploadsFrailty-Pathway-prompt-cardspdf (accessed 3 March 2018)

Madeira S Melo M Porto J Monteiro S Pereira de Moura JM Alexandrino MB and Moura JJ(2007) ldquoThe diseases we cause iatrogenic illness in a department of internal medicinerdquoEuropean Journal of Internal Medicine Vol 18 No 5 pp 391-399

Mainz J (2003) ldquoDefining and classifying clinical indicators for quality improvementrdquo InternationalJournal for Quality in Health Care Vol 15 No 6 pp 523-530

Manzano-Santaella A (2010) ldquoFrom bed-blocking to delayed discharges precursors andinterpretations of a contested conceptrdquo Health Services Management Research Vol 23 No 3pp 121-127

McColl-Kennedy JR Vargo SL Dagger TS Sweeney JC and van Kasteren Y (2012) ldquoHealthcare customer value co-creation practice stylesrdquo Journal of Service Research Vol 15 No 4pp 370-389

Mead N and Bower P (2000) ldquoPatient-centredness a conceptual framework and review of theempirical literaturerdquo Social Science and Medicine Vol 51 No 7 pp 1087-1110

Melis RJF Olde Rikkert MGM Parker SG and van Eijken MIJ (2004) ldquoWhat is intermediatecare An international consensus on what constitutes intermediate care is neededrdquo The BritishMedical Journal Vol 14 No 329(7462) pp 360-361

Mileski M Topinka JB Lee K Brooks M McNeil C and Jackson J (2017) ldquoAn investigation ofquality improvement initiatives in decreasing the rate of avoidable 30-day skilled nursingfacility-to-hospital readmissions a systematic reviewrdquo Clinical Intervention in Aging Vol 12pp 213-222

Moore L Lavoie A Bourgeois G and Lapointe J (2015) ldquoDonabedianrsquos structure-process-outcomequality of care model validation in an integrated trauma systemrdquo The Journal of Trauma andAcute Care Surgery Vol 78 No 6 pp 1168-1175

National Audit Office Department of Health (2016) ldquoDischarging older patients from hospitalrdquo availableat wwwnaoorgukreportdischarging-older-patients-from-hospital (accessed 3 March 2018)

National Healthcare System Benchmarking Network (2017) ldquoDelayed transfers of carerdquoavailable at wwwnhsbenchmarkingnhsukprojects2017410delayed-transfers-of-care(accessed 5 March 2018)

2121


Naylor M and Keating SA (2008) ldquoTransitional care moving patients from one care setting toanotherrdquo The American Journal of Nursing Vol 108 No 9 pp 58-63

Naylor MD Brooten DA Campbell RL Maislin G McCauley KM and Schwartz JS (2004)ldquoTransitional care of older adults hospitalized with heart failure a randomized controlled trialrdquoJournal of the American Geriatrics Society Vol 52 No 5 pp 675-684

Noseworthy TW McGurran JJ and Hadorn DC (2003) ldquoSteering Committee Of TheWestern Canada Waiting List Project waiting for scheduled services in Canada developmentof priority setting scoring systemsrdquo Journal of Evaluation in Clinical Practice Vol 9 No 1pp 23-31

Ozcan YA Tagravenfani E and Testi A (2017) ldquoImproving the performance of surgery-based clinicalpathways a simulationndashoptimization approachrdquoHealth Care Management Science Vol 20 No 1pp 1-15

Paton JM Fahy MA and Livingston GA (2004) ldquoDelayed discharge ndash a solvable problem Theplace of intermediate care in mental health care of older peoplerdquo Aging amp Mental Health Vol 8No 1 pp 34-39

Philp I Mills KA Thanvi B Ghosh K and Long JF (2013) ldquoReducing hospital bed use by frailolder people results from a systematic review of the literaturerdquo International Journal ofIntegrated Care Vol 13 e048 pp 1-19 available at wwwijicorgarticles105334ijic1148

Pitchforth E Nolte E Corbett J Miani C Winpenny E van Teijlingen E et al (2017) ldquoCommunityhospitals and their services in the NHS identifying transferable learning from internationaldevelopments ndash scoping review systematic review country reports and case studiesrdquo HealthServices and Delivery Research Vol 5 No 19 pp 1-248

Porter ME (2010) ldquoWhat is value in health carerdquo New England Journal of Medicine Vol 363 No 26pp 2477-2481

Proudlove N Boaden R and Jorgensen J (2007) ldquoDeveloping bed managers the why and the howrdquoJournal of Nursing Management Vol 15 No 1 pp 34-42

Rachoin JS Skaf J Cerceo E Fitzpatrick E Milcarek B Kupersmith E and Scheurer DB (2012)ldquoThe impact of hospitalists on length of stay and costs systematic review and meta-analysisrdquoThe American Journal of Managed Care Vol 18 No 1 pp e23-e30

Reilly S Hughes J and Challis D (2010) ldquoCase management for long-term conditions implementationand processesrdquo Ageing and Society Vol 30 No 1 pp 125-155

Rockwood K Song X MacKnight C Bergman H Hogan DB McDowell I and Mitnitski A (2005)ldquoA global clinical measure of fitness and frailty in elderly peoplerdquo The Canadian MedicalAssociation Journal Vol 173 No 5 pp 489-495

Rodriacuteguez-Mantildeas L Feacuteart C Mann G Vintildea J Chatterji S Chodzko-Zajko W Gonzalez-ColaccediloHarmand M Bergman H Carcaillon L Nicholson C Scuteri A Sinclair A Pelaez MVan der Cammen T Beland F Bickenbach J Delamarche P Ferrucci L Fried LPGutieacuterrez-Robledo LM Rockwood K Rodriacuteguez Artalejo F Serviddio G and Vega E onbehalf of the FOD-CC group (2013) ldquoSearching for an operational definition of frailty a Delphimethod based consensus statement The frailty operative definitionndashconsensus conferenceprojectrdquo The Journals of Gerontology Series A Biological Sciences and Medical Sciences Vol 68No 1 pp 62-67

Rousseau DM (2005) ldquoEvidence-based management in health carerdquo in Korunka C and Hoffmann P(Eds) Change and Quality in Human Service Work Hampp Munich pp 33-46

Salive ME (2013) ldquoMultimorbidity in older adultsrdquo Epidemiologic Reviews Vol 35 No 1 pp 75-83

Sanmartin CA and Steering Committee of the Western Canada Waiting List Project (2003) ldquoTowardstandard definitions for waiting timesrdquo Healthcare Management Forum Vol 16 No 2 pp 49-53

Schnekenberg RP (2011) ldquoHospital medicine in South Americardquo Hospitalist-in-Training reports fromPASHA the First Congress of the Pan-American Society of Hospitalists November available atwwwacphospitalistorgarchives201102studenthtm (accessed 12 May 2018)

2122

MD5610

Scott WC Bernstein SL Coble YD Eisenbrey AB Estes EH Karlan MS Kennedy WRNumann PJ Skom JH Steinhilber RM Strong JP Wagner HN Hendee WR McGivneyWTAnderson MS Gilchrist A Mondeika T and Schwartzberg JG (1990) ldquoAmerican MedicalAssociation White Paper on elderly health Report of the Council on Scientific Affairsrdquo Archives ofInternal Medicine Vol 150 No 12 pp 2459-2472

Shi P Chou MC Dai JG Ding D and Sim J (2016) ldquoModels and insights for hospital inpatientoperations Time-dependent ED boarding timesrdquo Management Science Vol 62 No 1 pp 1-28

Shortell SM Rundall TG and Hsu J (2007) ldquoImproving patient care by linking evidence-basedmedicine and evidence-based managementrdquo The Journal of the American Medical AssociationVol 298 No 6 pp 673-676

Siciliani L Moran V and Borowitz M (2014) ldquoMeasuring and comparing health care waiting times inOECD countriesrdquo Health Policy Vol 118 No 3 pp 292-303

Silvester KM Mohammed MA Harriman P Girolami A and Downes TW (2014) ldquoTimely carefor frail older people referred to hospital improves efficiency and reduces mortality without theneed for extra resourcesrdquo Age and Ageing Vol 43 No 4 pp 472-477

Singh D and Ham C (2006) Improving Care for People with Long-Term Conditions A Review of UKand International Frameworks NHS University of Birmingham Birmingham available atwwwbirminghamacukDocumentscollege-social-sciencessocial-policyHSMCresearchlong-term-conditionspdf (accessed 5 September 2017)

Song X Mitnitski A and Rockwood K (2010) ldquoPrevalence and 10-year outcomes of frailty in olderadults in relation to deficit accumulationrdquo Journal of the American Geriatrics Society Vol 58No 4 pp 681-687

Steiner A (2001) ldquoIntermediate care ndash a good thingrdquo Age and Ageing Vol 30 No 3 pp 33-39

Stuck AE (1997) ldquoMultidimensional geriatric assessment in the acute hospital and ambulatorypracticerdquo Schweizerische Medizinische Wochenschrift Vol 127 No 43 pp 1781-1788

Stuck AE Siu AL Wicland GD Adam J and Rubenstein LZ (1993) ldquoComprehensive geriatricassessment a meta-analysis of controlled trialsrdquo The Lancet Vol 342 No 8878 pp 1032-1036

Swanson JO and Hagen TP (2016) ldquoReinventing the community hospital a retrospective population-based cohort study of a natural experiment using register datardquo The British Medical JournalOpen Vol 6 No 12 pp 1-9

Szlejf C Farfel JM Curiati JA De Barros Couto Junior E Jacob-Filho W and Azevedo RS (2012)ldquoMedical adverse events in elderly hospitalized patients a prospective studyrdquo Clinics Vol 67No 11 pp 1247-1252

Titler MG Kleiber C Steelman VJ Rakel BA Budreau G Everett LQ Buckwalter KCTripp-Reimer T and Goode CJ (2011) ldquoThe Iowa model of evidence-based practice to promotequality carerdquo Critical Care Nursing Clinics of North America Vol 13 No 4 pp 497-509

United Nations Department of Economic and Social Affairs Population Division (2017) ldquoWorldPopulation Ageing 2017 ndash STESASERA408rdquo available at wwwunorgdevelopmentdesaageingwp-contentuploadssites24201705WPA-2017-Launch-to-the-IDOP-5-October-2017pdf (accessed 3 February 2018)

Van Eeden K Moeke D and Bekker R (2016) ldquoCare on demand in nursing homes a queuing theoreticapproachrdquo Health Care Management Science Vol 19 No 3 pp 227-240

Venkatasalu MR Clarke A and Atkinson J (2015) ldquo lsquoBeing a conduitrsquo between hospital and homestakeholdersrsquo views and perceptions of a nurse-led palliative care discharge facilitator service inan acute hospital settingrdquo Journal of Clinical Nursing Vol 24 Nos 1112 pp 1676-1685

Villanyi D Fok M and Wong RY (2011) ldquoMedication reconciliation identifying medicationdiscrepancies in acutely ill hospitalized older adultsrdquo The American Journal of GeriatricPharmacotherapy Vol 9 No 5 pp 339-344

Wachter RM (2002) ldquoThe evolution of the hospitalist model in the United Statesrdquo The Medical Clinicsof North America Vol 86 No 4 pp 687-706

2123


Wachter RM and Goldman L (1996) ldquoThe emerging role of lsquohospitalistsrsquo in the American health caresystemrdquo The New England Journal of Medicine Vol 335 No 7 pp 514-517

Wachter RM and Goldman L (2002) ldquoThe hospitalist movement five years laterrdquo The Journal of theAmerican Medical Association Vol 287 No 4 pp 487-494

Wachter RM and Goldman L (2016) ldquoZero to 50000 ndash the 20th anniversary of the Hospitalistrdquo TheNew England Journal of Medicine Vol 375 No 11 pp 1009-1011

Walshe K and Rundall TG (2001) ldquoEvidence-based management from theory to practice inhealthcarerdquo The Milbank Quarterly Vol 79 No 3 pp 429-457

White P and Hall ME (2006) ldquoMapping the literature of case management nursingrdquo Journal of theMedical Library Association Vol 94 S2 pp E99-E106

Woo J Yu R Wong M Yeung F Wong M and Lum C (2015) ldquoFrailty screening in the communityusing the FRAIL Scalerdquo Journal of the American Medical Directors Association Vol 16 No 5pp 412-419

World Health Organization (2016) ldquoWHO framework on integrated people-centered health servicesrdquoavailable at wwwwhointservicedeliverysafetyareaspeople-centred-careen (accessed 16August 2017)

Yousefi V and Wilton D (2011) ldquoRe-designing hospital care learning from the experience of hospitalMedicine in canadardquo Journal of Global Health Care Systems Vol 1 No 3 pp 1-10

Corresponding authorPaolo Landa can be contacted at PLandaexeteracuk


2124

MD5610

Application of evidence-basedmanagement to chronic disease

healthcare a frameworkSaligrama Agnihothri

School of Management Binghamton University (SUNY) BinghamtonNew York USA andRaghav Agnihothri

Independent Researcher New York New York USA

AbstractPurpose ndash The purpose of this paper is to develop a framework for the application of evidence-basedmanagement to chronic disease healthcareDesignmethodologyapproach ndash Chronic healthcare is specially characterized by recursive patient-physicianinteractions in which evidence-based medicine (EBM) is applied However implementing evidence-based solutionsto improve healthcare quality requires managers to effect changes in many different areas organizationalstructure procedures technology and in physicianprovider behaviors To complicate matters further they mustachieve these changes using the tools of resource allocation or incentives The literature contains many systematicreviews evaluating the question of physician and patient behavior under various types and structures ofincentives Similarly systematic reviews have also been done regarding specific changes to the healthcare processand their effectiveness in improving patient outcomes Yet these reviews uniformly lament a lack of appropriatedata from well-organized studies on the question of ldquoWhyrdquo solutions may work in one instance while not inanother The authors present a new theoretical framework that aids in answering this questionFindings ndash This paper presents a new theoretical framework (Influence Model of Chronic Healthcare) thatidentifies the critical areas in which managers can effect changes that improve patient outcomes the influencethese areas can have on each other as well as on patient and physician behavior and the mechanisms by whichthese influences are exerted For each the authors draw upon and present the evidence in the literature Ultimatelythe authors recognize that this is a complex question that has not yet been fully researched The contribution of thismodel is twofold first the authors hope to focus future research efforts and second provide a useful heuristic tomanagers who must make decisions with only the lesser-quality evidence the literature contains todayOriginalityvalue ndash This model can be used by managers as a heuristic either ex ante or ex post todetermine the effectiveness of their decisions and strategies in improving healthcare quality In additionit can be used to analyze why actions or decisions taken achieved a given outcome and how best to proceedto effect further improvements on patient outcomes Last the model serves to focus attention on specificquestions for further researchKeywords Evidence-based medicine Evidence-based management Chronic healthcareClinical decision support system Healthcare informatics Physician incentivesPaper type Research paper

1 IntroductionEvidence-based management (EBMgt) developed as an attempt to take the principles ofevidence-based medicine (EBM) and adapt them to business management by refiningmanagement guidelines and best practices However what is the definition of EBMgt inhealthcare Managers should make decisions that the best evidence shows is most effectivein supporting the practice of EBM (Shortell et al 2007) Therefore it is important formanagers to know the principles of EBM criticisms of EBM and solutions as well as themajor challenges to the practice of EBM In Table I we present a summary as context for



0025-1747DOI 101108MD-10-2017-1010


19 March 201829 April 2018

Accepted 15 May 2018


The authors would like to thank Dr RA Ramanujan Diabetic Care Associates Binghamton NYfor invaluable discussions on his practice of patient-centered care for chronic diseases The authorsthank the reviewers and special issue editors for their valuable comments that improved thepaper significantly

2125

Application ofevidence-basedmanagement


Principles

Criticism

sSolutio

nsandim

plications

tomanagers

Challeng

ein

transferring

EBM

into

clinical

practice

1The

bestavailable

practiceshould

beused

2Evidenceshould

bebasedon

anevaluatio

nof

allevidence

3Baseclinical

decisionson

patientsrsquovalues

andpreferences

Randomized

ControlT

ests

(RCT

s)are

considered

ashigh

estqu

ality

evidence

Shortcom

ings

ofRCT

s1RCT

smay

lack

applicability

because

aItisbasedon

ldquoaveragerdquo

rand

omized

patient

bThe

clinical

guidelines

that

are

basedon

RCT

smay

ignore

variablesthat

may

affect

treatm

entoutcom

es2RCT

sdo

notpresentresults

that

consider

effectsof

multi-

orco-

morbidity

3Indu

stry

pressuresmay

indu

cebias

inresultsintrodu

ceconflicts

ofinterest

Solutio

ns1Broadeningof

RCT

popu

latio

nsm

akeRCT

smorerealistic

2Curapersonalismdashcare

ofthewholepersonmdashsee

principle3of

EBM

aAttem

ptto

mitigate

problem

ofno

sing

legu

idelineapplying

specifically

toan

individu

alpatient

bCa

nusethekn

owledg

eof

patientsrsquodistinct

profile

iTodo

thisn

eedto

usecompu

ting

technology

andinform

atics

iiCa

ndevelopldquom

edicine-basedevidencerdquo

Implications

ofthesesolutio

nsto

managers

1Managersshould

prioritizetheintegrationof

patient

preferencesandvalues

into

the

healthcare

theirorganizatio

nprovides

2Managersshould

understand

theshortcom

ings

ofgu

idelines

andavoidmeasuring

performance

very

narrow

lywith

respectto

guideline

adherence

3Managersshouldun

derstand

biases

ofresearch

andconsider

theinflu

ence

ofincentives

onall

partsof

theirorganizatio

n4Im

plem

entatio

nof

ldquomedicine-basedevidencerdquo

requ

ires

useof

healthcare

inform

atics

Problems

1Not

enough

timeforph

ysicians

tokeep

upwith

theincreasedrate

ofavailabilityof

evidence

(medical

know

ledg

e)2Researchevidence

istranslated

into

practice

usinggu

idelinesG

uidelin

eshave

issues

aHighdegree

ofvariationinuseofgu

idelines

bGuidelin

esdo

notconsiderconsequences

interm

sof

financialor

otherresourcesskills

orotherchanges(th

eseareim

portantto

managers)

cPa

tientsdo

notco-operate

infollowing

guidelinesorhave

differentexpectations

andaskfortreatm

entthat

deviatefrom

guidelines

dGuidelin

esthem

selves

arelow

quality

eDogmaticrelianceon

guidelines

isnot

optim

alSolutio

nsIncrease

adherenceanduseof

guidelines

by1Ch

anging

thepatterns

ofcaremdashuseprovider

educationremindersp

atient

education

decision

supp

ortincentives

2Im

provingleadership

acceptance

(som

ething

managerscanim

pact)

Table ISummary ofprinciples criticismssolutions andchallenges of EBM

2126

MD5610

the ideas presented within this paper Chronic healthcare is specially characterized byrecursive patient-physician interactions in which EBM is applied The Chronic Care Model(CCM) is an evidence-based model that identifies and organizes changes that improvepatient outcomes into discrete elements of effective healthcare systems for chronic illnesswith the goal of shifting the orientation and design of practice (Wagner et al 2005)This paper broadens the CCM (itself an EBMgt tool) identifying additional elements criticalto improved outcomes patient decision aids (PtDA) and healthcare informatics A newmodel that serves as a guide for chronic healthcare management is formalizedmdashtheInfluence Model of Chronic Healthcare This model can be used by managers both ex anteand ex post to determine why actions achieved a given outcome and how best to proceed

2 EBM criticisms solutions and challenges21 Principles of EBMIn this section we present the principles of EBM major criticisms and suggested solutionsand some challenges Table I provides a summary of this section

As defined by Sackett et al (1996) ldquoevidence-based medicine is the conscientiousexplicit and judicious use of current best evidence in making decisions about the care ofindividual patientsrdquo There exist three epistemological principles in EBM first the bestavailable practice should be used second that evidence should not be selected simplybecause it favors a claim but rather based on an evaluation of all evidence and third thatclinical decisions should be based in part on patientsrsquo values and preferences (Djulbegovicand Guyatt 2017) The most important facet of EBM is that individual clinical expertiseshould be integrated with the best available external clinical evidence from systematicresearch Two major risks to the practice of EBM are failure to include clinical expertise inthe decision process of a provider and failure to use a bottom up approach that considerspatientsrsquo choice (Sackett et al 1996)

22 Criticisms of EBMThe first criticism is that EBM relies on reductionism of the scientific method by strictadherence to the evidence hierarchy pyramid in which randomized controlled trials (RCTs)are prized as the highest quality evidence (Djulbegovic and Guyatt 2017) Empiricalstudies fail to confirm the superiority of RCTs in assessing the benefits of a given therapyit has been found that well-designed observational and randomized designs produceequivalent results (Horwitz and Singer 2017) In addition RCT data are often not availablefor issues important to clinical practice such as etiology diagnosis and prognosisof disease (Horwitz and Singer 2017) Results arising from RCTs lack a degree ofapplicability as they are for an ldquoaveragerdquo randomized patient and not for patients whosecharacteristics depart from those in the trial (Fava 2017) Compounding this problem isthe fact that the test population used in RCTs is highly selected to meet inclusion criteriaand excludes many of the patients who would be candidates for treatment It is estimatedthat studies of medications for asthma have excluded 95 percent of the target populationand a recent review of trials showed that women older adults and minorities wereunderrepresented (Horwitz and Singer 2017 Horwitz et al 2017) Particularly withchronic illnesses patients often have multiple conditions that rarely map to a singleguideline treatment of one condition must consider how the therapy may cause orexacerbate another RCTs typically do not present results that consider the effects ofmulti- or co-morbidity each of which affects every patient differently (Fava 2017Greenhalgh et al 2014 Horwitz and Singer 2017) For many conditions interventionsresulting in large improvements have already been developed and the science focuses onmaking marginal gains (Greenhalgh et al 2014) The implication of this is that RCTs are

2127


designed with large sample sizes which while enabling the achievement of marginaltreatment gains may overstate potential benefits and also underestimate potential harms(Horwitz and Singer 2017 Greenhalgh et al 2014) Further the use of placebo controlsleads to exaggerated assessments of benefits particularly when new therapies are nottested against existing ones (Horwitz et al 2017)

Clinical guidelines that are based upon RCTs may exclude information such asimpairment distress and well-being that can be assessed by reliable methods in favor ofldquohard datardquo such as laboratory measurements They may also ignore variables that affecttreatment outcomes such as expectations preferences motivations and patient-physicianinteractions By doing so such guidelines replace ldquojudgement with overly simplistic methodsthat create the appearance of quantitative precision that does not existrdquo (Djulbegovic andGuyatt 2017 Horwitz and Singer 2017 Fava 2017) As a result use of these guidelines canencourage formulaic approaches to medicine and automatic decision-making that disregardsthe potential lack of applicability of RCT results in clinical practice (Horwitz and Singer 2017Horwitz et al 2017 Djulbegovic and Guyatt 2017 Greenhalgh et al 2014)

Industry pressures exacerbate the shortcomings of RCTs Respected practitioners havenoted that the research agenda is set by industry with influential randomized trials largelydone by and for their own benefit (Greenhalgh et al 2014 Ioannidis 2016) Industry alsodefines what constitutes a disease or pre-disease ldquorisk staterdquo decides which tests andtreatments will be compared in studies chooses outcome measures for establishingldquoefficacyrdquo conducts trials in a way that is optimized to the ldquoqualityrdquo analysis that istypically done to gauge significance and publishes results in advance of non-industry trials(Greenhalgh et al 2014 Ioannidis 2016) Further investigators with substantial financialconflicts of interest serve on panels concerned with clinical guidelines the industry sponsorsmeta-analysis aiming to receive favorable conclusions and creates an incentive problemwith ldquogift authorshipsrdquo wherein ldquoclinical investigators flock to try to get co-authorship inmulticenter trials meta-analyses and powerful guidelines to which they contribute little ofessencerdquomdashall sources of bias (Ioannidis 2016 Fava 2017)

23 SolutionsThe medical community has proposed solutions to deal with the problems just describedWith respect to RCTs the US Food and Drug Administration has encouraged trialists tobroaden the populations studied in RCTs and some studies now use ldquopragmatic RCTsrdquo thatldquoemulate more closely the actual practice of medicine and foster more comparativeeffectiveness studiesrdquo (Horwitz et al 2017)

The most important solution proposed is to follow cura personalis or care of the wholeperson (Richardson 2017) Care of the whole person considers the patientrsquos feelings andexperience of illness and integrates ldquopsychological and social factors to achieve a fullerunderstanding of illness and to guide treatment and to paying greater attention to healthpromotionrdquo (Wagner et al 2005) This approach recognizes that the patientndashphysicianrelationship is critical and that shared decision-making should be a goal allowing bothpatient and physician to make care decisions that may not reflect what the ldquobest evidencerdquosuggests (Fava 2017 Greenhalgh et al 2014 Wagner et al 2005 Richardson 2017) Curapersonalis reflects an awareness that no individual guideline applies completely to anyindividual patient and that it is often unclear what a likely outcome would be when a giventreatment is administered to a particular patient with their own distinctive biological andbiographical (life experience) profile (Burke 2013 p 67 Institute of Medicine 2015 p 59Horwitz and Singer 2017 Horwitz et al 2017)

The preference for RCTs and the rejection of physician experience was warranted whenldquoexperiencerdquo was limited to the physician in question but due to advancements incomputing and informatics it is now possible to compile the collected experiences of tens of

2128

MD5610

thousands of physicians caring for hundreds of thousands of patients producing a data setlarger than any clinical trial and enabling consideration of patientsrsquo clinical courses underdifferent treatments and for patients with different histories (Horwitz et al 2017)Personalized care in this context must begin with a complete characterization of the patientusing data describing not only their physiology but their environment psychology socialand behavioral characteristics etc These data points collected repeatedly at differentpoints in the patientrsquos clinical course would form a patient profile The profiles could becompiled to form an archive physicians would then be able to inform decisions by findingapproximate matches that describe how similar patients responded to the proposedtreatment or to alternative treatments (Horwitz and Singer 2017) This has been termedldquomedicine-based evidencerdquo

24 ObservationsWe make a few observations that inform effective EBMgt First the idea of cura personalisis not only a solution to the ills of EBM as it is practiced today but actually a core tenet ofEBM Effective EBMgt should make integration of patient preferences and values intohealthcare delivery a priority Physicians exercise judgment as decision makers and thebest decisions for the care of patients rely on physicians using their judgment in evaluatingevidence and its applicability to a given patient Therefore while it is tempting to measureand minimize variance in physician adherence to best practice guidelines it is important tounderstand the shortcomings of these guidelines (particularly with respect tomultimorbidity) and to allow for variations in adherence that arise from patientsrsquopreferences and values Second we must recognize the influence of biases introduced byindustry (pharmaceutical and medical device companies) on the body of knowledge that isused to make decisions by managers and providers alike These biases are alsorepresentative of the outsized role financial incentives play in all aspects of healthcareThird EBM may be improved by using medicine-based evidence however achieving theimprovement in care quality represented by personalized care relies heavily on healthcareinformatics The use of healthcare informatics and financial incentives are furtherdiscussed as elements of the Influence Model of Chronic Healthcare (Section 4)

25 Challenge the effective transfer of EBM into practiceOne of the primary requirements of EBM is that high quality research evidence should betransferred into practice Implementation can be complex especially because changingprovider behavior can be difficult (Davies 2002) Research has consistently shown that thereis a gap between evidence and practice patients often do not receive care in accordance withscientific evidence or even receive care that is harmful or not needed ultimately increasingthe costs of care (Grol and Grimshaw 2003 Grol 2000) Studies have also shown thatidentification of the best treatment with high quality evidence to support its use is availableonly about 10ndash20 percent of the time (Institute of Medicine 2011 p 40) The amount ofmedical knowledge available is continually increasing and the rate of its increase is onlyaccelerating (Institute of Medicine 2011 p 41) Given the demands on physiciansrsquo time it isunsurprising that they are unable to keep pace it has been estimated that general internistswould need to read about 20 articles a day every day of the year in order to maintain theirknowledge of current practices (Institute of Medicine 2001 pp 41-42 Grol and Grimshaw2003 Institute of Medicine 2015 p 59)

Typically evidence is transferred into practice using guidelines Research examining theuse of guidelines in decision-making showed high degrees of variation an indication thatdissemination efforts tools to promote adoption of best practices and incentives mustbe expanded (Grol 2000 Institute of Medicine 2001 pp 13-14) Studies have examined thecharacteristics of guidelines that lead to better compliance and have shown that compliance

2129


is directly related to the type of health problem addressed (less compliance exhibited inchronic care) the quality of evidence supporting the recommendations the compatibility ofthe recommendation with existing values the complexity of the decision-making needed thelevel of clarity with which the desired performance is described and the amount of newskills and organizational change needed to follow the recommendations (Grol andGrimshaw 2003 Kitson et al 1998)

A case study of guideline implementation with respect to cholesterol managementshowed that obstacles to change included doubts about the scientific basis of the guidelineresistance to motivating patients to change their lifestyle perceptions that the guideline wastoo complex and increased workload and that patients demanded unnecessary tests(Grol 2000) This example highlights some common problems with guidelines Not only isthere often only evidence for a small portion of the decisions addressed by a guideline theconsequences of a guidelines use (in terms of financial considerations resources skills ornecessary organizational changes) have typically not been considered (Grol 2000)In addition patients who do not follow their treatment plan inherently do not co-operate inmaking the guidelines effective This may be due to a difference in a patientrsquos expectationthat leads to demanding actions or treatments that are unnecessary in the context of theguideline Again patients their knowledge and decision-making are determinants of carequality and the success of EBM They become even more important in chronic healthcareconsidering the recursive patient-physician interactions We discuss this as a criticalelement of the Influence Model of Chronic Healthcare (Section 4)

While evidence-based guidelines are a powerful and necessary tool in increasing thequality of care dogmatic reliance on guidelines should be avoided and their use ldquomakessense when practitioners are unclear about appropriate practice and when scientificevidence can provide an answerrdquo (Grol 2000) Some have noted that there are too manyguidelines of low quality not based on evidence not developed systematically or thatinclude vested interests of specific parties driven by ldquoa guideline industry and a potentialoverproduction of guidelines in many western countriesrdquo (Grol and Grimshaw 2003Greenhalgh et al 2014 Ioannidis 2016) Together these detract from the use of guidelinesby causing confusion and by creating a negative opinion of guideline use among clinicians(Grol and Grimshaw 2003) The opinions of physicians toward aspects of clinical practiceinfluence the quality of care this is also discussed as an element of the Influence Model ofChronic Healthcare (Section 4)

A critical factor in successful implementation of evidence is a healthcare organizationsrsquo(HCO) structure management and willingness to pursue quality of care (Grol andGrimshaw 2003) Even if physicians are aware of evidence and aim to change theirpractice it is not fully within their control to do so as it can be difficult to alter wellestablished patterns of care if the clinical environment does not support these efforts(Grol and Grimshaw 2003) An organizationrsquos capability to change and infrastructuredetermine the likelihood of success in implementations of medical guidelines(Davies 2002) Unfortunately both financial and organizational resources to assistproviders in implementation are often scarce (Davies 2002 Grol 2000 Grol andGrimshaw 2003) However changes to organizational structure marshaling of resourcesand initiatives to improve quality can all be achieved through effective managementdecisions Common strategies for quality improvement include provider educationprovider reminder systems and decision support audit and feedback patienteducation and shared decision-making organizational change and financial incentivesregulation and policy (Shojania and Grimshaw 2005 Grol 2000)

While no single strategy can be consistently relied upon to produce improvements the useof multiple strategies has indeed succeeded Lastly in the Diffusion of Innovations model theprimary driver of idea adoption is observing the proven success of peers who have already

2130

MD5610

adopted the idea (Rogers 1995) While this may be viewed as a ldquocatch-22rdquo scenario what ismost important is the achievement of a ldquocritical massrdquo which when reached spreads the ideaTo this end it is important to incentivize early adoption to ensure leadership acceptance of theidea to narrate to participants that the idea is spreading and desirable and to incept the ideainto groups where feedback and interactions will result in a foundation for idea adoption(Rogers 1995) Leadership acceptance is of the utmost importance and is directly informed bythe management of the organization vocal and visible leaders are necessary to promotechanges in organizational culture and priorities Ultimately an EBMgt approach to healthcaremust facilitate the implementation of changes needed to optimally practice EBM at every levelof the health care system patient provider and organization

3 Chronic healthcare challenges and solutionsBelow we focus on challenges in managing chronic diseases and summarize the CCMintroduced by Wagner et al (2005) that addresses some of these challenges In the nextsection we discuss gaps in CCM and propose an improved model which we call theInfluence Model of Chronic Healthcare

Unlike acute conditions many of the most common chronic conditions can be directlyattributed to specific patient behaviors The single most important behavioral risk factor isobesity which itself is rooted in a lack of physical activity and poor nutrition Along withtobacco use and excess alcohol consumption they represent behaviors that can be changedbut that account for 40 percent of all premature death in the USA (Milani and Lavie 2015)As a result one of the most important goals in effective chronic healthcare should be thechanging of patient behaviors

Current chronic healthcare delivery typically relies on the primary care physician as thefirst point of contact Given that the median length of these interactions are less than15 minutes and cover six topics little time is available to assess and address patient behavior(Milani and Lavie 2015) This is reflected in a 2006 study in which only 65 percent of obesepatients were advised to lose weight by their physicians (Milani and Lavie 2015) Even whenpatients are advised to change their lifestyles the rate at which they adhere to this advice isvery low (Milani and Lavie 2015) Thus a first challenge in chronic healthcare is that existingchronic healthcare delivery systems are not effective in changing patient behavior

In addition chronic disease patients typically receive only half the recommended processof care making additional interventions necessary and increasing the total cost ofhealthcare (Milani and Lavie 2015 Wagner et al 2001 2005) Thus a second challenge inchronic healthcare is that the quality of chronic healthcare that patients receive is deficientDeficient care is a result of four factors physician time demands rapidly expanding medicaldatabase therapeutic inertia and lack of supporting infrastructure (Milani and Lavie 2015Wagner et al 2001)

We now examine how use of the CCM addresses the challenges presented above

31 Challenge existing chronic healthcare delivery systems are not effective in changingpatient behaviorPatient involvement in the delivery of care is in keeping with the principles of EBM thesolutions to criticisms of EBM and ldquowith a cultural change in medicine over the past 20 yearsthe growing emphasis on patient autonomy and the associated priority given to shareddecision-makingrdquo (Djulbegovic and Guyatt 2017) In fact Wagner et al (2001) recount thefinding of a Cochrane Collaboration review which found that ldquoeven complex interventionsthat only target providersrsquo behavior did not change patient outcomes unless accompaniedby interventions directed at patientsrdquo underscoring the importance of systematic effortsto increase the knowledge skills and confidence in self-management that patients have

2131


Having defined support for patient self-management as one of the critical elements of theCCMWagner et al (2005) identify a well-tested strategy with five steps that should be appliedroutinely as the basis of a systematic approach to providing self-management support assesspatient behaviors attitudes and goals advise patients based on science agree on the problemgoal and plan of action assist patients in developing realistic goals and identify barriers toand strategies for reaching a goal and arrange for additional resources support etc

Much research has been done into patient compliance with their treatment plan A detailedlist of factors that influence patient compliance is given in the first column of Table II

32 Challenge the quality of chronic healthcare that patients receive is deficientEarlier we noted four factors causing deficient care here we examine each and how it isaddressed by the CCM

321 Physician time demands Wagner et al (2005) note that practices with low patientsatisfaction measures are often linked to ldquounhappy stressed providers who are eager forguidance in how to work with their patients more effectivelyrdquo Large overhead timedemands are a stressor that result in providers who feel they are not working with theirpatients effectively They go on to state that ldquovisit time is frequently implicated as afundamental barrier to more patient-centered interactionsrdquo and that ldquolonger morestructured (planned) visits are an important feature of effective chronic care and providegreater opportunity for assessment of patient concerns and progress collaborative supportfor self-management and treatment planningrdquo Managers aligning their organizationrsquos

Table IIInfluence factors onpatient and physicianbehavior

2132

MD5610

practices with the CCM evaluate the composition of practice teams and the division of tasksas part of delivery system design and reduce the time demands on physiciansImplementation of decision support and clinical information systems also reduce thedemands on physiciansrsquo time by streamlining their workflow

322 Medical knowledge base Nearly 2m articles are published a year and the doublingtime of medical knowledge is rapidly decreasing from 10 years (Milani and Lavie 2015Burke 2013) In addition to the volume of information a large percentage of the studiespublished are contradictory also known as medical reversals To cope with the expandingmedical database and to use medicine-based evidence to improve EBM management initiativesto implement decision support systems are important and are an element of the CCM

323 Therapeutic inertia The failure of a provider to increase or modify therapy whentreatment goals are not met is therapeutic inertia (Milani and Lavie 2015) The factorsinfluencing therapeutic inertia involve all three facets of care the provider the patient andthe healthcare system Methods to change patient behavior address their understandingsocial setting and preference-setting mechanisms and are addressed by self-managementsupport Approaches that include care of the whole person (cura personalis) andself-management support lead to activated patients In doing so they produce betterestimates of patient need and combined with reduced overhead time demands lead toproactive interventions Outright failures to initiate treatment are often a result of a failureto consider all available data points regarding patient health and are significantlyinfluenced by shortages in time Again decision support and clinical information systemscan have a positive impact along with delivery system redesign

324 Lack of supporting infrastructure In previous work Wagner et al (2001) note thatsimply taking existing systems and stressing them is not effective in improving carebut that systems themselves must be changed instead In the CCM there is a clear takeawaythat planning communications coordination and establishing roles are criticalmdashall issues thatmanagers can act on as part of the delivery system redesign and in the process create neededsupporting infrastructure (Wagner et al 2005) Further managersrsquo allocation of resourcesto implement decision support and clinical information systems necessarily create thesupporting infrastructure that is needed for improved chronic care

4 Influence model of chronic healthcare41 Why is there a need for this modelWhile the CCM achieves its purpose in compiling evidence-based practice changes thathave been shown to improve chronic care it does have drawbacks Typical managementdecisions may involve implementation of incentives the allocation of resources or thechange of operating policies and procedures As the CCM itself points out implementationcan mean re-evaluating the ldquostructure organization and functioning of practicesystemsmdashincluding their measurement systems incentives information handlingvisit design team function and so onrdquo (Wagner et al 2005) Studies of the effectivenessof CCM-based quality improvement efforts have shown considerable variation(Coleman et al 2009) This variation is unsurprising when one considers the variety ofpractice changes that may be implemented because of differences in organization countryincentive system existing IT infrastructure etc In addition changes resulting fromaddressing one element of the CCM may impact others

In complex organizations such as healthcare it is important for managers to have a senseof what effects will result from a decision and why without this knowledge organizationalcomplexity can lead to unequal unintended or cascading impacts on other areas that mayeven be out of the scope or control of the manager If the effects of a decision can beanticipated before decisions are made managers may be able to make better decisions If the

2133


effects cannot fully be anticipated it is still beneficial for managers to understand the keyareas of their organization A clear implication of how the key areas are linked and theinfluences one area has on another are missing from the CCM

Further information technology (IT) has only taken on an increased role in improving thequality of chronic healthcare and several elements of the CCM involve or benefit from itsexpanded use Research reveals that smaller practices or those with limited IT or non-physicianclinical staff would have greater difficulty implementing the CCM and improving outcomes(Coleman et al 2009) We feel that the use of IT in improving chronic healthcare can be bettercharacterized in the context of the following the use of medicine-based evidence improvedtools for self-management support and improved tools for communication coordination andplanning Researchers should also better understand why a technology solution may positivelyimpact behavior in theory but perhaps not always in practice as well as whether or not thesolution is cost-effective Reviewing cases of CCM implementation shows that the impact onhealthcare costs and revenues is uncertain and that ldquothe CCM recommends services and modesof delivery that are generally poorly reimbursed or not reimbursed at all in most fee-for-service(FFS) schemesrdquo (Coleman et al 2009) This quote brings focus to an omission in the CCM that isof importance to managersmdashhow do payment structures and financial incentives influencephysician and patient behavior

The Influence Model of Chronic Healthcare aims to fill these gaps and is presentedin Figure 1

bull Knowledge Base Managementbull Disease Registrybull Electronic Health Record

Computerized Clinical Decision SupportSystem f g

a

e c

dI II

IIIb

HealthcareInformaticsbull Can develop medicine- based evidence

Patient Decision Aids Traditional Influences

Prioritized Influences

bull Financial Incentivesbull Management-Implemented Incentives Secondarybull Knowledge shaping

Healthcare organisation1) Communication and Planning

Care CoordinationPhysician-staff CommunicationPatient OutreachVisit Planning

2) Self-Management Support

EBM

Legend

Healthcare DeliverySystem

PhysicianPatientMotivations

Informations Systems

JointDecision

Provider- PatientSynergy

Primary

Patient Behavior

Physician Behavior

Figure 1Influence Model ofChronic Healthcare

2134

MD5610

The elements of practice change found in the CCM are incorporated as are additionalelements At the heart of the model is the patientndashphysician relationship Both physician andpatient are influenced by the HCO which we define as the people organizational structurepolicies and procedures In keeping with the CCM the HCO influences both patient andphysician through communication and self-management support A list of abbreviationsused in this paper is provided in the Glossary at the end

Computerized clinical decision support systems (CCDSS) can impact provider behaviorand have been identified as a solution addressing care deficiencies and variations inguideline adherence (two challenges mentioned in Section 3) Similarly another form of ITcan impact patient adherence to treatment plans and aid in self-management PtDAThe effectiveness of the HCO in communication and planning CCDSS and PtDA isdependent on quality of healthcare informatics an element we draw attention toThe effectiveness of CCDSS and the impact of the HCO on physician behavior is modulatedby other influences some of which can have a very large impact We attempt to enumerateand define these influences as Prioritized Influences on Physician Behavior Similarly theeffectiveness of the HCO and PtDA in impacting patient behavior is modulated by otherinfluences (psychological social behavioral) that are typical to all humans which we callTraditional Influences and briefly discussed in Section 3 Next we will examine theconstruction of the model and each element in further detail summarizing key points fromthe literature with regard to their efficacy and mechanism of impact on one another

42 Construction of the model421 The patientndashphysician relationship (see ldquoardquo in Figure 1) From the initial diagnosisevery interaction is defined by physician behavior and by patient behavior Physicianbehavior can be defined as being composed of first the decision process of determiningwhat medical intervention should be undertaken and second the formulation and executionof a treatment plan More generally physician behavior is the application of EBM to thespecific case presented by the patient Patient behavior can be defined as being composed offirst adherence to the treatment plan and second implementation of lifestyle changes thatare either preventative or aid in management of the chronic condition While the behavior ofphysician and patient is separate a third subset of the patientndashphysician relationship mustalso be considered the synergy between patient and physician that results in degrees ofjoint (shared) decision making

422 Healthcare Organizationmdashcommunication and planning self-management support(see ldquobrdquo in Figure 1) The CCM identifies key elements of practice change that improve chronichealthcare through the redesign of systems toward a more patient-centered approach Fromthe perspective of chronic healthcare delivery improvement a critical function of the HCO is todefine team member roles and tasks and communicate and coordinate treatment plansbetween patients physicians and other support staff In this model we define the HCO as thepeople organizational structure and policies and procedures that are needed to provide thisfunction as well as the person-to-person component of self-management support (educationfollow-up etc) which the CCM also identifies as a key change element This definitionseparates the information and technology infrastructure with the purpose being to highlightthese human activities as being shaped by a unique set of managerial decisions dealingspecifically with personnel Some examples that conform to the CCMrsquos key elements of changewould include choosing a support staff to physician ratio allocation of tasks betweenphysicians and staff changing policies and procedures for treatment leadership support ofimprovement development of agreements facilitating care coordination or even teaming withcommunity organizations to fill gaps in patient education These are all areas in whichmanagers and the decisions they take can have significant impact Later we discuss this

2135


impact in the context of management-implemented incentives to explicitly change physicianbehavior but it should be noted that the chosen organizational structure policies andprocedures and resource allocation decisions adopted by management all have anindirect impact on physician behavior in that they help to define the environment in whichphysicians operate This is the meaning of the linkage in Figure 1 (noted as II) between theHCO and physician behavior

Organizations with good communication and planning can conduct more effectivepatient outreach are able to better assess patient concerns may be able to give patientslonger and more structured visits and are able to give collaborative support forself-management All are components of patient-centered chronic care that EBM has shownto lead to engaged patients who are more likely to adhere to a prescribed treatment planThe quality of communication between physician and staff is directly related to the abilityto coordinate care Care coordination reduces the demands on physiciansrsquo time and in doingso removes a barrier to the optimal practice of EBM In other words the change effortsdescribed above and suggested by the CCM encourage a patient-centered approach thatinherently attempts to change patient behavior improving treatment compliance andhopefully resulting in better outcomesmdashindicated by the linkage in Figure 1 (noted as III)between the HCO and patient behavior

423 Traditional influences on patient behavior (see ldquocrdquo in Figure 1) In the previoussection we identified the HCO and its role in creating a patient-centered approach to chronichealthcare using in part a greater focus on the provision of self-management supportThe criticality of self-management support in the CCM is reflective of the outsized role thatpatient behavior has in treatment adherence and improvement outcomes Three questionsnaturally follow what are the influence factors on patient behavior how might they impedetreatment adherence and can they be mitigated or changed by the HCO Earlier in Section 3we detailed a systematic approach to providing self-management supportmdashinherent in thisapproach is an attempt to modulate the factors that influence patient behavior The firstcolumn of Table II details an unexhaustive list of major influence factors that each impactthe level of resolve that a person has in adhering to their treatment The improved caredelivery efforts HCOs undertake are provided against a backdrop of these mitigatingfactors While they are straight-forward and self-explanatory they can be quite challengingfor the HCO to address for example the influence of the patientrsquos social support network issignificant and can reach three degrees of separation (Milani and Lavie 2015 Wagner et al2001) In the next section we detail technological methods of improving self-managementsupport care coordination and directly aiding patientsrsquo decisions Finally we note that welater discuss financial incentives and physician payment systems the structure of thesesystems can indirectly have impacts on patient expectations and satisfaction with theirtreatment Some payment systems have the impact of limiting patientsrsquo options withregards to specialist services which can conceivably reduce patient satisfaction Patientsatisfaction is a critical factor in treatment adherence and improved clinical outcomes aswell studies have shown that better outcomes result from providers listening thoughtfullyand that even flu shots may be more effective depending on the mood of the patient (Owen2018) Logically patients who feel they are treated well are more likely to exhibit ldquobuy-inrdquo toa treatment plan and consequently exhibit improved adherence

424 Healthcare informatics (see ldquodrdquo in Figure 1) Informatics can be defined as amultidisciplinary area which draws on computer and social sciences to study the interactionbetween humans and computer information systems A key philosophical underpinning toinformatics is the use of computer technology and information systems in a manner thatallows improved human decision-making that is knowledge-based (eg statistical analysisof data) or in other words evidence-based Naturally then the use of informatics is a

2136

MD5610

priority in EBMgt (Wan 2006) Technology is only a component of informatics insteadinformatics recognizes that designing technologies having them implemented in a real-world setting and the effect they have on the individuals groups and organizations is not apurely technical matter Healthcare informatics ldquodraws upon the social and behavioralsciences to inform the design and evaluation of technical solutions and the evolution ofcomplex economic ethical social educational and organizational systemsrdquo (AmericanMedical Informatics Association 2011) We use this term to specifically highlight that whilean element of the CCM is clinical information systems it is important to go beyond patientregistries reminder systems and information sharing across teams and providers the terminformatics can be used to include other commonly used terms in the literature such ashealth information technologies and information and communications technology (ICT)In addition informatics addresses the question of what technological designs are effectiveand the reasons why technology may not be adopted by using social and behavioral scienceIn our model two elements highlight and encapsulate these reasons the TraditionalInfluences on Patient Behavior discussed above and the Prioritized Influences on physicianbehavior we present later In Section 23 we discussed medicine-based evidence and thepresence of large-scale data and computing technology could make its application practicalagain this is a challenge that falls squarely within informatics Research has shown thatweb-based patient-specific decision support showed the potential to improve diabetes careinternet-based health education was effective in glycemic control and interventions usingICTs for the control of hypertension and treatment compliance were effective (Garcia-Lizanaand Sarria-Santamera 2007)

Broadly speaking we believe the impact of health informatics can be categorized into threeareas Technology applied to directly impact change and improve the following patientbehavior physician behavior or the communications planning and self-management supportfunctions of the HCO

CCDSS are the primary evidence-based IT tool for addressing physician behavior andimproved guideline adherence Similarly the literature defines a type of evidence-basedtool to which IT is being applied and specifically impacts patient behavior PtDA Theseinformation system components of the model are highlighted in green in Figure 1

Finally it should be noted that a single application of health informatics can span all threeareas ICTs can include telemedicine data collection PtDA and internet-based patienteducation perhaps even united through a single interface (Celler et al 2003 Dorr et al 2007Garcia-Lizana and Sarria-Santamera 2007) The telemedicine aspect is an example ofinformatics influencing the HCOs care coordination patient outreach and visit planning(see I in Figure 1) Simultaneously such an ICT could influence physician behavior via thebenefits on physician time demands made possible by fewer scheduled office visits andsimultaneously influence patient behavior by improving satisfaction and treatment adherence(see II III respectively in Figure 1) This discussion reinforces the three important areas weidentified where healthcare informatics can have an influence an organizationrsquos IT andinformatics capabilities drive its patient education programs which are part of the HCOrsquosself-management support and can take the form of internet-based education The same systemscan also use informatics in automated scheduling and medication reminders as well as thetreatment plan a physician chooses based on the best available evidence The latter is driven byCCDSS which require informatics expertise in maintaining and managing a clinical knowledgebase a disease registry or very importantly mdashan electronic health record

In fact the impetus for developing the model explained in this paper originated because ofa research collaboration with an endocrinologist who has been practicing physician for over35 years (Banerjee et al 2016) Recently the physician has developed a robust flexible userfriendly web-based patent pending proprietary mobile health application (app) called

2137


CheckMyVitalsreg In its current form the app is being used by the physician in his clinicalpractice for over four years The app has a built-in CCDSS enabling providers to make timelyand informed patient interventions The app can be implemented on a large population ofpatients without making major infrastructural changes is independent of operating systemslocation and access to internet communicates instantly with the provider to make immediatetreatment modifications if needed allows multiple providers in the group to communicateinstantaneously through one portal to create a single continuum of care model for the patientssends alerts to patients reminding them to enter vitals on time keeps complete track of patienthistory and archive data when needed allows broadcasting chats and connecting providersreal time with patients to intervene allows for patients to request refills and medicationchanges and sends a summary document automatically to a patientrsquos electronic medicalrecord so that they can have a macro view of their readings So far this app has been used bymore than 2200 patients in his diabetes and hypertension clinic

This new software enabled a better method of communication between patients andproviders and overcame the issues related to mobility and cost The resulting timelyinterventions had the effect of providing preventative care that reduced the likelihood ofpatients needing care in emergency departments or in patient hospitals As far as we knowthis is the only fully integrated app that is in regular use in a clinical practice in the USA thatenables patients to continuously communicate data on their vitals while the providermonitors intervenes and gives timely feedback More information about the app is providedin Banerjee et al (2016)

4241 Patient decision aids (see ldquoerdquo in Figure 1) As mentioned previously patients oftendo not self-manage and there exists the possibility for ICTs to play a role in addressing thisproblem (Celler et al 2003) PtDAs are an evidence-based tool that can positively impactpatient behavior and the quality of chronic healthcare PtDAs are particularly suited for usein chronic healthcare because they are designed to aid in decisions that can be characterizedas ldquopreference-sensitiverdquo the best choice depends on patientsrsquo values or preferences for thebenefits harms and scientific uncertainties of each option (OrsquoConnor et al 2004) PtDAs arealso another area in which technology can be used to great effect Mobile applicationsoftware and wearables have been shown to have positive results in effecting lifestylechange for chronic disease patients (Milani and Lavie 2015) They engage patients inthe care process which leads to patients having greater satisfaction and turns them intoactive rather than passive participants simply receiving care (Milani and Lavie 2015Wagner et al 2001) PtDAs supplement the patientndashphysician interaction providinginformation about the choices facing the patient and the outcomes that can be expected(OrsquoConnor et al 2004) Another example is ICT allowing chronic care patients to monitorblood pressure and sugar levels at home while participating in a remote consultation with ahealth professionalmdashthe very definition of a PtDA (Wan 2006)

OrsquoConnor et al (2004) state three key elements common to the design of PtDAsinformation provision values clarification and guidance in coaching in deliberation andcommunication Studies have shown that when PtDAs are used to supplement counselingthey have positive effects on decision quality as evidenced by smaller proportion of patientswith unrealistic perceptions of the chances of benefits and harms less psychologicaluncertainty because of feeling uninformed and lower proportion of patients who are passiveor undecided Despite this four barriers to the implementation of PtDAs have also beenidentified awareness of the existence of an appropriate PtDA for a particular clinicaldecision situation accessibility of PtDAs acceptability issues (eg PtDAs must be up-to-date attractive and easy to use not require additional cost time or equipment) andmotivations to use PtDAs (eg saving time avoiding repetition not requiring extra callsfrom patients potentially decreasing liability and wait-list pressures)

2138

MD5610

OrsquoConnor also notes that patient ldquodecision support as a consciously planned clinicalintervention is particularly needed for highly prevalent preference-sensitive situations inwhich poor-quality decision-making is likely to generate unwarranted disparities inhealth carerdquo this perfectly describes the most common chronic conditions encounteredThis decision support could be provided via clinic or hospital-based patient educationprograms freely on the Internet or through insurance plans (OrsquoConnor et al 2004Garcia-Lizana and Sarria-Santamera 2007) The latter are particularly incentivized to do soas PtDAs can contain the costs they face for a given patient a topic we touch upon inSection 425 in a discussion of financial incentives and physician payment systems

4242 Computerized clinical decision support systems (see ldquofrdquo in Figure 1)Computerization of clinical decision support is another important area of application ofICT and is captured in the CCM as an element along with another element that is necessarilyencapsulated by healthcare informatics clinical information systems (Celler et al 2003)The literature finds some essential functions of CCDSS as follows (Roshanov et al 2011Garg et al 2005)

bull Characteristics of individual patients are matched to a computerized knowledge baseand software algorithms generate patient-specific recommendations

bull Practitioners healthcare staff or patients can manually enter patient characteristicsinto the computer system alternatively electronic medical records can be queried forretrieval of patient characteristics

bull Computer-generated recommendations for diagnosis treatment patient educationadequate follow-up and timely monitoring of disease indicators are delivered to theclinician through the electronic medical record

Because of the prevalence of chronic disease and its nature as a condition that must bemeasured managed and treated over time it is possible to generate large volumes ofdata from which evidence can be extracted In Section 23 the topic of medicine-basedevidence was discussed as a solution enabling patient-centered care The design andimplementation of a patient profile archive with matching functionality is one example ofan application of healthcare informatics in fact a public-private partnership (HealthcareCost and Utilization Project) assembled healthcare data system across the entire USAusing informatics (Wan 2006)

Research into CCDSS has been wide varied and generally accepts that there existspotential to improve care in many instances improving processes of care such astreatment and monitoring patient outcomes such as blood pressure and cholesterol levelslevels of guideline and treatment adherence and user satisfaction (Roshanov et al 2011Dorr et al 2007) CCDSS have been shown to enhance clinical performance for diagnosisdrug dosing preventive care diabetes and hypertension Research also shows thatCCDSS used together with an electronic medical record produced greater improvementsthat using automatic prompts rather than user initiation had better performance thatreminders and information brought to the attention of a physician should be timely andrequire their acknowledgment that physicians should be given personalized feedback toimprove adherence that CCDSS should be integrated into workflow and be designedwith a view toward speed (a major determinant of user satisfaction and acceptance)(Garg et al 2005 Hunt et al 1993 Bates et al 2003) We note that the last few pointsaddress usability a topic central to informatics

The literature also identifies some issues again the research is dominated by anemphasis on RCTs (whose drawbacks were discussed above in the context of EBM) whichare very useful for studying system performance or specific changes in clinical practicebehaviors However here too they have a drawback they are not well suited for

2139


determining the factors that influence whether systems are used why they may not beused or explain variations in the effectiveness of a system in different environmentsSimultaneously very few CCDSS have been independently evaluated in clinicalenvironments and while CCDSS were shown to be cost-effective in some cases thisaspect has not been well studied (Hunt et al 1993 Kaplan 2001 Garg et al 2005Roshanov et al 2011 Dorr et al 2007)

Some studies found that physicians failed to use CCDSS systems despite demonstratedbenefits a symptom of the problem that physicians often fail to comply with guidelineswhether or not they are incorporated into a CCDSS and even in cases where they agreedwith the systemrsquos recommendations (Kaplan 2001 Garg et al 2005) Few studies haveexamined why CCDSS may be effective or may fail or why user experiences may fall shortof expectations (Kaplan 2001 Roshanov et al 2011) This highlights the need forunderstanding CCDSS in the context of informaticsmdashusability is important and bothbehavioral and cognitive science play a role for example simple one screen interventionshave proven more effective as has limiting requests for information from physicians butphysicians still strongly resist suggestions when alternatives are not given even if theaction they go ahead with may be counterproductive (Bates et al 2003) For managers to beable to make more informed decisions future trials with ldquoclear descriptions of systemdesign local context implementation strategy costs adverse outcomes user satisfactionand impact on user workflowrdquo are needed (Roshanov et al 2011) Finally most studieslooked at CCDSS that were implemented using research funding commercially availablesystems face added costs for support personnel as well as the constraints of compatibilityamongst information systems system maturity and upgrade availability (Garg et al 2005Roshanov et al 2011)

This section has dealt with health informatics and identified three key areas in whichspecific IT systems can be used to improve healthcare in accordance with the evidence-basedchanges identified in the CCM The previous discussion of CCDSS illustrated the human sideof implementation It showed that systems should be designed with the user in mind andthat in some cases it can be difficult to change behavior even if the correct informationand evidence is being communicated At the same time authors of systematic reviews ofthese IT systemsrsquo performance and efficacy have lamented a lack of understandingregarding why systems succeed in changing physician behavior in specific instancesThis is caused by several factors a preference in publications for RCTs which areconsidered rigorous but are not the gold standard in behavioral research but also a lackof research from a multidisciplinary perspective While the literature contains mentions ofstudies of usability user satisfaction and user workflow there are larger questions thatremain unaddressed what incentives are in place that may influence physician behaviorand what are the effects of these incentives In the next section the element we introduce tothe model provides a taxonomy of the broad structural incentives that are commonlypresented to physicians managersrsquo ability to change these incentives and the impact theseincentives have on healthcare quality

425 Prioritized influences on physician behavior (see ldquogrdquo in Figure 1) In an idealizedsetting physician behavior would always result in achievement of a baseline goal thehighest quality healthcare resulting from clinical practice in accordance with the principlesof EBM However in practice physician behavior often deviates from this optimal scenarioEarlier this paper discussed some of the impediments to the practice of EBM The causes ofdeviation in those cases were due to influences exerted because of shortcomings inprocesses or organizational configuration However there exist influences at a highersystemic level that impact physicians and unlike other influences are difficult to mitigatethrough the action of an individual physician or in some cases even their managers

2140

MD5610

Among these higher-level influences some are more impactful than others It is ourcontention that these influences are prioritized directly by the nature and structure of thehealthcare system in general and then the HCO in specific These ldquoPrioritized Influences onPhysician Decisionsrdquo can be further categorized into two types The first type are ldquoPrimaryInfluencesrdquo overt identifiable incentives that we label as either ldquoFinancial Incentivesrdquo(relating to physician payment systems and characterized by limited manager ability tochange) or ldquoManagement-Implemented Incentivesrdquo (designed to enact change within anorganization) The next type are ldquoSecondary Influencesrdquo subtle and not uniquelyidentifiable they instead serve to shape the knowledge and opinions of physicians Assubordinate influences they are also susceptible to modification by Primary InfluencesThese influences are detailed in the second column of Table II and discussed further below

4251 Primary Influences-Financial Incentives While antithetical to professionalmedical practice wherein practitioners have a duty to their patients above all else financialincentives are inextricable from the capitalist ideology and healthcare is by no meansimmune to their influence Physicians may claim that they are immune to the effects but thestructure of physician payment systems today and the widespread use of explicit financialincentives indicates that they may have an impact Indeed there is precedent for the ideathat physicians may be subliminally influenced reflected in the acknowledged need fordouble-blind clinical trials (Hillman 1990) Some have argued that using financial incentivesto change clinical behavior asks physicians to consider their self-interest and in doing socompromises the patient-centered approach that has been described in this paper as centralto improving chronic healthcare but researchers have also found that intrinsic motivationsplay an important role in physician decision making and strong ethics dilute or remove theimpact of incentives to provide poor care as a result of physicians prioritizing their own self-interest (Rodwin 2004 Gosden et al 2001) Situations in which more than one treatmentoption is available and a clear decision is not available are at most risk for being influencedby financial incentives but also by CCDSS (Bates et al 2003 Hillman 1990 Gosden et al2001) Overall the influence of financial incentives is far-ranging eg the structure and formof regulation on healthcare the co-opting of clinical research by pharmaceutical companiesand medical suppliers and even private firms that encourage physicians to prescribe orrefer patients by offering ownership stakes (Rodwin 2004) From the perspective ofimplementing evidence-based changes outlined in the CCM researchers have stated ldquothatsome type of external financial incentive and quality improvement support may be essentialfor widespread practice change especially for small practicesrdquo (Coleman et al 2009)

From the perspective of physicians these high-level incentives are most immune to changewhen considering physician payment systems all of which create incentives (Rodwin 2004)Medical services have traditionally been provided as FFS and providers would decide theappropriate treatment However the FFS model of physician payment creates incentives thatresult in overtreatment some have argued that physicians pursue target incomes and sowould raise prices or induce demand (Gosden et al 2001 Frolich et al 2007) Research hasfound that retrospective payment structures such as FFS ldquocrowd outrdquo intrinsic motivationsand they resulted in lower quality of service (Green 2014)

In response managed care organizations proliferated and instead began payingphysicians through capitation wherein physicians are paid a fixed amount per patient permonth (Green 2014 Robinson et al 2004) Capitation creates its own problems Capitationas well as salary-based payment systems may result in under-treatment (Gosden et al 2001Robinson et al 2004 Hillman 1990) Under a salary-based payment system a physicianrsquosincome is fixed and an incentive arises to minimize personal costs (such as effort) byselecting low-risk patients writing prescriptions and making referrals (to shortenconsultations) or by minimizing the number of office visits (Gosden et al 2001) Capitation

2141


payment systems incentivize limitations on referrals which compromises care in manycases using withholding accounts that reduce physician pay and in the process reduce jobsatisfaction of physicians (Grumbach et al 1998 Hillman 1990) Capitation can reduce costsby broadening the scope of services provided but also shifts to the physician the risk ofattracting patients who need less care than what the capitated payment is and alsocreates inadequate rewards for new procedures that may have positive cost-benefits(Robinson et al 2004 Rodwin 2004) At the same time capitation creates an incentive toprovide preventive care as this would reduce future costs and result in dollar-for-dollarincreases in physician payments (Gosden et al 2001)

As these issues became apparent capitation payment systems have adapted today toinclude some measures of quality but this poses problems for managers and researchersalike as there is not a universal definition of quality and instead measures such as patientsatisfaction process compliance or patient outcomes such as readmission rates are used(Armour et al 2001 Porter and Kaplan 2016) As payment systems further evolved to betterincentivize pay for performance has been introduced (Green 2014) Unfortunately pay forperformance simply encourage the overprovision of services listed under the defined qualitymeasures and there is not clear evidence of reduced costs or improved service qualityThe debate is unresolved and many hybrid payment systems exist to combine the need forproductivity encouraged by FFS and the need for cost-reduction which is encouraged bycapitation If not by design physicians contracting with multiple organizations may be insuch a hybrid system de facto (Green 2014 Robinson et al 2004)

Other alternatives include systems where physicians are paid salaries keyed tomeasures of productivity (claims billed paid visits procedures etc) FFS adapted toprovide reimbursement for care coordination or services outside of traditional office visiton a capitated basis (ie additional monthly payment for these services) and capitationwith added FFS for preventive vaccinationsscreenings (Robinson et al 2004) Bundledpayments are an emerging payment system that purports to fix some problems In thissystem a single payment is made for care for a patientrsquos medical condition across theentire care cycle (Porter and Kaplan 2016) Hybrid payment systems create even morecomplications in determining what the impact of incentives are in explaining how theywork to physicians and finally in administering them but some have suggested thatimpacts may be estimated as a linear combination (Gosden et al 2001 Armour et al 2001Robinson et al 2004)

Another important consideration is the organizational structure of payments Typicallymanaged care organizations act as intermediaries between the insurer and physicianin a dual-principal agent relationship and localized medical groups and independentphysician associations (IPAsmdashnetworks of physicians who contract together) can addanother layer of complexity In the process they can blunt the impact of incentives as thephysician is paid on a contract designed by someone who is not receiving the services(Robinson et al 2004 Armour et al 2001 Green 2014) In addition there can be an incentivemisalignment if a physician is paid FFS but the group is contracted on capitated basis(Robinson et al 2004) Research exists characterizing the tendencies of medical groups andIPAs toward FFS or capitation larger groups can use peer monitoring and pressure toensure productivity while also being at risk of free-riding but large IPAs might begeographically diverse (Robinson et al 2004)

Unfortunately for managers changing the physician payment system is likely out oftheir control Nevertheless we have illustrated the range of incentives and effects thatvarious types of systems in use today create We have also mentioned that physicians havestrong intrinsic motivations and peer monitoring and pressure likely only strengthens theseagainst incentives that rely on self-interest (these would be classified as Secondary

2142

MD5610

InfluencesmdashKnowledge Shaping described later) Managers would benefit fromconsidering the backdrop of incentives they cannot control and understand what impactthey may have on aspects they can control For example if a manager is investigating adeficiency in service provided it may be worth considering whether the inability to receivepayment for a service may be to blame In the next section we consider the set of incentivesthat managers can control

Primary Influences-Management-Implemented Incentives Management-implementedincentives are the method by which quality improvement can occur When used effectivelythey should maximize profit quality andor efficiency and should align with and supportthe practice of EBM These incentives can be changed or influenced by management andthey may be financial or non-financial (eg extra on-call duty) The use of non-financialincentives particularly penalties may mitigate the expected results of financial incentivessignificantly (Hillman 1990) To the extent that management can institute or changeperformance-based incentives they may be able to change physician behavior and weseparate them to highlight this fact though they may be financial in nature Simultaneouslymanagement-implemented incentives may arise indirectly out of resource allocationdecisions or from redesign of the HCO

We have established that patient behavior through greater treatment adherence is amajor driver of better chronic healthcare outcomes it can dominate the role of the physicianwhich means that incentives should be designed with this in mind Research seems to showthat small rewards do not motivate physicians toward improved preventive care(Town et al 2005) Studies that have been done have multiple shortcomings including alack of data on the size of incentives and whether they were cost-effective Simultaneouslymany have found a consistent lack of awareness of the size and magnitude of financialincentives by physicians themselves (Town et al 2005 Grumbach et al 1998)

The literature identifies many unanswered questions ldquoHow large an incentive does ittake to change behavior Are incentives cost-effective What is the best way to structure anincentive How does the framing of the incentive affect behavior What role does thephysician practicersquos organizational structure play in determining the effectiveness of anincentive What is the threshold at which specific financial incentives reduce the quality ofcare Are financial incentives the best way to induce practice changes that are persistent inthe long run instead of IT How do non-financial measures magnify or counterbalancefinancial incentives (Town et al 2005 Hillman 1990)rdquo

4252 Secondary Influences-Knowledge Shaping This paper has argued that EBMpractice should allow physicians to exercise judgment especially in the context oftreatment decisions that reflect the values and preferences of patients We must alsorecognize that provider decision making is not always going to rely on the strongest orbest evidence but is also subtly influenced by factors that shape each individualphysicianrsquos body of knowledge and personal opinions Examples of knowledge shapinginfluences are an individual physicianrsquos cumulative clinical experience the clinicalresearch that they have read (as discussed this secondary influence has itself beensubverted by the primary influence of financial incentives) and their contacts andcommunication within their professional network A physicianrsquos opinions may beinfluenced by the norms in his practitioner community As mentioned in the discussion offinancial incentives peer monitoring and pressure is thought to positively impactphysicians by mitigating the impact of financial incentives reinforcing intrinsicincentives influencing physicians to adhere to cost constraints or to ensure quality Thedirection and magnitude of these impacts are not obvious and should be investigated(Town et al 2005 Hillman 1990) Alternatively those same norms could be influenced bypharmaceutical representatives and corporate marketing We note that this is perhaps an

2143


example of a secondary influence that can be subverted by the primary influenceof management-implemented incentives (eg restrictions on marketing to physicianson premises)

43 Takeaways and use of the modelIt is clear from the previous section that the effects of financial incentives can be variedbased on payment system organizational structure and many other factors

Simultaneously our presentation of the other important influences and elements thatshould be considered in improving chronic healthcare has highlighted the need formanagers to understand the ldquobig-picturerdquo which our model aims to better illustrate

Use of the model will be dependent on the context of user Without being exhaustivewe provide some examples for managers

bull If managers are looking at ways to improve physician performance in chronic carewe posit this can be done by implementing computerized clinical decision supportor in the form of management incentives to change physician behavior

bull For managers who may be able to define the form of incentives offered by changingthe payment system or by offering explicit incentives it is useful to carefullyconsider how physician clinical behavior may be impacted

bull For managers who have control over redesign of chronic care delivery systems wehave highlighted that reducing physician time demands is beneficial so perhaps thisnecessitates focusing on workflows and task distribution something that is alsoideally done with a view toward patient outreach and self-management support(which we have identified as the HCO) This may also suggest the use of IT such asweb education if that were just a beginning it may be improved by integration ofvisit planning data collection and patient decision aid perhaps in the form of amobile application

bull For managers considering implementing or allocating additional resources towardimproving efficiency and the quality of care the model makes clear that a focus oninformatics is important and that IT in the forms of PtDA and CCDSS can havebenefits In addition when the effectiveness of these systems is being evaluatedmanagers must consider also the influence factors that may be impeding uptake ofnew systems either by physician or patient

It is hoped that presentation of this model may even influence managers and researchers toconsider and investigate these factors pre-implementation or even in study design as manyother authors have also called for

5 ConclusionChronic healthcare is specially characterized by recursive patient-physician interactionsin which EBM is applied As a result effective EBMgt of chronic healthcare mustrecognize that quality of care is improved through EBM This paper presented the currentpractice of EBM and the criticisms and challenges to EBM that are borne out ofdeficiencies in care quality The discussion of the CCM to improve the practice of EBM andchronic healthcare led to the synthesis of a new model that serves as visual guide forchronic healthcare managementmdashthe Influence Model of Chronic Healthcare This modelcan be used by managers either ex ante or ex post to determine the effectiveness of theirdecisions and strategies in improving healthcare quality In addition it can be used toanalyze why actions or decisions taken achieved a given outcome and how best toproceed to effect further improvements on patient outcomes

2144

MD5610

GlossaryEBM Evidence-based MedicineEBMgt Evidence-based ManagementRCT Randomized Controlled TrialCCM Chronic Care ModelHCO Healthcare OrganizationIT ICT Information (and Communication) TechnologyPtDA Patient Decision AidCCDSS Computerized Clinical Decision Support SystemsFFS Fee-For-ServiceIPA Independent Physician Association

References

American Medical Informatics Association (2011) ldquoWhat is informaticsrdquo available at wwwamiaorgfact-sheetswhat-informatics (accessed October 10 2017)

Armour BS Pitts MM Maclean R Cangialose C Kishel M Imai H and Etchason J (2001)ldquoThe effect of explicit financial incentives on physician behaviorrdquo Archives of Internal MedicineVol 161 No 10 pp 1261-1266

Banerjee A Ramanujan RA and Agnihothri S (2016) ldquoMobile health monitoring development andimplementation of an app in a diabetes and hypertension clinicrdquo 2016 49th Hawaii InternationalConference on System Sciences (HICSS) IEEE pp 3424-3436

Bates DW Kuperman GJ Wang S Gandhi T Kittler A Volk L Spurr C Khorasani RTanasijevic M and Middleton B (2003) ldquoTen commandments for effective clinical decisionsupport making the practice of evidence-based medicine a realityrdquo Journal of the AmericanMedical Informatics Association Vol 10 No 6 pp 523-530

Burke J (2013) In Health Analytics Gaining the Insights to Transform Health Care John Wiley ampSons Inc Hoboken NJ

Celler BG Lovell NH and Basilakis J (2003) ldquoUsing information technology to improve themanagement of chronic diseaserdquo The Medical Journal of Australia Vol 179 No 5 pp 242-246

Coleman K Austin BT Brach C and Wagner EH (2009) ldquoEvidence on the chronic care model inthe new milleniumrdquo Health Affairs Vol 28 No 1 pp 75-85

Davies BL (2002) ldquoSources and models for moving research evidence into clinical practicerdquo Journal ofObstetric Gynecologic amp Neonatal Nursing Vol 31 No 5 pp 558-562

Dixon-Fyle S Gandhi S Pellathy T and Spatharou A (2012) ldquoChanging patient behavior thenextfrontier in healthcare valuerdquo Health International Vol 12 No 12 pp 64-73

Djulbegovic B and Guyatt GH (2017) ldquoProgress in evidence-based medicine a quarter century onrdquoThe Lancet Vol 390 No 10092 pp 415-423

Dorr D Bonner LM Cohen AN Shoai RS Perrin R Chaney E and Young AS (2007)ldquoInformatics systems to promote improved care for chronic illness a literature reviewrdquo Journalof the American Medical Informatics Association Vol 13 No 2 pp 156-163

Fava GA (2017) ldquoEvidence-based medicine was bound to fail a report to Alvan Feinsteinrdquo Journal ofClinical Epidemiology Vol 84 pp 3-7

Frolich A Talavera JA Broadhead P and Dudley RA (2007) ldquoA behavioral model of clinicianresponses to incentives to improve qualityrdquo Healthy Policy Vol 80 No 1 pp 179-193

Garcia-Lizana F and Sarria-Santamera A (2007) ldquoNew technologies for chronic disease managementand control a systematic reviewrdquo Journal of Telemedicine and Telecare Vol 13 No 2 pp 62-68

Garg AX Adhikari NKJ McDonald H Rosas-Arellano MP Devereaux PJ Beyene J Sam J andHaynes RB (2005) ldquoEffects of computerized clinical decision support systems on practitionerperformance and patient outcomes a systematic reviewrdquo JAMA Vol 293 No 10 pp 1223-1238

2145


Gosden T Forland F Kristiansen IS Sutton M Leese B Giuttrida A Sergison M and Pedersen L(2001) ldquoImpact of payment method on behaviour of primary care physicians a systematic reviewrdquoJournal of Health Services Research amp Policy Vol 6 No 1 pp 44-55

Green EP (2014) ldquoPayment systems in the healthcare industry an experimental study of physicianincentivesrdquo Journal of Economic Behavior amp Organization Vol 106 pp 367-378

Greenhalgh T Howick J and Maskrey N (2014) ldquoEvidence based medicine a movement in crisisrdquoBMJ Vol 348 No g3745 pp 1-7

Grol R (2000) ldquoBetween evidence-based practice and total quality management the implementation ofcost-effective carerdquo International Journal for Quality in Health Care Vol 12 No 4 pp 297-304

Grol R and Grimshaw J (2003) ldquoFrom best evidence to best practice effective implementation ofchange in patientsrsquo carerdquo The Lancet Vol 362 pp 1225-1230

Grumbach K Osmond D Vranizan K Jaffe D and Bindman AB (1998) ldquoPrimary care physiciansrsquoexperiences of financial incentives in managed-care systemsrdquo The New England Journal ofMedicine Vol 339 No 21 pp 1516-1521

Hillman AL (1990) ldquoHealth maintenance organizations financial incentives and physiciansrsquojudgmentsrdquo Annals of Internal Medicine Vol 112 No 12 pp 891-893

Horwitz RI and Singer BH (2017) ldquoWhy evidence-based medicine failed in patient care andmedicine-based evidence will succeedrdquo Journal of Clinical Epidemiology Vol 84 pp 14-17

Horwitz RI Hayes-Conroy A Caricchio R and Singer BH (2017) ldquoFrom evidence based medicine tomedicine based evidencerdquo The American Journal of Medicine Vol 130 No 11 pp 1246-1250

Hunt DL Haynes RB Hanna SE and Smith K (1993) ldquoEffects of computer-based clinical decisionsupport systems of physician performance and patient outcomesrdquo The Journal of the AmericanMedical Association Vol 280 No 15 pp 1339-1346

Institute of Medicine (2001) Crossing the Quality Chasm A New Health System for the 21st CenturyThe National Academies Press Washington DC

Institute of Medicine (2011) Engineering a Learning Healthcare System A Look at the FutureWorkshop Summary The National Academies Press Washington DC

Institute of Medicine (2015) Integrating Research and Practice Health System Leaders Working TowardHigh-Value Care Workshop Summary The National Academies Press Washington DC

Ioannidis JP (2016) ldquoEvidence-based medicine has been hijacked a report to David SackettrdquoJournal of Clinical Epidemiology Vol 73 pp 82-86

Kaplan B (2001) ldquoEvaluating informatics applicationsmdashclinical decision support systems literaturereviewrdquo International Journal of Medical Informatics Vol 64 pp 15-37

Kitson A Harvey G and McCormack B (1998) ldquoEnabling the implementation of evidence basedpractice a conceptual frameworkrdquo Quality in Health Care Vol 7 pp 149-158

Milani RV and Lavie CJ (2015) ldquoHealth care 2020 reengineering health care delivery to combatchronic diseaserdquo The American Journal ofMedicine Vol 128 pp 337-343

OrsquoConnor AM Llewellyn-Thomas HA and Flood AB (2004) ldquoModifying unwarrantedvariations in health care shared decision making using patient decision aidsrdquo Health AffairsSupplement Web Exclusive pp VAR63-72

Owen D (2018) ldquoThe happiness buttonrdquo The New Yorker February pp 26-29

Porter ME and Kaplan RS (2016) ldquoHow to pay for health carerdquo Harvard Business ReviewJuly-August pp 88-100

Richardson WS (2017) ldquoThe practice of evidence-based medicine involves the care of whole personsrdquoJournal of Clinical Epidemiology Vol 84 pp 18-21

Robinson JC Shortell SM Li R Casalino LP and Rundall T (2004) ldquoThe alignment and blendingof payment incentives within physician organizationsrdquo Health Services Research Vol 39 No 5pp 1589-1606

2146

MD5610

Rodwin MA (2004) ldquoFinancial incentives for doctors have their place but need to be evaluated andused to promote appropriate goalsrdquo BMJ Vol 328 pp 1328-1329

Rogers EM (1995) Diffusion of Innovations 4th ed The Free Press New York NYRoshanov PS Misra S Gerstein HC Garg AX Sebaldt RJ Mackay JA Weise-Kelly L

Navarro T Wilczynski NL and Haynes RB (2011) ldquoComputerized clinical decision supportsystems for chronic disease management a decision-maker-researcher partnership systematicreviewrdquo Implementation Science Vol 6 No 92

Sackett DL Rosenberg WMC Gray JAM Haynes RB and Richardson WS (1996) ldquoEvidencebased medicine what it is and what it isnrsquotrdquo British Medical Journal Vol 312 No 7023 pp 71-72

Shojania KG and Grimshaw JM (2005) ldquoEvidence-based quality improvement the state of thesciencerdquo Health Affairs Vol 24 No 1 pp 138-150

Shortell SM Rundall TG and Hsu J (2007) ldquoImproving patient care by linking evidence-basedmedicine and evidence-based managementrdquo JAMA Vol 298 No 6 pp 673-676

Town R Kane R Johnson P and Butler M (2005) ldquoEconomic incentives and physiciansrsquo delivery ofpreventive care a systematic reviewrdquo American Journal of Preventive Medicine Vol 28 No 2pp 234-240

Wagner EH Austin BT Davis C Hindmarsh M Schaefer J and Bonomi A (2001) ldquoImprovingchronic illness care translating evidence into actionrdquo Health Affairs Vol 20 No 6 pp 64-78

Wagner EH Bennett SM Austin BT Greene SM Schaefer JK and Vonkorff M (2005) ldquoFindingcommon ground patient-centeredness and evidence-based chronic illness carerdquo The Journal ofAlternative and Complementary Medicine Vol 11 No S1 pp S-7-S-15

Wan TT (2006) ldquoHealthcare informatics research from data to evidence-based managementrdquoJournal of Medical Systems Vol 30 No 1 pp 3-7

About the authorsSaligrama Agnihothri is Professor of Operations and Business Analytics in the School of Managementat Binghamton University He holds BSc and MSc Degrees from Karnatak University Dharwad Indiaand MS and PhD Degrees from the University of Rochester His research interests include improvingefficiency and quality in healthcare operations process flexibility and cross-training decisions inservices and managing field service operations He has published in leading operations managementjournals including Operations Research Production and Operations Management IIE TransactionsNaval Research Logistics Decision Sciences and Interfaces He was Associate Editor of ManagementScience and is currently on the editorial board of Production and Operations Management journalSaligrama Agnihothri is the corresponding author and can be contacted at agnibinghamtonedu

Raghav Agnihothri CFA CMT is a former healthcare entrepreneur who is currently a PortfolioManager at a large multi-national bank in New York City He graduated with an AB in Economics fromCornell University and a MS in Finance from the University of Rochester


2147


Configurations of factors affectingtriage decision-making

A fuzzy-set qualitative comparative analysisCristina Ponsiglione and Adelaide Ippolito

Department of Industrial Engineering University of Naples Federico IINaples Italy

Simonetta PrimarioIndustrial Engineering University of Naples Federico II Naples Italy and

Giuseppe ZolloDepartment of Industrial Engineering Universita degli Studi di Napoli Federico II

Napoli Italy

AbstractPurpose ndash The purpose of this paper is to explore the configuration of factors affecting the accuracy of triagedecision-making The contribution of the work is twofold first it develops a protocol for applying a fuzzy-setqualitative comparative analysis (fsQCA) in the context of triage decision-making and second it studiesthrough two pilot cases the interplay between individual and organizational factors in determining theemergence of errors in different decisional situationsDesignmethodologyapproach ndash The methodology adopted in this paper is the qualitative comparativeanalysis (QCA) The fuzzy-set variant of QCA (fsQCA) is implemented The data set has been collected duringfield research carried out in the Emergency Departments (EDs) of two Italian public hospitalsFindings ndash The results of this study show that the interplay between individual and contextualorganizationalfactors determines the emergence of errors in triage assessment Furthermore there are some regularities in thepatterns discovered in each of the investigated organizational contexts These findings suggest that we shouldavoid isolating individual factors from the context in which nurses make their decisionsOriginalityvalue ndash Previous research on triage has mainly explored the impact of homogeneous groups offactors on the accuracy of the triage process without considering the complexity of the phenomenon underinvestigation This study outlines the need to consider the not-linear relationships among different factors inthe study of triagersquos decision-making The definition and implementation of a protocol to apply fsQCA to thetriage process in EDs further contributes to the originality of the researchKeywords Fuzzy sets Qualitative comparative analysis Heuristics Individual and organizational factorsTriage accuracy Triage decision-makingPaper type Research paper

1 IntroductionNowadays growing attention is paid to the management of Emergency Departments (EDs) asthese healthcare units are continuously affected by overcrowding This stems from ldquofeweremergency departments being available for a greater number of patients seeking carerdquo(Stanfield 2015 p 396) The triage process is the first step in the path of patients withinhospitalsrsquo EDs It consists of the assessment and subsequent prioritization of patients based onthe level of severity of their symptoms and their health conditions (Hitchcock et al 2013)The correct prioritization of patients is crucial as it has a direct impact on patientsrsquo safety andtheir flow within the healthcare facility (Cioffi 1998) Moreover the accuracy of triageassessment affects the EDrsquos level of service quality as an incorrect sorting implies prolongedwaiting-room times an increased number of patients who leave without being seenand decreased patient satisfaction (Derlet and Richards 2000) Furthermore the accuracy ofassessment is often related to the effectiveness of the triage process (Marsden 2000Frykberg 2005) To accurately prioritize patients in a time when available resources are limited

Management DecisionVol 56 No 10 2018pp 2148-2171copy Emerald Publishing Limited0025-1747DOI 101108MD-10-2017-0999

Received 15 October 2017Revised 7 March 201821 May 201811 July 2018Accepted 16 July 2018


2148

MD5610


means in fact ldquoto provide care to those who seek itrdquo (Stanfield 2015 p 396) These elementsjustify the increasing attention paid by the literature on healthcare and emergency management(McMillan et al 1986 Chung 2005 Andersson et al 2006 Noon 2014 Vatnoslashy et al 2013Hitchcock et al 2013 Martin et al 2014) to the triage process

The decision-making process is the foundation of triage practice (Chung 2005Noon 2014) It is frequently described as a dynamic complex process (Cioffi 2001Goumlransson et al 2008 Noon 2014) that occurs mostly under conditions of uncertainty(Cioffi 1998 2001) and time pressure (Chung 2005 Wolf 2010) Because of thesecharacteristics some scholars (Cioffi and Markham 1997 Cioffi 1998) have classifieddecision-making in triage assessment as a heuristic process Tversky and Kahneman (1974)the pioneers of the Heuristics and Biases Program introduced the term ldquoheuristicsrdquo whichrefers to mental strategies that prevail over the laws of logic and rational choice Usingheuristics the decision-maker determines systematic deviations from optimal decisionscalled ldquobiasesrdquo cognitive illusions or ldquoirrationalityrdquo (Kahneman and Tversky 1977 1981)The Heuristics and Biases Program assumes that heuristics are ldquomental shortcomingsrdquo(Artinger et al 2015) that always lead to the second-best solution (Kahneman 2011) Thisapproach has been strongly criticized by Gigerenzer and his research group who proposedthe ldquofast and frugalrdquo (Gigerenzer et al 1999) view of heuristics They argued that heuristicscould lead to accurate and fast judgment in complex situations because they focus on alimited number of critical variables as happens in human reasoning (Gigerenzer 1996 Luanet al 2011 Meissner and Wulf 2017) Heuristics are ldquofast and frugalrdquo as the judgment isbased on few cues and is made in a short time (Martignon and Hoffrage 2002 Kuncel et al2011 Drechsler et al 2014) Central in this view of heuristics is the interplay between theenvironmentrsquos structure and the mental model of the decision-maker ldquoHeuristics allow foradaptive responses to the characteristics of an uncertain managerial environmentrdquo (Artingeret al 2015 p 833) The success of a heuristic is determined by its ldquoecological rationalityrdquonamely its match with a specific environmentrsquos structure (Gigerenzer et al 1999) Ecologicalrationality refers to how a bounded mind ldquoexploits the structure of the social and physicalenvironments in which it must reach its goalsrdquo (Chase et al 1998 p 212)

The crucial points of the ldquofast and frugalrdquo approach to heuristics from the perspective ofecological rationality can be also found in triage decision-making and can be summarizedas follows

The individual under conditions of uncertainty and limited cognitive and time resourcesfocuses only on a portion of the available information The decision can nevertheless beaccurate (Gigerenzer and Kurzenhaumluser 2005)

The structure of the information characterizing the decisional situation (task complexityuncertainty ambiguity) influences the judgment process and its accuracy (Cioffi 1998)

The match between the individual experience and beliefs the social-organizationalcontext in which the decision takes place and the nature of the decisional task are decisive indetermining the accuracy of the decisionrsquos outcome (Smith et al 2008)

The assumption of this research thus departs from adopting the ecological rationalityperspective to frame the decision-making process in triage as a dynamic complexprocess in which factors related to the individualrsquos biography (eg education trainingprevious work experience) interact with environmental factors (including social-organizational and situational factors) in producing a specific answer to a specific task(Todd and Gigerenzer 2012)

The literature on clinical and triage decision-making has extensively examined thesegroups of factors (Stanfield 2015) separately or via an additive approach The contributionof our work consists of the development of a methodological approach to analyze from anon-linear perspective the effect that combinations of individual and organizational factorshave on the accuracy of triage assessment taking into account the complex nature of the

2149

A fuzzy-setqualitative

comparativeanalysis

decision-making process and the different levels of uncertainty of situations in which thedecision has to be made

We explore different combinations of factors in terms of their causal link with the level oferrors made by triage nurses This can provide interesting insights into the identification ofconfigurations of levers to foster the accuracy and the quality of the triage process

The paper is structured as follows the next section presents a literature review ofsuggested relevant factors in terms of their impact on triage nursesrsquo decisions Section 3illustrates the main pillars of the adopted methodology namely the fuzzy-set qualitativecomparative analysis ( fsQCA) describes the steps of its implementation and the datacollection and elaboration phases Section 4 reports on the results while Section 5 discussesthem Section 6 addresses the implications of our findings for theory and practice

2 Factors affecting decision-making in the triage processBeginning in the end of the 1990s several studies have been published mainly in the field ofclinical decision-making and emergency nursing (Cioffi 1998 Cabana et al 1999 Croskerryand Sinclair 2001 Cone and Murray 2002 Chung 2005 Andersson et al 2006 Smith et al2008 Garbez et al 2011 Wolf 2010 2013 Martin et al 2014 Stanfield 2015) that analyzethe decision-making process in the practice of triage These studies adopt differenttheoretical approaches and research methods (qualitative or quantitative) and considerdifferent outcomes of the decision-making process In most cases the accuracy of theassignment of triage scores to patients is examined as the outcome (Cioffi 1998 Cooperet al 2002 Garbez et al 2011 Martin et al 2014) Gerdtz and Bucknall (2001) consider theduration of the triage process as the main outcome to be studied There are alsocontributions (usually exploratory qualitative studies) that focus on the description of thetriage assessment process or on the elements considered to make decisions (Chung 2005Andersson et al 2006 Smith et al 2008)

One of the aspects taken into consideration in studies dealing with theaccuracyvulnerability of the triage process is related to the complexity of the situationthat the operator must evaluate (Cioffi 1998 Chung 2005 Cioffi 2001) A shared definitionof ldquocomplexityrdquo is not traceable in this context mainly because some studies mentionthe complexity of the task as an element that can influence the decision but do notoperationalize this concept Empirical works using a taskrsquos complexity as a variable in theanalysis of the triage process classify real decisional situations on the basis of twodimensions (Cosier and Dalton 1988) the uncertainty of the situation and the availability ofrelevant information Situations with the lowest complexity are those in which the level ofuncertainty is limited and relevant information needed to make decisions is accessibleThe most complex situations are those with a high level of uncertainty (limited possibility topredict the value of the decisional variables) and little relevant information available

The use of objective parameters is one of the most-cited factors in the literature on thetriage process (Salk et al 1998 Gerdtz and Bucknall 2001 Wolf 2010 Vatnoslashy et al 2013)Objective parameters are vital signs that can be measured through different typologies ofdiagnostic tests There is evidence that referring to objective parameters slows down thedecision-making process and lengthens the time that the assessment takes (Gerdtz andBucknall 2001 Storm-Versloot et al 2014) The literature does not agree on the effect thatthe use of objective parameters has on the accuracy of scoring (Conen et al 2006) On theone hand vital signs can reveal possible changes in health conditions improving theaccuracy of triage assessment (Burchill and Polomano 2016) On the other hand decisionsbased mainly on vital signs can lead to nursesrsquo under- or over-assessing the assignedpriority code (Nakagawa et al 2003) In a study conducted by Vatnoslashy et al (2013) it ispointed out that the general tendency of triage operators is to neglect the use of vitalparameters This study also shows that the implementation of protocols and guidelines

2150

MD5610

fosters a reference to objective parameters Furthermore as the use of objective parametersincreases the number of patients classified at the highest levels of urgency decreasesVatnoslashy et al (2013) claim however that the effect of the use of vital signs on the accuracy ofthe assessment and on patientsrsquo safety is not clear Cooper et al (2002) state that ldquovisual cues(non-verbal communication) physical findings (limited physical examination) and vitalsigns all inform the decision-making process Each component likely plays an importantpart in accurate triage with the relative importance of each element varying on acase-by-case basisrdquo (Cooper et al 2002 p 231) Most experienced nurses tend to under-utilizeobjective parameters (Chung 2005) On the other hand the implementation of specificprotocols and guidelines in the ED can lead to an increase in their usage (Vatnoslashy et al 2013)

The role of visual cues protocols and guidelines in determining the decision of triagenurses is also studied (Salk et al 1998 Cone and Murray 2002 Cooper et al 2002 Chung2005) Salk et al (1998) look at the same group of nurses assigning a priority code tothe same group of patients in a two-stage triage in which the first stage consists of atelephone triage and the second of a face-to-face triage The use of formal protocols andobjective parameters does not determine an alignment between the scores of the operators inthe two phases This leads the authors to conclude that visual cues become decisive inin-person triage Guidelines and assignment criteria seem to represent a reference forthe decision especially for beginners but their presence is not considered decisive in thedecision-making process (Salk et al 1998) In particular expert nurses perceive the presenceof guidelines pre-established criteria and protocols negatively (Cone and Murray 2002)

Experience is one of the factors frequently analyzed in theoretical-qualitative studies andin those with a strong empirical and quantitative nature as a fundamental variableinfluencing the triage decision-making process and its outcomes Experience is usuallyframed as the frequency of nursesrsquo exposure to different emergency problems (Cioffi 1998)The most widespread measures of the specific experience and skills of nurses are thenumber of working years in EDs and those accumulated as a triage operator (Cioffi 1998Cone and Murray 2002 Andersson et al 2006 Martin et al 2014 Hitchcock et al 2013)Referring to all the activities performed in EDs Croskerry and Sinclair (2001 p 273) claimthat ldquothe level of experience of physicians and nurses is intrinsically linked to preventabilityof errorrdquo Hitchcock et al (2013) outline that nurses perceive the level of experience as havingan impact on the outcomes of the process and on the professional relationships among staffmembers Cone and Murray (2002 p 203) identify experience as ldquoan important characteristicthat included intuition confidence in judgment and trust in or reliance on peersrdquoFurthermore experience in EDs and in triage activities is considered as the primary factorfor performing safely in emergency situations Martin et al (2014) examine whetherexperience and attitude toward patients are discriminatory when determining accurateassignments of priority codes by nurses in triage This descriptive study concludes thatldquofindings did not achieve statistical significance to support the notion that attitude orspecified amount of experience contributed to accurate ESI score assignmentrdquo (Martin et al2014 p 467) Cioffi (1998) analyzes the role of nursesrsquo experience in the mechanisms used tomake triage assessment under conditions of uncertainty First the results of this work showa variation in the acuity levels assigned by more and less experienced nurses Second theperception of assigned acuity levelsrsquo accuracy is higher in more experienced nurses than inless experienced ones This is consistent with other research that relates self-confidence andtrust in onersquos intuitions courage and the ability to master stress to nursesrsquo work experience(Cone and Murray 2002 Andersson et al 2006) Additionally more experienced nursesusually collect less data when they assess triage cases and use more heuristics particularlyin situations of high uncertainty

The personal experience of nurses is often characterized as an individual factor inconnection with other elements such as intuition confidence in onersquos own assessments

2151


comparativeanalysis

motivation listening and communication skills and relationships with colleagues andpatients (Andersson et al 2006 Martin et al 2014) In other cases experience is related tothe level of knowledge acquired through education formal training and technical know-howin different disciplines (Cone and Murray 2002 Hitchcock et al 2013) The ldquoknowledgerdquovariable is a multidimensional concept In some cases the level of knowledge is framed interms of education and training (Chung 2005 Andersson et al 2006) in other casesknowledge is related to broad technical know-how and a diversified knowledge base(Cone and Murray 2002 Hitchcock et al 2013) Training activities are considered relevantfor reducing triage mistakes (Lampi et al 2017) Training is also related to the capability ofnurses to make decisions coherently with the guidelines of technical triage manuals(Arslanian-Engoren 2005)

The literature also points to several factors related to the social context and nursesrsquo workenvironments which affect the process and potential outcomes of triage (Croskerry andSinclair 2001 Wolf 2010 Hitchcock et al 2013 Wolf 2013) Some of these factors refer to theculture and tacit rules in a given context internalized through experience in the specific workenvironment and able to affect the perceptions and motivations of nurses For example asWolf (2010) suggests the culture developed in a context as well as the perception thatoperators have of their leaders and the level of collaboration and communication with patientsand among peers can determine the type of information and objective data that nurses takeinto consideration when assessing priority levels This also affects their perception of theusefulness of protocols and guidelines Hitchcock et al (2013) argue that nurses perceivecommunication collaboration and the intensity of teamwork as essential to reducing loss ofinformation and ensuring the quality of triage assessment Croskerry and Sinclair (2001) claimthat a lack of feedback by supervisors could compromise the maintenance of ED nursesrsquocognitive and procedural skills Wood and Bandura (1989) point out that judgment indecision-making is influenced by motivational mechanisms If operators have a goodperception of the effectiveness of procedures protocols and guidelines (Greenwood et al 2000Smith et al 2008) they might not feel isolated in their professional responsibility(Adriaenssens et al 2011 Melby et al 2011 Vatnoslashy et al 2013)

Finally the literature highlights the potential negative effect of nursesrsquo workload andcontinuous interruptions of their assessment job (Chung 2005 Andersson et al 2006)The EDrsquos overcrowding and patient volume (Hitchcock et al 2013 Wolf 2013) couldsignificantly affect the level of stress experienced by triage nurses and consequently theaccuracy of priority levelsrsquo assignment

All the factors discussed above are summarized in Table I the table characterizes factorsas mainly individual or related to the work environment (organizational or contextualfactors) and reports more relevant literature findings about their influence on the triageassessment process

The studies examined in this short literature review have different objectives andapproaches Some of them are qualitative and aim at highlighting the issues that nursesperceive as important in the triage decision-making process (eg Andersson et al 2006Hitchcock et al 2013) others are quantitative and generally study the impact ofhomogeneous groups of factors on triage outcomes (timing and accuracy of theassignments) with a typically additive approach (descriptive or inferential statistics)(eg Gerdtz and Bucknall 2001 Martin et al 2014)

Wolf (2010 p 245) concluding her ethnographic exploration of the clinical decision-making of emergency nurses claims that the process of acuity assignation observed in herstudy ldquoseems to be the result of an interplay of elements particular to the individual nursethe immediate environment of the unit and the general environment of carerdquo

Furthermore Todd and Gigerenzer (2012) describing the perspective of ldquoecologicalrationalityrdquo on the heuristic decision-making process declare ldquoOur intelligent adaptive

2152

MD5610

Factors ReferencesIndividualorganizationaland contextual Main themes and findings

Use of objectiveparameters

Gerdtz and Bucknall(2001) Nakagawaet al (2003) Chunget al (2005) Vatnoslashyet al (2013) Storm-Versloot et al (2014)

Individual affected bythe implementation ofspecific protocols andguidelines and byorganizational informalshared rules

Objective parameters are usuallyunder-utilized by nurses in particularby expert nurses It is not establishedhow the use of objective parameterscould impact on the accuracy ofTriages assessment Theimplementation of guidelines andprotocols increases the use of objectiveparameters among Triages nurses

Use of visualcues

Salk et al (1998) Individual dependentalso on the complexity ofthe task to be assessedand by organizationalinformal shared rules

Visual cues are fundamental sources ofinformation for nurses in in-persontriage

Use of formalproceduresguidelinesmanuals andprotocolscriteria

Salk et al (1998) Coneand Murray (2002)Adrianenssens et al(2011)

Organizational but alsoaffected by individualattributes

Formal procedures and guidelinesrepresent a reference for young nursesand make them comfortable and safewhen making decisions Pre-established criteria and formalguidelines are perceived as detrimentalby expert nurses

Experience Cioffi (1998) Cone andMurray (2002)Andersson et al(2006) Martin et al(2014) Hitchcock et al(2013)

Individual The experience affects negatively theuse of objective parameters and formalguidelines in making decision Highlevel of experience impact positively onthe intensity of teamwork on themotivation and on communication withpeers and physicians Moreexperienced nurses use extensively theheuristics in their judgment It is notstatistically proven that greaterexperience means better accuracy

Knowledgetraining andeducation

Cone and Murray(2002) Chung (2005)Andersson et al(2006) Hitchcock et al(2013)

Individual dependent insome cases byorganizationalprocedures

A broad technical know-how acquiredthrough advise by supervisors in otherdisciplines or by training could bebeneficial for the self-confidence ofnurses and consequently for theaccuracy of acuity levels assignmentKnowledge also contributes to effectivecommunication with peers and patients

Personal traitsand attitudes

Andersson et al(2006) Martin et al(2014)

Individual it is not clearly assessed the directimpact of attitude toward patientscourage intuition and motivation onthe accuracy of the assessment Allthese factors are reported as related tothe experience of nurses and areclassified as personal traits that cancontribute to the work environmentrsquosclimate

Communicationfeedback unitsleadership andteamwork

Croskerry andSinclair (2001) Wolf(2010) Hitchcock et al(2013) Wolf (2013)


All these factors can contribute toTriagersquos assessment accuracy becausereduce the loss of information inemergency situations help in

(continued )

Table IFactors affecting

triage process

2153


comparativeanalysis

behavior emerges from the interaction of both mind and wordrdquo (Todd and Gigerenzer 2012p 4) The ldquowordrdquo is defined as the ldquostructure of the environmentrdquo in which and upon whichthe individual acts ldquoThe environment also influences the agentrsquos actions in multiple waysby determining the goals that the agent aims to fulfill shaping the tools that the agenthas for reaching those goals and providing the input processed by the agent to guide itsdecisions and behaviorrdquo (Todd and Gigerenzer 2012 p 16) The input to be processed andthe weight assigned to it in the decision thus become part of the environment and areeventually filtered and interpreted according to individual and social-organizational frames

The issue addressed in the present paper departs from the premise highlighted byWolf (2010 2013) and it is analyzed in accordance with the theoretical perspective ofecological rationality (Gigerenzer et al 1999)

The research question we aim to answer with this research is

RQ1 What configurations of factors affect the accuracy of the decision-making processof triage nurses in assigning priority codes

In answering to this question we assume the complexity of the phenomenon underinvestigation and of the information structure of the decisional task (as suggested by theview of ldquoecological rationalityrdquo) The perspective of complexity implies the need to considerthat non-linear relationships of different factors play a role in the decisional processes oftriage nurses The methodological approach of qualitative comparative analysis (QCA)seems to be well suited to this aim To the best of our knowledge the QCA approach has notpreviously been used to study the effects of different factors on the accuracy of triageassessment The present study moreover aims at integrating the repertoire of qualitativemethodologies used in the analysis of clinical decision-making for this reason the test andcalibration of the methodological approach via two pilot cases constitutes a relevantobjective of the work

3 Method and dataThe QCA is a relatively new approach in the social sciences (Fiss 2009 Marx et al 2013Ragin 1987 Ragin 2000 Ragin 2008) that is receiving increasing attention in managerialstudies as demonstrated by the number of papers using this method that are published inhigh-quality journals (see eg Dy et al 2005 Fiss 2009 Greckhamer et al 2013 Ordaniniet al 2014)

QCA is a comparative case-oriented (Marx et al 2013) methodology based on theprinciples of Boolean algebra and set-theoretic analysis (Ragin 2008) The method movesfrom an in-depth knowledge and analysis of a small to intermediate number of empiricalcases (eg between 5 and 50) toward the identification of configurations of causally relevantconditions linked to the outcome under investigation (Marx et al 2013)

QCA is case-oriented The consequence of this view is that the effects of variables areassessed in the context of investigated cases and not in isolation cases are framed as


managing the stress and foster thelearning process of nurses

Overcrowdingworkloadinterruptions

Chung (2005)Andersson et al(2006) Hitchcock et al(2013) Wolf (2013)

Organizational-contextual

All these factors affect negatively theaccuracy of Triagersquos assessmentbecause increases the level of stress inthe work environment and eventuallyproduces loss of informationTable I

2154

MD5610

configurations of relevant causal conditions Furthermore the method is comparative as itdevelops through comparisons of cases to find cross-case similarities or differences ThusQCA allows researchers to continuously integrate within-cases with cross-cases analysis(Marx et al 2013) As outlined by Ragin who launched this methodology and its analyticaltools QCA ldquointegrates the best features of the case-oriented approach with the best featuresof the variable-oriented approachrdquo (Ragin 1987 p 84)

QCA is in fact a set-theoretic analytical approach in the sense that it identifies causalpatterns in a phenomenon under investigation by focusing on sets and subsetsrelationships The use of set-theoretic principles originates in the awareness that ldquoalmost allsocial science theory is verbal and as such is formulated in terms of sets and set relationsrdquo(Ragin 2008 p 13)

The use of set relations and Boolean algebra to identify and analyze causal patterns thatlead to a specific outcome strongly distinguishes QCA from traditional variable-orientedmethodologies In the latter the verbal relations between sets typically formulated insocial-science theories are translated into hypotheses of correlations among variables and thenstudied through correlation techniques (Ragin 2008) In this kind of approach variables ldquoaimto capture a dimension of variation across cases and distribute cases on this variationrdquo (Rihouxand Marx 2013 p 168) In QCA a symmetric relationship is divided into two asymmetricanalyses formalized by set and sub-set relationships one of the necessity of the conditionswith respect to the outcome and the other of their sufficiency This allows researchers to dealwith the complexity of real phenomena without any a priori simplifications QCA in factassumes the non-linearity of phenomena under investigation and is based on the principle ofcausal complexity This means that in most cases it does not make sense to isolate the effectof a single independent variable on the outcome but configurations of variables are identifiedthat are related to the outcome Moreover several different configurations can be recognized asldquocausal recipesrdquo for the same outcome (Ragin 1987)

This is one of the advantages in most social sciences of using QCA Its level ofanalytical formalization leads to other advantageous features it is possible to conductcomparative assessments of intermediate samples of cases that are too big for traditionalqualitative approaches and too small for correlation analyses and the use of Booleanalgebra and set operations enables the replication of research conducted through QCA(Rihoux and Marx 2013)

31 The implementation of fsQCAThe QCA research approach has been divided into three different versions based on analyticaland software tools (Ragin 2000 Rihoux 2006 Cronqvist 2005) the crisp set (csQCA) versionthe version based on fuzzy sets ( fsQCA) and the multi-value version (mvQCA)

In this study the fuzzy-set-based variant is used to consider the granularity ofinformation and data collected during the fieldwork The possibility to use both fuzzyvariables and crisp variables is another reason that makes this method well suited for thecontext of this study

The steps suggested to implement the fsQCA are the followingIdentification of relevant empirical cases causal conditions and outcomeBuilding a raw data table Generally this table has as many rows as there are cases Single

causal conditions and the outcome are listed in the columns and cells of the matrix representthe values of indicators through which the causal conditions have been operationalized

The raw-data table undergoes a dichotomization process in the crisp variant usingthresholds defined by the researcher based on herhis in-depth theoretical and empiricalknowledge (Rihoux and DeMeur 2008) In the fuzzy variant a calibration process of fuzzy setsrepresenting the causal conditions and the outcome is needed which again strictly depends onthe relevant theoretical and empirical knowledge of the researchers involved (Ragin 2000)

2155


comparativeanalysis

Building a truth-table The truth-table groups empirical cases based on the fact that theyshow the presence or absence of the outcome In the csQCA variant the truth-table shows asmany rows as there are combinations of causal conditions (2k rows where k is the number ofcausal conditions) and each case is assigned to a unique row The values in the cells aredichotomous values (0 1) In the fsQCA version building a crisp truth-table does notproceed automatically but requires intermediate steps In fact when conditions andoutcomes are fuzzy sets each case can have a unique combination of membership scoresassigned to the causal conditions and the outcome Ragin (2008) shows however that thereis a correspondence between the rows of the crisp truth-table and the 2k corners of themulti-dimensional space made by the fuzzy sets

The analysis of the truth-table allows researchers to identify explicit connectionsbetween configurations of causal conditions and the outcome A causal condition isnecessary for an outcome if instances of the outcome constitute a subset of the instances ofthe causal condition A condition is sufficient if the instances of the causal conditionconstitute a subset of the outcome When fuzzy sets are used the assessment of sufficiencyis not trivial The solution can be found by applying the logic of fuzzy-sets theory and theoperations on fuzzy sets

To assess the level of fitness of subset relations two parameters of fit (Legewie 2013) areused consistency and coverage They serve to assess the degree of approximation ofidentified set-theoretic relations in empirical cases Consistency measures the degree towhich a subset relation between a casual condition and an outcome is ldquometrdquo in real data(Legewie 2013) Consistency ranges from 0 to 1 with 1 indicating perfect consistency

Once the consistency of a subset relation has been assessed coverage measures itsempirical relevance (Legewie 2013) Coverage also ranges from 0 to 1 As Ragin (2006)outlines consistency and coverage of a subset relation are contrasting measures in manyresearch contexts and a trade-off between the two has to be found according to the specificobject of investigation and taking into account the number of causal conditions andavailable cases According to Raginrsquos (2006 2008) suggestions in this study the minimumacceptable level of consistency is used to assess the empirical relevance of sufficient sub-setrelations (Fiss 2011) that is 075

The last step of the QCA procedure is the identification and interpretation of consistentand empirically relevant patterns (causal configurations of conditions) pertaining to theoutcome The analysis of the truth-table is usually employed to identify sufficientcombinations of conditions for the outcome to occur The identification of necessaryconditions is an intermediate step implemented to simplify the truth-table (Fiss 2009) Thereare three types of solutions that the truth-table analysis provides A complex solution doesnot allow for any simplifying assumptions and displays all logically true combinations offactors sufficient for an outcome to occur (Legewie 2013) A parsimonious solution insteadis obtained automatically by applying the process of Boolean minimization and allsimplifying assumptions to the truth-table without applying any specific knowledge ofthe cases under investigation Finally intermediate solutions are obtained by allowing forsome simplifications and including the researcherrsquos previous empirical and theoreticalknowledge in the analysis of the truth-table (Fiss 2011)

Most of the steps described above are taken with the help of software specificallydeveloped in the context of QCA research In this study the package fsQCA 30 is adoptedThe next section illustrates how the protocol of fsQCA has been implemented in the presentresearch project

32 The application of the fsQCA protocol field research and dataField research has been conducted in the EDs of two Italian public hospitals named Alphaand Beta because of privacy concerns in the period JanuaryndashApril 2016 The two hospitals

2156

MD5610

are in the same city but they serve two different populations and significantly differ interms of the emergency activitiesrsquo organization Alpha serves mainly a city population Betaserves a very large user base which extends beyond the cityrsquos boundaries across the region

The ED of Alpha is classified as a level I Emergency and Acceptance Department(DEA I) According to the Italian classification of EDs a DEA I ensures additional servicessuch as patientsrsquo observation and short stay Alpha implemented the triage system in 2008

The ED of Beta is classified as DEA II In addition to the services provided by typicalfirst-level DEAs it ensures the highest-qualifying features related to emergency careincluding neurosurgery cardiac surgery neonatal intensive care thoracic surgery andvascular surgery It introduced the triage system in 2006

In Italy triage coding is mostly done on a color-code scale basis with highest prioritygiven to a red code followed by yellow green and white

Alpha and Betarsquos EDs exhibit two different organizational models with respect to theprioritization of patients In Alpha the whole process is performed in a linear way withoutinterruptions the nurse assigned to triage takes care of the patient from herhis entry intothe structure until shehe is called for a medical examination (global triage) In Beta theprocess is divided into two phases (two-steps triage) each taken charge of by a differentnurse In the first step the patient is identified and registered a first evaluation of theexpressed symptoms is performed and a temporary codification is assigned by one triagenurse in the next step a different triage nurse reassesses the patient and the color-code isdefinitively assigned confirming or not confirming the one previously attributed

During the research period Alpharsquos ED employed 31 nurses of whom 19 were regularlyassigned to triage activities Betarsquos ED accounted for 59 nurses 35 of whom were regularlyinvolved in the two steps of triage In Alpha triage nurses are those with an adequate basiccertification for the execution of the planned activities who regularly attend specific trainingcourses In Beta nurses working in triage are not regularly trained and in most cases havenot attended specific triage courses Furthermore in Alpharsquos ED there are specific protocolsand guidelines available to triage operators the same does not apply for Betarsquos ED The maincharacteristics of Alpha and Betarsquos emergency services are summarized in Table II

Table III reports on the number of training courses (basic and specialized training ontriage) attended by the triage nurses of Alpha and Betarsquos EDs during their working life

Number of attended courses Alpha () Beta ()

⩽2 16 55⩾3 and ⩽4 68 37W4 16 9

Table IIIPercentage of

attended courses bytriagersquos nurses

Alpha ED Beta ED

Number of accesses in 2015 52922 90566Triage model Global Two-stepsTriage shifts 3 shifts

(800 -1400 1400-2000 2000-800)3 shifts(800 ndash1400 1400-2000 2000ndash800)

Number of triage nurses per shift 2shift 2shift (I step)3shift (II step)

Re-evaluation of waiting patients Yes YesSpecific protocols and guidelinesfor triage

Yes No

Table IIMain characteristics ofemergency services in

Alpha and Beta

2157


comparativeanalysis

Figure 1 shows the distribution of triage nursesrsquo experience levels in the health sector EDsand the specific ED under investigation for Alpha and Betarsquos nurses Furthermore theaverage three levels of experience of the two samples are compared (right side of the figure)

The steps involved in implementing the fsQCA described in section 31 have beenintegrated in the field research

The first step was conducted as desk research It was the identification of the outcome(the dependent variable) and the causal conditions to be studied (the factors assumed tohave an impact on the outcome) In our study the accuracy of assigned priority codesrepresents the outcome of interest The accuracy is operationalized in terms of the level oferrors made by nurses and is measured as the ratio between the number of errors in theassignment of priority codes and the number of assessed cases by the same nurse

Most of the studies on factors affecting the effectiveness and quality of nursesrsquo decision-making processes in emergency situations refer to the accuracy of triage decisions and therelated error level in the assessment of priority codes as outcome variable (Croskerry andSinclair 2001 Martin et al 2014 Wolf 2013)

Causal conditions are factors assumed to have an impact on the chosen outcome Theselection of input variables for the research model was made according to the following criteria

bull variables related to different levels of analysis (individual and organizational)were chosen

bull context variables (workload interruptions overcrowding) were excluded because thecollection of data was executed in a controlled environment (like a laboratoryexperiment) through a simulative approach and

725

20

AverageYHS

AverageYED

AverageYTED

ALPHA

BETA

15

10

5

0

6

5

4

No

of N

urse

s

3

2

10 5 10 15 20

YearNotes Levels of experience in the health sector (YHS darker shade of color) levels ofexperience in emergency departments (YED intermediate shade of color) levels ofexperience in the specific emergency department (YTED lighter shade of color)

25 30 35 40 45

Figure 1Experiencersquos levelsdistribution

2158

MD5610

bull other variables especially those related to personal attitudes (courage attitudestoward patients) or to the work environment (perception of the unitrsquos leadership)have not been considered due to the unavailability of nurses to disclose information

Table IV presents the causal conditions and the outcome specifying for each variable theabbreviation and a crisp or fuzzy classification The choice of calibrating the value of avariable as crisp or fuzzy was based on the typology of the measures adopted and on thelevel of availability and granularity of information collected in the field Furthermorevariables representing causal conditions have been classified according to the literaturediscussed in Section 2 and consistently with the ecological rationality perspective asindividual-related or organization-related factors

The use of objective parameters (PO) refers to the tendency of nurses to consider vitalsigns when choosing priority levels It is considered an individual factor because it isdependent on a specific choice of individual nurses and is often related to their level ofexperience (Chung 2005) The years of experience in the health sector (YHS) is included inthe study as a proxy for the ldquoknowledge-baserdquo of nurses together with the number ofattended training courses (CT) Moreover these variables are classified as individualfactors since they can identify different experiences in terms of the education and trainingof nurses

The years of experience in EDs (YED) are used as a measure of individual nursesrsquoexperience and expertise as suggested by the literature analyzed in Section 2

The years of experience in the specific ED under analysis (YTED) is included in thisstudy as a proxy for the nursesrsquo internalization level of organizational formal and informalrules and of socially constructed norms In this sense this variable is classified as anorganization-related factor The perception of the reliability of work procedures andprotocols involved in the general triage methodology (PTM) is used as a measure of nursesrsquoattitude toward the use of formal guidelines and criteria established by the Health MinistryIt is considered an individual factor since it is assumed to be related to individual choicesand beliefs as in the case of objective parameters The perception of how the triagemethodology is adopted in the specific organization (PED) is related to the availability anduse of specific formal or informal shared rules in the organizational context of the ED underinvestigation Using this perspective this variable is classified as an organizational factor

In order to collect the data to be calibrated and used in the fsQCA 25 patient scenarioswere built and administered to triage nurses Each case simulates a situation in which thepatient arrives to the ED The simulation of clinical scenarios for data gathering is one of themethods used in triage research (Van der Wulp et al 2008 Gerdtz and Bucknall 2007)particularly in qualitative and exploratory research

An expert nurse (a trainer in the triage process) assisted in building patient scenariosThe expert having obtained specific work experience in triage activities acted as a trainer

Variable Acronym Individualorganizational Calibration

Use of vital signs and objective parameters PO Individual CrispExperience in the health sector YHS Individual FuzzyExperience in an emergency department YED Individual FuzzyExperience in this emergency department YTED Organizational FuzzyGood perception about triage methodology PTM Individual CrispGood perception about triage methodology asit is applied in this ED

PED Organizational Crisp

Number of attended training courses CT Individual FuzzyErrorsrsquo ratio OUTCOME na Fuzzy

Table IVVariables in

fsQCA analysis

2159


comparativeanalysis

of nurses in different hospitals in the region During the period in which the research wascarried out he was an independent trainer and did not belong to one of the two hospitalsunder investigation He elaborated patient scenarios according to his work experience andalso relied on his knowledge of real and most frequent triage situations which were tested inthe two EDs

For each scenario the triage trainer identified the right priority code to be assignedaccording to general triage protocols and guidelines Furthermore he elicited the key cuesthat were useful for making correct decisions Other cues reported in the scenariosrsquodescriptions were considered not necessary for providing the correct answer To ensure thereliability of patientsrsquo scenarios and the priority codes assigned by the expert scenarioswere analyzed by another trainer operating in a completely different context (Spain)He analyzed the scenarios and assigned them scores Despite small differences in prioritycodesrsquo scales in Italy and Spain the two experts made comparable assessments and definedthe same ranking for the patientsrsquo scenarios

We grouped these 25 scenarios into three classes based on their level of ldquocomplexityrdquofollowing the classification of clinical situations proposed by Cioffi (1998 2001) based onCosier and Daltonrsquos (1988) simple cases (the additional cues are compatible with the key cuerelevant information is available and the prediction of decision variables is possible)intermediate cases (the additional cues are not compatible with the key cue and the relevantinformation is not always available) complex cases (cues are contradictory and somerelevant information is lacking) Table V presents the distribution of clinical scenarios interms of their level of complexity and right color codes

Nurses involved in the field study numbered 19 for Alpha and 35 for Beta Thus all thetriage nurses of the two EDs participated in the study A simulation of prioritization wasmade allowing nurses to evaluate in a very short time (less than five minutes) the informationreported in each case and to assign a priority code (nurses of Beta were invited not to refer toa specific step of the ldquotwo-stepsrdquo procedure) After that using a semi-structured interviewnurses were asked to justify their decision explain the rationale of their choices according toindividual and organizational variables selected for the study and identify the informationselected for making the decision Additional information related to their previous experienceseducation and perception of the working context was collected

The simulation phase took place for each nurse separately when shehe was notinvolved in herhis work shift Nurses were not informed about the different levels ofcomplexity of patient scenarios presented to them This choice resembled situations usuallyexperienced by them in real cases

Raw-data tables (one for Alpharsquos nurses and one for Betarsquos nurses) include for eachoperator and each of the simulated clinical scenarios the values of the indicators used tomeasure causal conditions and the outcome Table VI shows the variables and the typologyof measures obtained through the interviews

The calibration of fuzzy sets was executed automatically by the software R based ondata and using qualitative anchor points provided by the investigators

The elaboration and analysis of truth-tables instead were performed through the fsQCA30 package

White priority code Green priority code Yellow priority code Red priority code Total

Simple 4 6 0 3 13Intermediate 0 0 6 0 6Complex 1 3 2 0 6Total 5 9 8 3 25

Table VClassification of caseswith respect to theirlevel of complexityand to theircolor codes

2160

MD5610

4 ResultsResults of the application of fsQCA are reported with reference to the two analyzed samples(Alpharsquos nurses and Betarsquos nurses) and to the three categories of clinical scenarios underanalysis simple intermediate and complex (Table VII) Complex solutions have beenchosen for the analysis of truth-tables as the present research is exploratory its aim is theidentification of all consistent andor empirically relevant combinations of factors leading tothe outcome to be further investigated or simplified through additional case studies (otherEDs) The analysisrsquo focus on sufficient configurations follows the assumption that thetriage-decisional process is complex and diverse combinations of causal conditions can belinked to the occurrence of the same outcome

As shown in Table VII none of the emergent configurations for Alpharsquos sample passedthe consistency test (threshold 075) in the case of simple scenarios This result is probablydue to the fact that in simple cases the coherence between the cues determines a lower levelof errors than in intermediate and complex ones

This means that it is difficult to find cases in which the subset relation between causalconfigurations and the outcome (presence of a certain level of errors) is verified Despite thisfact there is almost one solution related to Alpharsquos sample that is close to the consistencythreshold and that also exhibits a balance between consistency and row coverage

The third solutionrsquos row (POYHS simYTEDPEDPTMCT) presents a consistency ofabout 0725 and a row coverage of 04 This sufficient configuration shows that the recurrentuse of object parameters as vital signs (PO) long experience in the health sector (YHS) alack of specific experience in the ED under investigation (simYTED) combined with a goodperception of the reliability of the triage methodology (PTM) and of its implementation(PED) and with a high level of training on triage (CT) together lead to the occurrence oferrors in the assessment of priority codes by Alpharsquos triage nurses in simple scenariosIt seems that the reliance on vital signs and the good level of knowledge of nurses acquiredthrough both work experience in the health sector and training courses attended produce anoverconfidence of personnel that in turn is conducive to making mistakes Anotherindividual factor also contributes to this overconfidence nursesrsquo perception of therobustness of guidelines provided by the general protocols of triage methodology

The first solution displayed in Table VII for Alpha in simple scenarios (POYHSsimYEDsimYTEDPEDPTM) with a consistency of about 070 and a coverage slightly higher thanthe third solutionrsquos row partially confirms the result that emerged above This solutionshows that a limited or lacking work experience in EDs implies a susceptibility to errorsdespite a prolonged working history in other health operative units and the perceivedreliability of triage protocols

The Beta samplersquos results related to simple scenarios (Table VII-first box on the rightside) show substantial differences compared to what was just reported in the case of Alpha

Variable Measure

PO 1 if the decision has been made using vital signs0 if the decision has been made without using vital signs

YHS Number of years of experience in the health sectorYED Number of years of experience in an EDYTED Number of years of experience in this specific EDPTM 1 if the operator declares to be confident in the Triage methodology

0 if the operator declares to be not confident in the Triage methodologyPED 1 if the operator declares to be confident in the Triage methodology as it is applied in the specific ED

0 if the operator declares to be not confident in the Triagemethodology as it is applied in the specific EDCT Count of attended training courses on triage

Table VIVariable in the

fsQCA analysis andtheir measure

2161


comparativeanalysis

Alpha

Beta

Configuration

RAW

COVERAGE

UNIQUE

COVERAGE

CONSIST

ENCY

CONFIGURATION

RAW

COVERAGE

UNIQUE

COVERAGE

CONSIST

ENCY

Simple

POY

HSsimYEDsim

YTEDP

EDP

TM

0442105

00547369

0697674

simPO

YHSYEDY

TEDsim

PEDsim

PTMC

T00392402

00392402

1PO

simYEDsim

YTEDP

EDP

TMC

T0393684

000631577

0653846

POY

HSYEDY

TEDsim

PEDP

TMC

T0172524

0137345

0717517

POY

HSsimYTEDP

EDP

TMC

T0406316

00189474

0725564

simPO

simYHSsimYEDsim

YTEDsim

PEDsim

PTMC

T00652632

00652632

054386

Solutio

ncoverage

0532632

Solutio

ncoverage

0176586

Solutio

nconsistency

0575

Solutio

nconsistency

0765574

Interm

ediate

POY

HSsimYEDsim

YTEDP

TMsim

CT0472258

0104516

0831818

simPO

YHSsimYTEDP

EDP

TMsim

CT0137203

00764015

0918954

POsim

YEDsim

YTEDP

EDP

TMC

T0296774

00309677

0804196

POY

HSYEDY

TEDP

TMsim

CT0114192

00651566

0993228

POY

HSsimYTEDP

EDP

TMC

T0265806

00774436

simPO

simYHSsimYEDsim

YTEDP

EDP

TMC

T0111387

00548768

0938837

simPO

simYHSsimYEDsim

YTEDsim

PEDP

TMC

T0211613

0211613

0811881

POY

HSsimYEDsim

YTEDsim

PEDP

TMC

T00497569

00124166

0984593

POY

HSYEDY

TEDsim

PEDP

TMC

T0154839

00258064

0736196

Solutio

ncoverage

0767742

Solutio

ncoverage

0306993

Solutio

nconsistency

0750315

Solutio

nconsistency

0938447

Com

plex

POsim

YHSsimYEDsim

YTEDsim

PEDC

T0217628

00620239

1simPO

YHSYEDY

TEDsim

PEDsim

PTMC

T00278783

00278782

1simYHSsimYEDsim

YTEDsim

PEDP

TMC

T0309032

00261153

0879257

POY

HSYEDY

TEDsim

PEDP

TMsim

CT0172524

0172524

0946586

POsim

YHSsimYEDsim

YTEDP

TMC

T0311208

00772579

0953333

simPO

YHSYEDY

TEDP

EDP

TMsim

CT0249273

0249273

0958017

simPO

YHSYEDsim

YTEDsim

PEDP

TMsim

CT0198041

00707291

0764706

POY

HSYEDY

TEDsim

PEDP

TMC

T009358

00304679

1PO

YHSYEDsim

YTEDP

EDP

TMC

T0085963

000761694

0918605

Solutio

ncoverage

0635473

Solutio

ncoverage

0449675

Solutio

nconsistency

0870343

Solutio

nconsistency

0956075

Table VIIResults of fsQCA insimple intermediateand complex clinicalscenarios both forAlpha and Betaemergencydepartments

2162

MD5610

There is a solution that achieves the highest level of consistency although the degree ofcoverage does not display a high empirical relevance The fact that we can identify asolution with a high level of consistency (simple scenarios) in the case of Beta unlikethe case of Alpha can be interpreted in accordance with what was previously assumedIn Alpha in the case of simple scenarios the level of correct codes assigned by theoperators is equal to 7545 percent in the case of Beta more errors are identified (64 percentof correct codes)

The first row of Table VII for Betarsquos sample in simple scenarios (simPOYHSYEDYTEDsimPEDsimPTMCT) shows that in Betarsquos ED the high level of errors can beexplained by the lack of reference to objective information (simPO) associated with a highlevel of experience in the health sector (YHS) and in EDs (YED YTED) and with lowconfidence in the robustness and reliability of triage methodology (simPTM) including how itis applied in the specific ED (simPED) The theoretical knowledge acquired through attendingtraining courses (CT) also seems to be detrimental

To interpret these results we can recall some organizational characteristics of Betarsquos EDThe triage is normally performed in two steps and the use of vital parameters is oftenpostponed from the first phase to the second phase Betarsquos triage operators exhibit a slightlyhigher seniority than those of Alpha in the specific ED Finally in Beta there are no specificprotocols and guidelines on how to implement the triage In simple cases the availableinformation is limited and unambiguous and the use of objective elements should lead to thecorrect solution Instead in the case of Beta nurses tend to neglect the measurement of vitalparameters especially in clinical cases classified as ldquosimplerdquo because of practices acquiredin the specific organizational context it seems that there is an excessive recourse to basictheoretical knowledge and to experience gained in the field that when associated with a lackof confidence in manuals procedures and ministerial protocols leads to errors

In intermediate scenarios and for Alpharsquos sample four configurations are displayed thatpassed the consistency test and that exhibit an acceptable level of coverage

The most consistent configuration for the Alpha sample (POYHSsimYEDsimYTEDPTMsimCT) is also the most empirically relevant in the set of intermediateclinical scenarios This solution reinforces some of results discussed for simple scenariosLooking at all the configurations that emerged as solutions for Alpha and in the case ofintermediate clinical scenarios it can be observed that the weak experience in EDs (simYEDsimYTED) and the lack of coherence among cues are compensated for by an overconfidence ofnurses in the general guidelines available in the triage methodology (PTM) But this kind ofbehavior is not beneficial to the effectiveness of triage implementation

Referring to Beta in intermediate complex scenarios (Table VII-second box on the rightside) it can be noticed immediately that all the solutions passed the consistency test

The solution with the highest consistency (POYHSYEDYTEDPTMsimCT) showsthat in intermediate scenarios errors are mainly related to a reliance on objectiveparameters (PO) and work experience (YHSYEDYTED) accompanied by operatorsrsquoreference to general guidelines (PTM) and non-adequate theoretical knowledge acquiredthrough training (simCT) The experience of Betarsquos nurses seems to be the major driver ofassessment errors together with little attention to formal training

With respect to complex scenarios and Alpharsquos sample there are six emergentconfigurations representing sufficient conditions for the occurrence of the outcome All theidentified solutions present a consistency above the suggested threshold The coverage asexpected is noticeably less than in the cases discussed above for Alpharsquos sample

The configurations that exhibit a consistency equal to 1 (POsimYHSsimYEDsimYTEDsimPEDCT POYHSYEDYTEDsimPEDPTMCT) reveal that the highpropensity of nurses to consider the objective parameters (PO) in the assessment ofpriority codes associated with a high number of attended training courses (CT) and with a

2163


comparativeanalysis

lack of confidence in the specific triage guidelines of the ED under investigation (sim PED) aresusceptible to errors in complex scenarios for Alpha Furthermore as shown in the secondthird fifth and sixth rows of the last box of Table VII (left side) the combination of anintense perception of the effectiveness of the general triage methodology (PTM) and a highnumber of training courses (CT) attended probably determines nursesrsquo strong recourse totheoretical knowledge without considering other information and informal rules providedby the specific work context Additionally the use of vital signs to make decisions (PO) ispresent in most of the highly consistent solutions (rows 1 3 5 6 of table VII- third box on theleft side) as is the lack of experience in the specific ED This is also true for simple andintermediate clinical scenarios

Finally the third box on the left side of Table VII reports three complex solutions thatemerged from the elaboration of data referring to Betarsquos nurses in complex scenariosAll these configurations show a consistency above the threshold and an acceptablelevel of coverage The solution with greater consistency (simPOYHSYEDYTEDsimPEDsimPTMCT) shows that Betarsquos triage operators commit mistakes in complexscenarios when they rely too much on their knowledge base (YHS CT) and their experiencein EDs and in the specific ED (YEDYTED) paying limited attention to objectiveparameters (simPO) and lacking confidence in triage methodology and how it is applied in thespecific context under analysis (simPEDsimPTM) Another solution with high consistencyand with a level of coverage emerges higher than the solution examined above(simPOYHSYEDYTEDPEDPTMsimCT) In this case the Beta operators seem to relymainly on their experience and confidence in the general and organizational rules (even ifthese are unwritten rules because Beta does not have specific protocols and guidelines)Also in this case as in the previous one triage nurses do not rely very often on vitalparameters In the case of the first solution examined (with a consistency of 1 and a very lowcoverage) the error is determined by the high experience in the field and the theoreticalknowledge acquired through training courses in the case of the second examined solutionthe error seems to be determined again by recourse to individual work experience and alsoby a reference to formal (PTM) and informal rules (PED) available in Betarsquos ED It isinteresting to note that in the case of Betarsquos sample the solution with the highestconsistency in simple scenarios is also one of the solutions with higher consistency incomplex ones (simPOYHSYEDYTEDsimPEDsimPTMCT)

5 DiscussionThe results described in the previous section lead to three relevant findings representingthe main contribution of this research to the scientific debate on the decision-making processin triage

First factors usually analyzed by the literature as elements characterizing the triageprocess cannot be isolated from each other when assessing their impact on decision-makingoutcomes Groups of homogeneous factors (knowledge and experience recourse to objectiveparameters and guidelines perception of the reliability of guidelines protocols and informalrules of the organization) combine with each other and do so differently in the twoorganizational settings under investigation

This is in line with what emerged from the analysis of the literature summarized inTable I Numerous studies highlight through a descriptive approach that the experience ofnurses affects the intensity of their use of vital parameters (Chung 2005 Vatnoslashy et al 2013)The implementation of protocols and guidelines determines a greater use of vital parameters(Gerdtz and Bucknall 2001) furthermore the high level of nursesrsquo experience fosters aclimate of nursing satisfaction and greater trust (Andersson et al 2006) On the other handthe literature is unable to assess in a definite way the impact of single or homogeneousfactors on the outcomes of the triage process For example it has not been established

2164

MD5610

whether a high level of nursesrsquo experience positively affects the accuracy of acuity levelsrsquoassignments (Martin et al 2014) This lack of statistical evidence could be explained by thecomplex adaptive nature of the decisional process (deMattos et al 2012) which requiresmore attention to non-linear relationships that occur between factors related to differentlevels of analysis (individual groups organization) From the methodological point of viewthis implies avoiding traditional variable-oriented (Ragin 1987) approaches adopting linearand additive perspectives (eg linear regression factor analysis)

Second results clearly show no single pattern is able to explain the emergence oferrors We can observe that there are regularities in the configurations of factors leadingto a high level of mistakes and that these regularities are different in the twoorganizational contexts analyzed In the case of Alpharsquos sample the reliance on objectiveparameters (particularly for beginners) the scarce experience in the specific ED and inEmergency and confidence in the effectiveness of triage protocols and guidelines aremainly related to the highest levels of errors In practical terms it emerges clearly inAlpha the need of achieving a balance between the level of work experience in Emergencyand the level of work experience in other areas of healthcare This result could be reachedby structurally revising recruiting policies or by designing specific training on the jobinitiatives for beginners of triage

In the case of Beta instead the scarce recourse to objective parameters and the highamount of work experience particularly in the specific ED are related to the generation ofassessment mistakes In some cases the effect of these elements is amplified by areference to general protocols and a lack of confidence in the specific organizational rules(shared informal rules) The managerial levers to be considered for reducing errors in thiscontext above all in simple cases could involve training interventions aimed at sensitizingexpert operators to consider the vital parameters more carefully The creation of localguidelines which underline the importance of certain objective variables could be a furtherelement to consider

The finding above can be traced back to the research of Wolf (2010) which emphasizesthe importance of organizational rules ( formal and informal) in determining the ways inwhich nurses seek and assign meaning to the information used to make decisions Decisionsare an output of the interplay between nursesrsquo individual frames and frames socially sharedin a specific organizational context It also confirm the assumption of this research using theperspective of ecological rationality of Gigerenzer et al (1999) on heuristics and helps us indiscussing the third relevant finding of our study

In each of the considered EDs the configurations of factors leading to errors showspecific regularities that seem to be not strictly dependent on the level of complexity ofsimulated tasks The specificity of the decisional situations disappears in the face of thespecificity of organizational environments The ldquocomplexityrdquo of medical scenarios inour study represents what Todd and Gigerenzer (2012) name ldquothe structure of theinformationrdquo of situations assessed by nurses The complexity in fact is characterized interms of level of uncertainty and the availability or redundancy of information Todd andGigerenzer (2012) however highlight that ldquothe situationrdquo is conveyed or filtered by theenvironment Individuals choose to consider one piece of information rather than anotheror give weight to one piece of information rather than another based also on behaviorsand rules that are collectively shared in the environment in which the decision is madeOur results therefore remind us of the need to consider the complexity of the task in lightof the constraints and resources that characterize the specific organizational context inwhich nurses work

In summary our findings suggest that no single factors (or homogeneous groups offactors) could explain the outcomes of decision-making in triage assessment alone Factorsrelated to different levels of analysis (individual group situation organization) have to be

2165


comparativeanalysis

analyzed together adopting a perspective that is able to take into account their complexinteraction and the non-linearity of their relationships as well as the outcome of thedecision-making process This opens up a new perspective for research and practice

6 ConclusionsThis paper addresses a topic widely analyzed by the literature on clinical decision-makingthe identification of factors influencing triage nursesrsquo decision-making process and theevaluation of their impact on triage outcomes The workrsquos innovative contribution to thedebate is twofold

First the analysis of factors impacting triage decision-making was framed usingthe perspective of ecological rationality proposed by Gigerenzer et al (1999) to explain theperformance of fast and frugal heuristics This perspective informs Wolfrsquos research (20102013) although not explicitly and outlines the need to consider nursesrsquo decision-making intriage as a complex process in which different elements at different levels of analysis(individual organizational and environmental) interact and co-evolve in determiningprocess outcomes In other healthcare contexts where decision-making processes arecharacterized by uncertainty and time pressure the perspective of ecological rationality onheuristics is present (see for example Rudolph et al 2009) and drives researchers to modeldecision-making processes as complex adaptive and path-dependent The findings of thispaper could be applied in these different healthcare empirical settings as well in order toshed light on the interplay of factors affecting the accuracy of decisional processes

Second in accordance with the theoretical premise the paper adopts a qualitativemethodology that allows for integrating the richness of case-oriented approaches with theformalization of variable-oriented approaches (Ragin 2006) To the best of our knowledgethis is the first application of QCA to the topic under investigation The paper has thuscontributed by proposing a methodological approach that preserves the specificity of theanalyzed cases and their intrinsic complexity without resorting to reductionist hypotheses

The main findings of the study suggest some implications for research Errors in theassignment of triage priority codes are determined by the interplay between differentfactors some relating to the individual level and others related to the organizational levelThese groups of factors interact and co-evolve determining specific answers to specificsituations these latter being filtered and interpreted in the light of the constraints andresources of the context in which the decision is made It is therefore necessary to notisolate individual factors from each other and from the organizational and contextual onesin the analysis and to avoid linear and additive approaches The perspective inspired bythe theory of Complex Adaptive Systems (Holland 2006) could be particularly suitable forthis issue In Complex Adaptive Systems individual agents interact in a specificenvironment characterized by opportunities and threats following their local rules andpreferences (ldquointernal modelsrdquo or ldquomicro-specificationsrdquo) and co-evolving with theenvironment itself Their interactions are not linear and determine the emergence at thecollective level of macro-regularities that cannot be explained by completelydeconstructing the system and studying the local behaviors of agents To identifypossible explanations for aggregated properties it is necessary to adopt a ldquogenerativerdquoapproach (Epstein and Axtell 1996) using methodologies that are able to identify sets ofmicro-specifications sufficient to explain the emergence of the collective outcome In thisstudy the exploratory analysis has been conducted through fsQCA which allowed us tooutline different patterns of factors that determine the emergence of errors Based on thisresult further developments of the research could be proposed in order to develop anagent-based model calibrated through empirical data This model would be useful toevaluate the impact of additional contextual factors and assess ex-ante the effect of somemanagerial interventions on the accuracy of decision-making processes in triage and in

2166

MD5610

other healthcare contexts in which uncertainty and time pressure make decisionalprocesses complex dynamic and adaptive

This complexity could also inspire managerial practice The interventions aimed atimproving the effectiveness of triage practice and clinical decision-making in general shouldbe designed while avoiding two deviations hard managerial approaches (acting on formalrules procedures and structure) and soft approaches ( focused on the motivation of people)(Morieux and Tollman 2014) Managerial interventions should emerge instead froman in-depth knowledge of the organizational context and decision-making situationsand be aimed at fine-tuning the relationships between individuals and contextual resourcesand constraints

Some limitations affect this study First it was not possible to include contextual factorssuch as EDrsquos overcrowding patientsrsquo volume the effect of interruptions in the analysisfactors which can determine an increase in the level of operatorsrsquo stress and a potentialloss of information at the time of the decision (Hitchcock et al 2013 Wolf 2013)

Furthermore the absence of the patient at the moment of data collection prevented averification of the role of visual cues in the decision-making process Both these limitationsderive from the use of a simulative approach in the data collection step This choice wasdictated by the need to analyze the impact of situations characterized by different levels ofcomplexity and at the same time to keep research time limited Some measures have beenadopted to make the simulations closer resemble reality and increase the confidence of theresearchers about the resultsrsquo interpretation the data collection phase was preceded by aperiod of observation in the field limited time was given to the operators to assign prioritycodes to the analyzed scenarios as happens in real situations immediate interaction withother nurses was avoided as occurs during each work shift and finally scenarios proposedto nurses were enriched with information regarding the presentation of the patientat the door

Future research will revolve around adapting the protocol used during the fieldwork inorder to carry out a structured observational study during the situations experienced bynurses in the two organizational settings that were investigated By comparing theresults it will be possible to carry out a precise assessment of the implications of thesimulation approach

References

Adriaenssens J De Gucht V Van Der Doef M and Maes S (2011) ldquoExploring the burden ofemergency care predictors of stress‐health outcomes in emergency nursesrdquo Journal of AdvancedNursing Vol 67 No 6 pp 1317-1328

Andersson AK Omberg M and Svedlund M (2006) ldquoTriage in the emergency department ndash aqualitative study of the factors which nurses consider when making decisionsrdquo Nursing inCritical Care Vol 11 No 3 pp 136-145

Arslanian-Engoren C (2005) ldquoPatient cues that predict nursesrsquo triage decisions for acute coronarysyndromesrdquo Applied Nursing Research Vol 18 No 2 pp 82-89

Artinger F Petersen M Gigerenzer G and Weibler J (2015) ldquoHeuristics as adaptive decisionstrategies in managementrdquo Journal of Organizational Behavior Vol 36 No S1 pp S33-S52

Burchill CN and Polomano R (2016) ldquoCertification in emergency nursing associated with vital signsattitudes and practicesrdquo International Emergency Nursing Vol 27 No 4 pp 17-23

Cabana MD Rand CS Powe NR Wu AW Wilson MH Abboud PAC and Rubin HR (1999)ldquoWhy donrsquot physicians follow clinical practice guidelines A framework for improvementrdquoJAMA Vol 282 No 15 pp 1458-1465

Chase VM Hertwig R and Gigerenzer G (1998) ldquoVisions of rationalityrdquo Trends in CognitiveSciences Vol 2 No 6 pp 206-214

2167


comparativeanalysis

Chung JY (2005) ldquoAn exploration of accident and emergency nurse experiences of triage decisionmaking in Hong Kongrdquo Accident and Emergency Nursing Vol 13 No 4 pp 206-213

Cioffi J (1998) ldquoDecision making by emergency nurses in triage assessmentsrdquo Accident andEmergency Nursing Vol 6 No 4 pp 184-191

Cioffi J (2001) ldquoClinical simulations development and validationrdquo Nurse Education Today Vol 21No 6 pp 477-486

Cioffi J and Markham R (1997) ldquoClinical decision-making by midwives managing case complexityrdquoJournal of Advanced Nursing Vol 25 No 2 pp 265-272

Cone KJ and Murray R (2002) ldquoCharacteristics insights decision making and preparation of EDtriage nursesrdquo Journal of Emergency Nursing Vol 28 No 5 pp 401-406

Conen D Leimenstoll BM Perruchoud AP and Martina B (2006) ldquoRoutine blood pressuremeasurements do not predict adverse events in hospitalized patientsrdquo The American Journal ofMedicine Vol 119 No 1 pp 70-e17

Cooper RJ Schriger DL Flaherty HL Lin EJ and Hubbell KA (2002) ldquoEffect of vital signs ontriage decisionsrdquo Annals of Emergency Medicine Vol 39 No 3 pp 223-232

Cosier RA and Dalton DR (1988) ldquoPresenting information under conditions of uncertainty andavailability some recommendationsrdquo Systems Research and Behavioral Science Vol 33 No 4pp 272-281

Cronqvist L (2005) ldquoIntroduction to multi-value qualitative comparative analysisrdquo COMPASSSdidactics paper No 20054 MVQCA Maryland MD

Croskerry P and Sinclair D (2001) ldquoEmergency medicine a practice prone to errorrdquo CanadianJournal of Emergency Medicine Vol 3 No 4 pp 271-276

deMattos PC Miller DM and Park EH (2012) ldquoDecision making in trauma centers from thestandpoint of complex adaptive systemsrdquo Management Decision Vol 50 No 9 pp 1549-1569

Derlet RW and Richards JR (2000) ldquoOvercrowding in the nationrsquos emergency departments complexcauses and disturbing effectsrdquo Annals of Emergency Medicine Vol 35 No 1 pp 63-68

Drechsler M Katsikopoulos K and Gigerenzer G (2014) ldquoAxiomatizing bounded rationality thepriority heuristicrdquo Theory and Decision Vol 77 No 2 pp 183-196

Dy SM Garg P Nyberg D Dawson PB Pronovost PJ Morlock L and Wu AW (2005) ldquoCriticalpathway effectiveness assessing the impact of patient hospital care and pathwaycharacteristics using qualitative comparative analysisrdquo Health Services Research Vol 40No 2 pp 499-516

Epstein JM and Axtell R (1996) Growing Artificial Societies Social Science From the Bottom UpBrookings Institution Press Washington DC

Fiss PC (2011) ldquoBuilding better causal theories a fuzzy set approach to typologies in organizationresearchrdquo Academy of Management Journal Vol 54 No 2 pp 393-420

Fiss PC (2009) ldquoPractical issues in QCArdquo Presentation at Academy of Management 2009 available atwwwresearchgatenetprofilePeer_Fisspublication266471735_Practical_Issues_in_QCAlinks56bb757508ae7be8798bc0c4Practical-Issues-in-QCApdf

Frykberg ER (2005) ldquoTriage principles and practicerdquo Scandinavian Journal of Surgery Vol 94 No 4pp 272-278

Garbez R Carrieri-Kohlman V Stotts N Chan G and Neighbor M (2011) ldquoFactors influencingpatient assignment to level 2 and level 3 within the 5-level ESI triage systemrdquo Journal ofEmergency Nursing Vol 37 No 6 pp 526-532

Gerdtz MF and Bucknall TK (2001) ldquoTriage nursesrsquo clinical decision making an observationalstudy of urgency assessmentrdquo Journal of Advanced Nursing Vol 35 No 4 pp 550-561

Gerdtz MF and Bucknall TK (2007) ldquoInfluence of task properties and subjectivity on consistency oftriage a simulation studyrdquo Journal of Advanced Nursing Vol 58 No 2 pp 180-190

2168

MD5610

Gigerenzer G (1996) ldquoOn narrow norms and vague heuristics a reply to Kahneman and TverskyrdquoPsychological Review Vol 103 No 3 pp 592-596

Gigerenzer G and Kurzenhaumluser S (2005) ldquoFast and frugal heuristics in medical decisionmakingrdquo in Ribace R Laird JD Noller KL and Valsiner J (Eds) Science and Medicine inDialogue Thinking Through Particulars and Universals Praeger Westport CT pp 3-15

Gigerenzer G Todd PM and ABC Research Group T (1999) Simple Heuristics That Make us SmartOxford University Press

Goumlransson KE Ehnfors M Fonteyn ME and Ehrenberg A (2008) ldquoThinking strategies used byregistered nurses during emergency department triagerdquo Journal of Advanced Nursing Vol 61No 2 pp 163-172

Greckhamer T Misangyi VF and Fiss PC (2013) ldquoChapter 3 the two QCAs from a small-Nto a large-N set theoretic approachrdquo in Fiss PC Cambreacute B and Marx A (Eds) ConfigurationalTheory and Methods in Organizational Research Emerald Group Publishing Limitedpp 49-75

Greenwood J Sullivan J Spence K and McDonald M (2000) ldquoNursing scripts and the organizationalinfluences on critical thinking report of a study of neonatal nursesrsquo clinical reasoningrdquo Journalof Advanced Nursing Vol 31 No 5 pp 1106-1114

Hitchcock M Gillespie B Crilly J and Chaboyer W (2013) ldquoTriage an investigation of the processand potential vulnerabilitiesrdquo Journal of Advanced Nursing Vol 70 No 7 pp 1532-1541

Holland JH (2006) ldquoStudying complex adaptive systemsrdquo Journal of Systems Science and ComplexityVol 19 No 1 pp 1-8

Kahneman D (2011) Thinking Fast and Slow Macmillan London

Kahneman D and Tversky A (1977) Intuitive Prediction Biases and Corrective Procedures Decisionsand Designs Inc Mclean Va Oregon OR

Kahneman D and Tversky A (1981) ldquoThe simulation heuristicrdquo No TR-5 Department ofPsychology Stanford University California CA

Kuncel NR Goldberg LR and Kiger T (2011) ldquoA plea for process in personality prevaricationrdquoHuman Performance Vol 24 No 4 pp 373-378

Lampi M Junker J Berggren P Jonson CO and Vikstroumlm T (2017) ldquoPre-hospital triageperformance after standardized trauma coursesrdquo Scandinavian Journal of TraumaResuscitation and Emergency Medicine Vol 25 No 1 pp 53-58

Legewie N (2013) ldquoAn introduction to applied data analysis with qualitative comparative analysisrdquo InForum Qualitative SozialforschungForum Qualitative Social Research Vol 14 No 3 pp 1-45

Luan S Schooler LJ and Gigerenzer G (2011) ldquoA signal-detection analysis of fast and-frugal treesrdquoPsychological Review Vol 118 No 2 pp 316-338

McMillan JR Younger MS and DeWine LC (1986) ldquoSatisfaction with hospital emergencydepartment as a function of patient triagerdquo Health Care Management Review Vol 11 No 3pp 21-27

Marsden J (2000) ldquoAn evaluation of the safety and effectiveness of telephone triage as a method ofpatient prioritization in an ophthalmic accident and emergency servicerdquo Journal of AdvancedNursing Vol 31 No 2 pp 401-409

Martignon L and Hoffrage U (2002) ldquoFast frugal and fit Simple heuristics for paired comparisonrdquoTheory and Decision Vol 52 No 1 pp 29-71

Martin A Davidson CL Panik A Buckenmyer C Delpais P and Ortiz M (2014) ldquoAn examinationof ESI triage scoring accuracy in relationship to ED nursing attitudes and experiencerdquo Journalof Emergency Nursing Vol 40 No 5 pp 461-468

Marx A Cambreacute B and Rihoux B (2013) ldquoChapter 2 crisp-set qualitative comparative analysis inorganizational studiesrdquo in Fiss PC Cambreacute B and Marx A (Eds) Configurational Theory andMethods in Organizational Research Emerald Group Publishing pp 23-47

2169


comparativeanalysis

Meissner P and Wulf T (2017) ldquoThe effect of cognitive diversity on the illusion of control bias instrategic decisions an experimental investigationrdquo European Management Journal Vol 35No 4 pp 430-439

Melby V Gillespie M and Martin S (2011) ldquoEmergency nurse practitioners the views of patients andhospital staff at a major acute trust in the UKrdquo Journal of Clinical Nursing Vol 20 Nos 1‐2pp 236-246

Morieux Y and Tollman P (2014) Six Simple Rules How to Manage Complexity Without GettingComplicated Harvard Business Review Press Massachusetts MA

Nakagawa J Ouk S Schwartz B and Schriger DL (2003) ldquoInterobserver agreement in emergencydepartment triagerdquo Annals of Emergency Medicine Vol 41 No 2 pp 191-195

Noon AJ (2014) ldquoThe cognitive processes underpinning clinical decision in triage assessment atheoretical conundrumrdquo International Emergency Nursing Vol 22 No 1 pp 40-46

Ordanini A Parasuraman A and Rubera G (2014) ldquoWhen the recipe is more important than theingredients a qualitative comparative analysis (QCA) of service innovation configurationsrdquoJournal of Service Research Vol 17 No 2 pp 134-149

Ragin CC (2008) Redesigning Social Inquiry Fuzzy Sets and Beyond Vol 240 University of ChicagoPress Chicago IL

Ragin CC (1987) The Comparative Method Moving Beyond Qualitative and Quantitative MethodsUniversity of California Berkeley CA

Ragin CC (2000) Fuzzy-Set Social Science University of Chicago Press Chicago IL

Ragin CC (2006) ldquoSet relations in social research evaluating their consistency and coveragerdquo PoliticalAnalysis Vol 14 No 3 pp 291-310

Rihoux B (2006) ldquoQualitative comparative analysis (QCA) and related systematic comparativemethods Recent advances and remaining challenges for social science researchrdquo InternationalSociology Vol 21 No 5 pp 679-706

Rihoux B and De Meur G (2008) ldquoCirsp-set qualitative comparative analysis (csQCA) and relatedtechniquesrdquo in Ragin C and Rihoux B (Eds) Configurational Comparative MethodsQualitative Comparative Analysis (QCA) and Related Techniques Sage PublicationsCalifornia CA pp 33-68

Rihoux B and Marx A (2013) ldquoQCA 25 years after lsquothe comparative methodrsquo mapping challengesand innovations ndash mini-symposiumrdquo Political Research Quarterly Vol 66 No 1 pp 167-235

Rudolph JW Morrison JB and Carroll JS (2009) ldquoThe dynamics of action-oriented problemsolving linking interpretation and choicerdquo Academy of Management Review Vol 34 No 4pp 733-756

Salk ED Schriger DL Hubbell KA and Schwartz BL (1998) ldquoEffect of visual cues vital signs andprotocols on triage a prospective randomized crossover trialrdquo Annals of Emergency MedicineVol 32 No 6 pp 655-664

Smith M Higgs J and Ellis E (2008) ldquoFactors influencing clinical decision makingrdquo in Higgs J et al(Eds) Clinical Reasoning in the Health Professions 3rd ed Elsevier Churchill LivingstoneEdinburgh and New York NY

Stanfield LM (2015) ldquoClinical decision making in triage an integrative reviewrdquo Journal of EmergencyNursing Vol 41 No 5 pp 396-403

Storm‐Versloot MN Verweij L Lucas C Ludikhuize J Goslings JC Legemate DA andVermeulen H (2014) ldquoClinical relevance of routinely measured vital signs in hospitalizedpatients a systematic reviewrdquo Journal of Nursing Scholarship Vol 46 No 1 pp 39-49

Todd PM and Gigerenzer G (2012) Ecological Rationality Intelligence in the World Oxford UniversityPress

Tversky A and Kahneman D (1974) ldquoJudgment under uncertainty Heuristics and biasesrdquo ScienceVol 185 No 4157 pp 1124-1131

2170

MD5610

Van der Wulp I Van Baar ME and Schrijvers AJP (2008) ldquoReliability and validity of the Manchestertriage system in a general emergency department patient population in the Netherlands results ofa simulation studyrdquo Emergency Medicine Journal Vol 25 No 7 pp 431-434

Vatnoslashy TK Fossum M Smith N and Sletteboslash Å (2013) ldquoTriage assessment of registered nurses inthe emergency departmentrdquo International Emergency Nursing Vol 21 No 2 pp 89-96

Wolf L (2010) ldquoDoes your staff really lsquogetrsquo initial patient assessment Assessing competency intriage using simulated patient encountersrdquo Journal of Emergency Nursing Vol 36 No 4pp 370-374

Wolf L (2013) ldquoAn integrated ethically driven environmental model of clinical decision making inemergency settingsrdquo International Journal of Nursing Knowledge Vol 24 No 1 pp 49-53

Wood R and Bandura A (1989) ldquoSocial cognitive theory of organizational managementrdquo Academy ofManagement Review Vol 14 No 3 pp 361-384

Corresponding authorCristina Ponsiglione can be contacted at ponsigliuninait


2171


comparativeanalysis

Assessing the conformityto clinical guidelines

in oncologyAn example for the multidisciplinary

management of locally advancedcolorectal cancer treatment

Jacopo LenkowiczFondazione Policlinico A Gemelli IRCCS ndash Universitagrave Cattolica del Sacro Cuore

Rome ItalyRoberto Gatta

Fondazione Policlinico Universitario A Gemelli IRCCS Rome ItalyCarlotta Masciocchi

Fondazione Policlinico A Gemelli IRCCS ndash Universitagrave Cattolica del Sacro CuoreRome Italy

Calogero Casagrave and Francesco CelliniFondazione Policlinico Universitario A Gemelli IRCCS Rome Italy

Andrea DamianiFondazione Policlinico A Gemelli IRCCS ndash Universitagrave Cattolica del Sacro Cuore

Rome ItalyNicola Dinapoli

Fondazione Policlinico Universitario A Gemelli IRCCS Rome Italy andVincenzo Valentini


AbstractPurpose ndash The purpose of this paper is to describe a methodology to deal with conformancechecking through the implementation of computer-interpretable-clinical guidelines (CIGs) and presentan application of the methodology to real-world data and a clinical pathway for radiotherapy-relatedoncological treatmentDesignmethodologyapproach ndash This methodology is implemented by a software able to use the hospitalelectronic health record data to assess the adherence of the actual executed clinical processes to a clinicalpathway monitoring at the same time management-related efficiency and performance parameters andideally suggesting ways to improve themFindings ndash Three use cases are presented in which the results of conformance checking are used to comparedifferent branches of the executed guidelines with respect to the adherence to ideal process temporaldistribution of state-to-state transitions and overall treatment efficacy in order to extract data-drivenevidence that could be of interest for the hospital managementOriginalityvalue ndash This approach has the result of applying management-oriented data mining techniqueon sequential data typical of process mining to the result of a conformity check between the preliminaryknowledge defined by clinicians and the real-world data typical of CIGsKeywords Conformance checking Evidence-based practice Process mining Clinical guidelinesClinical pathwayPaper type Research paper


Received 29 September 2017Revised 20 February 201824 April 2018Accepted 2 May 2018


2172

MD5610


1 IntroductionThe aim toward evidence-based management in a health care setting has to confront itselfwith the general fact that evidence-based management like evidence-based medicine entailsa mind-set that clashes with the way many managers and companies operate as it features awillingness to put aside belief and conventional wisdom and replace these with anunrelenting commitment to gather the necessary facts to make more informed andintelligent decisions (Pfeffer and Sutton 2006) Also if on the one hand the adoption ofevidence-based practices found a fertile ground in the clinician culture because it is rootedin a formal body of shared knowledge on the other hand the managerial culture is muchless homogeneous and has often little involvement in or experience with the evidence ofscientific research In other words as Walshe and Rundall (2001) put it in their paperldquoHealth care managers and researchers in health care management are not one communitybut twordquo To complete the picture it is worth to add the key point that also from theclinical literature emerges the fact that health care practices must be guided byevidence-based information and management and that an effective application of clinicalgovernance and evidence-based performance management are increasingly needed forhealth care decision makers and hospitals (Trabacchi et al 2008 Dwyer et al 2012Travaglia et al 2011 Ravaghi et al 2013)

A way to test the chance of a mature adoption of evidence-based management inhealth care is to ask to which degree the management is able or willing to embrace themind-set of a researcher in his or her job and symmetrically how much the healthcare workers are willing to accept such kind of shift in the management mind-setSteps toward merging these diverging points of view can be done by proposingframeworks that are able to capture different aspects of an organizational workflow thusresponding at the same time to the needs of the different stakeholders Such kind ofproposal would be for instance a platform able to assess the adherence to a set of clinicalguidelines which at the same time monitors management-related efficiency orperformance parameters and suggests ways to improve them the common interest inthis case would be that not acknowledging the difference between the pathways actuallyfollowed by patients and the desired one can lead the clinical decision makers to asub-optimal management of the clinical case and from the management point of view theresult is an inefficient allocation of resources creating bottlenecks on services andwasting time and money

A suitable strategy to achieve this kind of interaction between the health care workersand the management is to exploit concepts and methods from an emerging but promisingtopic namely process mining (van der Aalst 2016a b) and blend them with those from amore established one computer-interpretable-clinical guidelines (CIGs) Indeed processmining has its main focus on the real-world data and aims at discovering the processesactually in use with no or little preliminary knowledge of the real organizational processesthe scope of CIGs (Hripcsak 1994 Wang et al 2002 Peleg 2013) on the other hand is toprovide tools and frameworks to implement a wide corpus of preliminary knowledge in theform of validated clinical guidelines on computers allowing measurements of the adherenceof the clinical practice to such guidelines

Such a mixed approach would have the result of applying management-oriented datamining techniques on sequential data typical of process mining to the result of a conformitycheck between the preliminary knowledge defined by clinicians and the real-world datatypical of CIGs From this perspective there is a wide range of applications that span boththe clinical and management fields suggesting what to do in a specific clinical caseallowing the definition of the legally supported clinical procedures check the degree ofadherence of a care unit to best practices analyze the time spent by patients in any step oftheir clinical pathway and relative cost compared to the benefit assess if the human and

2173

Assessing theconformityto clinicalguidelines

financial resources can be better allocated to remove bottlenecks and improve processesand even support the education of new physicians for specific diseases treatment ordiagnostic procedures Finally it is crucial to underline that the implementation of practicesmeasuring the distance from the real-world clinical practice to what requested by clinicalguidelines enhances financial accountability and can be seen as a relevant factor during thenegotiation of budget objectives for hospitals

11 Conformance checking in process miningConformance checking on processes is a family of process mining techniques that compare aprocess model with an event log of the same process (van der Aalst 2016a b) It is used tocheck if the actual execution of a process as recorded in the event log conforms to the modeland vice versa

The interpretation of non-conformance depends on the purpose of the model

bull If the model is intended to be descriptive discrepancies between model and logindicate that the model needs to be improved to capture reality better

bull If the model is normative then such discrepancies may be interpreted in two waysthey may expose undesirable deviations (ie conformance checking signals the needfor a better control of the process) or may reveal desirable deviations (ie workersmay deviate to serve the customers better or to handle circumstances not foreseen bythe process model)

A raising interest in process mining applications to the health care domain has beenunderway for the last few years Reviews of the main efforts toward this goal can be found intwo different authors (Rojas et al 2016 Kurniati et al 2009) where it is highlighted thepreeminent role on the subject played by conformance checking analysis of the clinicalguideline implementation to validate the patientrsquos clinical pathways adherence Consequentlymany techniques were developed to perform conformance checking analysis such as theconformance checking based on alignments which is now one of the state-of-the-arttechniques (Adriansyah 2014 Bose and van der Aalst 2012) and available as a ProM plugin(Bose and van der Aalst 2012) In the analysis for this paper we use instead a self-developedtechnique described in Section 22 which is designed and developed in accordance with thespecific needs of the oncology department of a medium-size city hospital (around 1600 beds)

12 Computer-interpretable-clinical guidelinesAs defined in Field and Lohr (1990) clinical guidelines are ldquosystematically developedstatements to assist practitioner and patient decisions about appropriate health carefor specific clinical circumstancesrdquo They may offer concise instructions on which diagnosticor screening tests to order how to provide medical or surgical services how long patientsshould stay in hospital or other details of clinical practice (Woolf et al 1999)

CIG investigates how to represent a clinical guideline in order to make it computer ablegiven a patientrsquos pathway or sequence of clinical events to check if it complies with theguideline and suggest the ldquonext steprdquo to perform

One of the main challenges of CIG is the definition of suitable structured languages Theimportance of languages is twofold first the language should support physicians inrepresenting their clinical guidelines second the language should be easy to deal with byautomated tools Examples of existing languages include Arden Syntax (Hripcsak 1994)Asbru (Shahar et al 1998) and others described in other evidence (such as Peleg 2013)

In this paper we describe a particular methodology to deal with conformance checkingthrough the implementation of CIGs and we show an application of this methodology to areal-world event log and a clinical pathway with an integrated workflow which consists of

2174

MD5610

extracting data from an electronic health record (EHR) and turning them into an event logdefine the clinical pathway (Valentini et al 2012) in a computer interpretable way run theevent log against the guideline and output results in terms of patientsrsquo flow from state tostate and show some use cases that are integrated in an online dashboard to comparepathways actually followed by patients with respect to adherence to the ideal processesdefined by the clinical pathway temporal distribution of state-to-state transitions within theclinical pathway and overall treatment efficacy

2 BackgroundThis section addresses the software engine we used to define the CIG and to do conformancechecking with and the clinical pathway itself As to the latter we chose to work with aconsensus-based clinical pathway for the treatment of locally advanced rectal cancer asdefined in previous evidence (Valentini et al 2012) on the one hand this guideline is prettystraightforward and useful as a proof of concept for the methodology on the other hand ituses all the relevant data that a radiotherapy ward usually records and thus it makes thedata acquisition process less demanding

As to the software choice there are many software available for doing process miningsuch as PROM (van Dongen et al 2005) DISCO (Guumlnther and Rozinat 2012) and pMineR(Gatta et al 2017) Since our research center is also a hospital and consequently our effortsare markedly oriented to the practical needs of doctors in our case we have adoptedpMineR a software developed internally and released on CRAN (the official platform for therelease of packages in R httpsCRANR-projectorg) as designed and developed inaccordance with the specific needs of the oncology department of medium-size city hospital(around 1600 beds) These specific requirements are related for example to the possibilityof having data within a statistical analysis framework (R) some specific features for themanagement of time constraints in the attributes of some events the availability of arepresentation language of the guidelines more similar to the way of thinking of ourclinicians (according to them) with respect to the classical formalism of Petri nets (commonin process mining) or Arden Syntax GLIF Asbru (more common in the domain of CIGs)

21 The clinical guidelineThe guideline includes instructions on how to deal with the clinical management of locallyadvanced rectal cancer patients from the diagnosis to the post-surgery treatment which is aquite common kind of treatment pathway in radiotherapy departments and thus it can begeneralized to other guidelines and other pathologies The expression ldquolocally advancedrdquorefers to either an extramural extension or a regional lymph-nodal involvement without anydeep infiltration of surrounding pelvic organs precluding a microscopically radical surgicalresection (Valentini et al 2012)

For patients with this kind of diagnosis a neoadjuvant (ie pre-surgery) radiotherapytreatment is advised in combination with chemotherapy Moreover the guideline states thatthe time interval between the end of the neoadjuvant chemo-radiotherapy treatment and thesurgery itself has to be no less than 28 days and no more than 70 days

Figure 1 is shown a schematic representation of the guideline The three blocks on theleft depict the three entry points which are dependent on the clinical staging of T(tumor length) and N (lymphnodes involvement) at diagnosis (M is always equal to 0 sincethis is a guideline for non-metastatic patients) The second line of blocks from the left holdsthe information about the radiotherapy total dose and the combination of chemotherapyagents for the three different branches The third line of blocks states which type of surgeryis prescribed The last blocks on the right describe the details of the post-surgery treatmentwhich in agreement with our clinician was excluded from the analysis since the data that wehad did not allow for a straightforward representation of that clinical pathwayrsquos section

2175


Finally it is to notice that the way the guideline is represented in Figure 1 which is thediagram provided by our reference paper (Valentini et al 2012) has too high a level ofabstraction thus being of little use in order to build a computer interpretable version of itFor this reason a close collaboration with a team of radiotherapy oncologists was requiredto remove all the ambiguities and the ldquounknownsrdquo when defining the conditions of statesand transitions A more detailed explanation of this point can be found in the section onmaterials and methods

22 Pseudo-workflow language (PWL)The conformance checking utilities in pMineR are a set of tools specialized in conformancechecking In particular there is a class able to work with an internal formalism called PWLfor representing WorkFlow-like diagrams PWL is based on three main constructs

(1) events

(2) statuses and

(3) triggers

Given an event log the engine reads the list of events and for each event it tests if atrigger can be fired A trigger is an item composed by two main sections condition andeffects The condition part can check elements of the just read event log or other statusesof the patient (eg currently active statuses) If the condition applies the effects listed inthe subsequent section are executed For instance if the current status of the treatment isldquoin progressrdquo and a dismission report event is read the status of the patient has to beupdated according to the list of set and unset items Using this approach statuses areautomatically updated while the events are processed sequentially from the first tothe last

Figure 2 provides an example of the computation of a PWL for a dummy set of event log(on the left) and details about a specific patient (on the right) On the left the workflow isgraphed starting from the given XML used for defining triggers (squared boxes) andstatuses (rounds) On the top right an original event log which is an input of thecomputation On the bottom right the result of the computation for the same event log

T3 N0

Any T N1-2

T4 andor unresectabledisease

Clinical stage Primary treatment Adjuvant treatment

CI-5FURT orcapecitabineRT or 5times5

CI-5FURT orcapecitabineRT


Surgical resection

Resection if possible Any pT

De GramontCapecitabine orFOLFOXXELOX


Notes Tmdashclinical staging value T Nmdashclinical staging value N Mmdashclinical staging value MCImdashcontinuos infusion 5-FUmdash5-Fluoracile RTmdashradiotherapy FOLFOXXELOX treatmentschemas as defined in Valentini et al (2012) The three blocks on the left depict the three entrypoints which are dependent on the clinical staging of T (tumor length) and N (lymph-nodesinvolvement) at diagnosis (M is always equal to 0 since this is a guideline for non-metastaticpatients) The second line of blocks from the left holds the information about the radiotherapytotal dose and the combination of chemotherapy agents for the three different branches The thirdline of blocks states which type of surgery is prescribed The last blocks on the right describe thedetails of the post-surgery treatment which for the sake of simplicity when presenting the resultof conformance checking we ignored in the present implementation of the clinical guideline

Figure 1The clinical guidelineas it is in theoriginal document

2176

MD5610

plotted under the form of the ldquoactivation timerdquo of the different statuses Here the activationtime starts when the trigger for a state activation is fired and ends with the firing of thecorresponding unset trigger for that state

3 Materials and methods31 From clinical pathway to CIGsIn coding the clinical pathway into CIGs a pivotal challenge is to tackle the linguistic gapbetween the natural language (adopted to write the clinical guidelines) and a formallanguage (the only language which can be parsed by a computer) the former is the commonlanguage of the clinical domain and requires domain experts to be decoded the latter ismore commonly adopted in computer science and due to its relatively high complexityrequires specific technical skills to be properly handled

For this reason in order to build a computer interpretable version of the clinical pathwayin Figure 1 we worked in close collaboration with a team of radiotherapy oncologists

Because a ldquoone-hoprdquo translation was unfeasible we first translated the clinical pathwayin a semi-formal representation defining with the clinicians a graphical language able toreduce the ambiguity of the natural language guideline and which can be easily translatedin a PWL This ldquolanguage in the middlerdquo played as a linguistic contact point betweenclinicians and computer scientists

32 The dataThe data in process mining are normally stored in the event log In our case the event log wassubsequently build from this data set in such a way that the ldquoeventrdquo column of the event logencoded the type of eventmdashnamely clinical staging neoadjuvant radiotherapy neoadjuvantchemotherapy surgerymdashand the corresponding values as exemplified in Table I where eventtypes event values and the number of occurrences in the event log are shown

BEGIN0

Imaging Detected100

Waiting for a visit100

Visit detected100

Surg int detected40

RT detected60

Treated100

Patient treated with radio60

Patient treated with radiochemo10

End of Treatment

CHT detected10

EOT100

Patient operated40

Not treated yet100

Waiting for therapy100

Time-event for Patient 5

Waiting for a visit 11 days

ImagingJanuary 02 2000

VisitJanuary 13 2000

RadiotherapyJanuary 27 2000

DismissedFebruary 16 2000Chemotherapy

January 27 2000

January 02 2000 February 02 2000

25 days

14 days

0 days

20 days

20 days

Waiting for therapy

Patient treated with radio

Patient treated with radiochemo

Treated

Not treated yet

Figure 2An example of theoutput provided bypMineR after the

computation of a PWL

2177


The event log built this way has 9018 rows and 4 columns (id event start dateand end date) for a total of 3229 patientsrsquo traces The different event types in the eventlog are

bull ldquostaging Crdquo which is the clinical staging defined by the values of T (related totumor length) N (related to presence of positive lymphnodes) and M (relatedto the presence of metastasis) These parameters which we call attributes of theldquostaging Crdquo event can have the following values Tfrac14 01234 Nfrac14 012 andMfrac14 01

bull ldquonad rtrdquo is the event associated to the administration of a radiotherapic treatmentbefore the surgery Its attribute is the total delivered dose during the treatmentwhich is a numeric value expressed in gray

bull ldquonad ctrdquo is the event associated to the administration of a chemotherapic treatmentconcurrent to the radiotherapic one described above Its attribute is the list ofchemotherapy agents administered during the therapy

bull ldquosurgeryrdquo is the event associated to the surgical procedure the patient underwent toIts attribute is the type of surgical procedure which in this clinical setting can beAnterior resection APR Hartmann procedure and local excision

As we encoded the attributes in the event definition the number of distinct events in theevent log is 230 Also we did not filter patientsrsquo traces to avoid missing values inthe eventsrsquo attributes and decided instead to use the whole data set as input to thecomputation model

To this event log we applied the guideline written in PWL language which is made of16 statuses (circles in Figure 3) and 15 triggers (boxes in Figure 3) These statuses andtriggers define the computer-interpretable version of the guideline and can be described interms of the four horizontal layers of white boxes in Figure 3

bull First layer it is the definition of the conditions to enter in one of the branches of thePWL guideline This has to do with the value of the ldquostaging crdquo attributes asexplained in greater detail in the ldquoResultsrdquo section

bull Second layer it is the representation of the possible types of radiotherapy treatmentprescribed in the clinical pathway Indeed depending on the dose value the patientflows in the ldquoshort courserdquo branch (dose equal to 25 gray) or in the ldquolong courserdquo(dose equal to or greater than 45 gray)

bull Third layer it is the definition of the conditions on the concurrent chempotherapytreatment for which this clinical pathway prescribes a well-defined combination ofchemotherapy agents 5-Flouroulacil and Capecitabine

Event type Occurrences Value type Different values No of missing values

staging c 3241 (TNM) 91 5nad rt 1129 Total radiation dose (gray) 30 9nad ct 1051 Chemotherapy agent type 12 37surgery 1649 Type of surgery 6 25Notes For instance event type ldquostaging crdquo when joined with its attribute has the form ldquostaging c value Tvalue N value Mrdquo There are 91 such different combination of event type and event value for staging cLikewise for radiotherapy dose the 30 possible combinations are of the form ldquonad rt total radiation doserdquo(for instance ldquonad rt 45rdquo) The number of missing values in the attributes for each event type is also reported(for instance ldquonad rt NArdquo)

Table ICharacteristics of theevent log in terms ofrelevant events typesand event values

2178

MD5610

BE

GIN

No

17

74

in p

ath

BN

o 9

51in

pat

h C

No

141

in p

ath

AN

o 2

30

attiv

a pa

th A

(T

3 N

0)N

o 2

30

is n

ad R

T d

ose

25 G

ray

No

0

nad

RT

dos

e is

25

Gra

yN

o 0

no c

hem

io p

ath

A

No

0ch

emo

is fl

uoro

OR

cap

ecit

pat

h A

N

o 8

3ch

emo

is fl

uoro

OR

cap

ecit

pat

h B

N

o 6

23ch

emo

is fl

uoro

OR

cap

ecit

pat

h C

N

o 1

00

wai

ting

for

surg

ery

B1

2N

o 6

23w

aitin

g fo

r su

rger

y C

No

100

wai

ting

for

surg

ery

A1

2N

o 8

3w

aitin

g fo

r su

rger

y A

3N

o 0

is s

urge

ry p

erfo

rmed

pat

h A

3

No

0

surg

ery

perf

orm

ed p

ath

A3

No

0su

rger

y pe

rfor

med

pat

hA

12

No

55

surg

ery

perf

orm

ed p

ath

B1

2N

o 3

70su

rger

y pe

rfor

med

pat

hC

No

36

is s

urge

ry p

erfo

rmed

pat

h A

12

N

o 5

5

is s

urge

ry p

erfo

rmed

pat

h B

12

N

o 3

70is

sur

gery

per

form

ed p

ath

C

No

36

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

A

No

123

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th A

No

123

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

B

No

623

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th B

No

623

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

C

No

100

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th C

No

100

attiv

a pa

th B

(T

1-4

N1-

2)N

o 9

51at

tiva

path

C (

T4

N0-

3 T

1-4

N3)

No

141

Not

es T

he g

reen

circ

le o

n th

e to

p is

the

entry

poi

nt in

the

clin

ical

pat

hway

lab

eled

as ldquo

begi

nrdquo

and

cont

ains

17

74 p

atie

nts

From

ther

est

art t

he th

ree

bran

ches

A (l

eft)

B (m

iddl

e) a

nd C

(rig

ht)

whi

ch c

onta

in 2

30 9

51 a

nd 1

41 p

atie

nts

resp

ectiv

ely

The

box

es in

bet

wee

nth

e st

ates

repr

esen

t the

trig

gers

that

act

ivat

e th

e re

spec

tive

trans

ition

Figure 3Result of theconformance

checking computation

2179


bull Fourth layer it defines the conditions on surgery after the chemo-radiotherapytreatment According to the prescription of this clinical pathway it has to occurbetween 28 and 70 days after the end of the treatment and the surgical procedure hasto be ldquoAnterior resectionrdquo

The complete PWL language guideline with statuses and triggers definition in xml format canbe found in the supplementary materials of this paper An example of a trigger that defines atransition between two statuses in the PWL language for this guideline is shown below

As presented above the trigger tag has a name attribute and envelopes the condition tagin which the logical expression that allows the transition is encoded it also envelopes theset and unset tags that define the two statuses between which the transition happens(in general set and unset can refer to more than one status) This example refers to thetransition in the B branch of the clinical pathway from the end of the treatment (the patientin state ldquowaiting for surgery B12rdquo) to the surgery execution (an event surgery is found inthe patientrsquos trace) Also a temporal condition is defined the new state of the patient is setto surgery performed if the time spent in the ldquowaiting for surgery staterdquo is greater than28 days and smaller than 70 days These temporal conditions are encoded in the structureldquoafmtdrdquo (active for more than days) and ldquoafltdrdquo (active for less than days) of the PWLWe then ran this clinical pathway against the event log and the result is a pseudo-workflow graph which shows the number of patients that flow from a state to the nextaccording to the respective triggering condition Also a final xml log of the computation isgenerated which can be queried to extract the id of in and out patients at each transitionas well as the time elapsed in the transition between any two connected points of theclinical pathway These data were used to investigate the adherence of the executedprocess to the prescribed one to have a general idea of the most critical points of theclinical pathway and check where the discrepancies came from compare the branches ofthe clinical pathway in terms of waiting time from the end of treatment to the surgery(Kolmogorov-Smirnov test) compare the overall execution time for the three differentbranches (log-rank test) and compare the branches of the clinical pathway in terms of theclinical response to the treatment ( χ2-test)

4 ResultsOut of the 3229 patients in the event log 1774 patients met the clinical staging conditionfor one of the three clinical pathwayrsquos entry points (green circle in Figure 3) 230(13 percent) of which had a clinical staging fitting the branch A of the clinical pathway(clinical staging T3N0) 951 (54 percent) fitting the branch B (clinical staging any Texcept 4 N1 or N2) and 141 (8 percent) fitting the branch C (clinical staging T4) inFigure 3 these information are to be found in the three upper circles called respectivelyldquopath Ardquo ldquopath Brdquo and ldquopath Crdquo Moving downwards in the graph the flow of patientsinside the clinical pathway is readable in the same way no patients had the radiotherapydose compatible with the ldquoshort courserdquo treatment defined in the branch A3 (the one onthe left) while 123 patients entered the ldquolong courserdquo branch of the path A (53 percent oftotal path A patients) 623 the ldquolong courserdquo branch of the path B (65 percent of total

2180

MD5610

path B patients) and 100 the ldquolong courserdquo path of the branch C (71 percent of total path Cpatients) Below that there is the level of chemotherapy agents which complete theneoadjuvant treatment and make the patients ready for surgery we find 83 patients(67 percent) ready for surgery in path A 623 (100 percent) in path B and 100 (100 percent)in path C The last condition involves both the surgery execution and time betweenthe end of neoadjuvant treatment and surgery execution The overall result of theconformance checking with this clinical pathway is that 55 of path A patients had surgeryperformed in the proper timespan (66 percent of those who were waiting and 24 percent ofthose who entered path A in the first place) for path B and path C the analogous resultsare respectively 370 (59 percent and 38 percent of the total path B patients) and 36(36 percent and 25 percent of the total path C patients)

5 DiscussionFirst we noted that a relevant number of patients that are eligible for the clinical pathwaydo not go all the way through to the last state in the computation model This is due toseveral causes such as a sub-optimal imputation of data resulting for instance in missingvalues in the type of chemotherapy agent or in the value of radiotherapy dose whoselevels are reported in Table I If this is the case it might be a clue that the data entryworkflow needs to be monitored closer From this perspective the presented frameworkcan also be exploited to check the quality of the data in the EHR Another possibleexplanation is that the evidence-based clinical guideline has stricter or slightly differentconditions than the executed clinical pathways and this is something that needs to beinvestigated further with the hospital managers in order to figure out the sources of thisdiscrepancy and decide whether the clinical pathway needs to be extended to capture thereal process behavior This second case is also remarkable because it brings out thecoverage of the given clinical guideline with respect the clinical staging of the patients inthe care unit

Here we are interested in presenting three possible types of data analysis that giventhese results can be helpful to compare different branches of the executed clinical pathwaywith respect to temporal and outcome parameters

51 Waiting time for surgeryThis allows to check if the waiting times between the end of treatment and the surgery aresimilar or significantly different across the three branches of the clinical pathway If asignificant difference is found it can be a clue for instance that patients in a particularbranch of the clinical pathway are reserved higher priority in surgery-room allocationand therefore it is possible to check the conformity of this evidence to the hospital policy onthe matter

In order to do this analysis we extracted the time from waiting for surgery to surgeryperformed for the 55 patients which underwent this transition in path A for the 370 inpath B and for the 36 in path C A comparison of these transition time distributions isshown in Figure 4

The median waiting time is 58 days for path A 57 days for path B and 59 days forpath C which account for a skew toward higher time values with respect to center of thetime range allowed by the clinical pathway (between 28 and 70 days) The two sampleKolmogorov-Smirnov test confirms that the curves are similar in the sense that are verylikely to come from the same statistical distribution with the resulting p-values in Table IIWe can conclude that the waiting time for surgery is equally distributed in the threebranches of the clinical pathway and no major anomalies are detected and they areconsistent with the ongoing recommendations

2181


52 Overall execution timeAnother way to look at transition times along the clinical pathway and to compare them isthrough time-to-event analysis and Kaplan-Meier curves In this case we supposed that thegoal was to check if the overall time behavior from the entry point to the end state of the clinicalpathway (in this case surgery execution) was significantly different for the three paths A Band C In order to build the Kaplan-Meier curves we defined an ldquoeventrdquowhen a patient reachesthe surgery performed state and the event time the time between the occurrence of the eventand the entrance in the guideline (which is the staging date) Also we defined a censoring onthose patients that enter the computation model but did not reach the surgery performed stateand the relative censoring time is the time of the furthest state they get to In Figure 5 theresulting Kaplan-Meier curves are shown for the three different paths in the guidelineA log-rank test was performed to asses statistical difference among the curves which resultedin rejecting the null hypothesis of no difference with a p-value o0001 The implication of thisevidence since we already know that the waiting time for surgery does not differ significantly

Distributions Two sample KS p-value

Path A path B 08549Path A path C 07157Path B path C 04126Note Null hypothesis they come from the same statistical distribution

Table IITwo sample KS testfor the three pairsof distributions

005

004

003

002

001

000

30

Den

sity

40 50 60 70 80

Path APath BPath C

Notes The time unit is days (x-axis) The two-sample Kolmogorov-Smirnov test p-values are 085 for path A and path B 071 for path Aand path C and 041 for path B and path C

Figure 4Time distribution forsurgery waiting timein the three pathways

2182

MD5610

in the three groups is probably to be found in the different percentage of censoring and that inturn can be investigated further by for instance analyzing if the higher rate of censoring is dueto an higher rate of toxicities during treatment for a particular group of patients

53 Outcome analysisAnother use case involves the comparison of clinical pathway branches with respect to aclinical outcome measuring treatment efficacy As such a value we used the TumorRegression Grade (TRG) which can have values in the set 12345 where lower valuesmean better response and TRGfrac14 1 is complete tumor regression We are interested inchecking how this indicator is distributed in the three groups of patients who reached thefinal state as defined in the computer-interpretable guideline Table III shows the occurrence

Panel A received surgerymdashχ2-test po0001path A path B path C

No 17 186 18Yes 33 162 14

Panel B all patientsmdashχ2-test pfrac14 003path A path B path C

No 76 408 68Yes 109 369 39Note χ2 H0 populations are not different with respect to the clinical outcome TRG

Table IIITRGfrac14 12

(labeled ldquoYesrdquo) andTRGfrac14 345(labeled ldquoNordquo)distribution for

the 3 paths

10

08

06

04

02

00

0 100 200 300 400

Path APath BPath C

Notes The x-axis represents time (days) while the y-axisrepresents the percentage of patients reaching the final statesLog-rank test was performed to asses statistical difference amongthe curves which resulted in rejecting the null hypothesis of nodifference with a p-value lt0001

Figure 5Kaplan-Meier for the

three paths from entrypoint to surgery

execution

2183


of TRG frac14 1 or 2 (labeled as ldquoyesrdquo) in the three groups of patients The χ2-test and its p-valuesuggest that the occurrence of the clinical indicator is related to the typology of clinicalpathway the patients were in entering the modeling Indeed from the data in Table III(Panel A) we can see that the proportion of patients which had tumor regression comparedto the negative cases among those who received surgery in path A is roughly two to onewhereas for paths B and C the ratio is respectively 87 and 77 percent This statisticalevidence confirms what is observed in the clinical practice that N0 patients have in generalhigher TGR rates than N1 or N2 patients

Coherently with the goal of merging clinical and management needs in an integratedplatform it is worth to point out that since the data are directly exported from thedepartment EHR it is almost straightforward that the results and the analysis of theresults are made available through a real-time dashboard whose widgets allow the usermanager or clinician that be to monitor the performance indicators and to checkvariations of the health care services provide depending on the management strategiesthat will be adopted

Future work To enhance further the potential of this methodology of clinical pathwayanalysis for management-oriented information extraction some developments and newdashboard tools should be thought of Here is a brief summary of the main ones that shouldbe proposed to the health care management to help them take the route towards evidence-based decision making

(1) The conformance checking itself will be done considering both hard and softconstraint so to allow a kind of fuzzy indicator of conformance instead of the binaryin or out This way a degree of conformance can be defined

(2) Adding the information about costs in the event log will lead to a monitoring systemof financial resources to balance costs and benefits in a quantitative way

(3) Develop software agents to alert a user if the performances of the care unit are goingunder a wished threshold if a patient (or a group of patients) is moving towardstatistically dangerous pathways or if the current trend let estimate a future criticalworkload for some resources

6 ConclusionsIn this paper we described a methodology to deal with conformance checking through theimplementation of CIGs and we showed an application to a real-world event log through aclinical pathway Also some use cases were presented in which the results of conformancechecking were used to compare different branches of the executed guideline with respect toadherence to ideal process temporal distribution of state-to-state transitions and overalltreatment efficacy In particular it was shown how many patients flew from the entry pointsto the end of the guideline and how many exited at each step also time behavior and clinicalefficacy of the different paths were analyzed and compared in a quantitative way in order tocheck substantial differences among them and to extract data-driven evidences that couldbe of interest for the hospital management

References

Adriansyah A (2014) ldquoAligning observed and modeled behaviorrdquo PhD thesis Technische UniversiteitEindhoven Eindhoven (cit on pp 18 21 27 61 78 87-91 116 179 182)

Bose JRPC and van der Aalst WMP (2012) ldquoProcess diagnostics using trace alignmentopportunities issues and challengesrdquo Information Systems Vol 37 No 2 pp 117-141 available athttpsdoiorg101016jis201108003

2184

MD5610

Dwyer AJ Becker G Hawkins C McKenzie L and Wells M (2012) ldquoEngaging medical staff inclinical governance introducing new technologies and clinical practice into public hospitalsrdquoAustralian Health Review Vol 36 pp 43-48

Field MJ and Lohr KN (Eds) (1990) Clinical Practice Guidelines Directions for a New ProgramNational Academy Press Washington DC

Gatta R Lenkowicz J Vallati M Rojas E Damiani A Sacchi L De Bari B Dagliati AFernandez-Llatas C Montesi M Marchetti A Castellano M and Valentini V (2017)ldquopMineR an innovative R library for performing process mining in medicinerdquo in ten Teije APopow C Sacchi L and Holmes JH (Eds) Proceedings of the 16th Conference on ArtificialIntelligence in Medicine (AIME) ISBN 978-3-319-59758-4 Springer International PublishingBasel pp 351-355

Guumlnther CW and Rozinat A (2012) ldquoDisco discover your processesrdquo in Lohmann N and Moser S(Eds) BPM (Demos) CEUR-WSorg Tallin pp 40-44

Hripcsak G (1994) ldquoWriting Arden syntax medical logic modulesrdquo Computers in Biology andMedicine Vol 24 No 5 pp 331-363

Kurniati AP Johnson O Hogg D and Hall G (2009) ldquoProcess mining in oncology a literaturereview information communication and management (ICICM)rdquo International Conference IEEEOctober 29 2016 pp 291-297

Peleg M (2013) ldquoComputer-interpretable clinical guidelines a methodological reviewrdquo Journal ofBiomedical Informatics Vol 46 No 4 pp 744-763


Ravaghi H Heidarpour P Mohseni M and Rafiei S (2013) ldquoSenior managerrsquos viewpoints towardchallenges of implementing clinical governance a national study in Iranrdquo International Journalof Health Policy and Management Vol 1 No 4 pp 295-299

Rojas E Munoz-Gama J Sepuacutelveda M and Capurro D (2016) ldquoProcess mining in healthcarea literature reviewrdquo Journal of Biomedical Informatics Vol 61 June pp 224-236

Shahar Y Miksch S and Johnson P (1998) ldquoThe Asgaard project a task-specific framework for theapplication and critiquing of time-oriented clinical guidelinesrdquo Artificial Intelligence in MedicineVol 14 Nos 12 pp 29-51

Trabacchi V Pasquarella C and Signorelli C (2008) ldquoEvolution and practical application of theconcept of clinical governance in Italyrdquo Annali Di Igiene Vol 20 No 5 pp 509-515

Travaglia JF Debono D Spigelman AD and Braithwaite J (2011) ldquoClinical governance a review ofkey concepts in the literaturerdquo Clinical Governance An International Journal Vol 16 No 1pp 62-77

Valentini V Anti M Barbaro B Cellini F Coco C Corsi DC Cosimelli M DrsquoAprile M Doglietto GBFabiano A Ferri M Garufi C Gentile PC Laghi A Osti MF and Vecchio FM (2012)ldquoCriteri di appropriatezza clinica ed organizzativa nella diagnosi terapia e follow-up delle neoplasiedel rettordquo Rete oncologica Lazio Criteri di Appropriatezza Diagnostico Terapeutici pp 133-151available at httpsfrancescocognettifileswordpresscom201203impaginatopdf

van der Aalst W (2016a) Process Mining Data Science in Action Springer Berlin

van der Aalst W (2016b) Process Mining Discovery Conformance and Enhancement of BusinessProcesses Springer Berlin

van Dongen BF de Medeiros AKA Verbeek HMW Weijters AJMM and van der Aalst WMP(2005) ldquoThe prom framework a new era in process mining tool supportrdquo in Ciardo G andDarondeau P (Eds) Applications and Theory of Petri Nets 2005 Vol 3536 Lecture Notes inComputer Science Springer Berlin and Heidelberg pp 444-454


2185


Wang D Peleg M Tu S Boxwala A Greenes R Patel V and Shortliffe E (2002) ldquoRepresentationprimitives process models and patient data in computer-interpretable clinical practiceguidelinesrdquo International Journal of Medical Informatics Vol 68 Nos 13 pp 59-70

Woolf S Grol R Hutchinson A Eccles M and Grimshaw J (1999) ldquoClinical guidelinespotential benefits limitations and harms of clinical guidelinesrdquo BMJ Vol 318 No 7182pp 527-530

Corresponding authorCarlotta Masciocchi can be contacted at carlottamasciocchiunicattit


2186

MD5610

An integrated approach toevaluate the risk of adverseevents in hospital sector

From theory to practiceMiguel Angel Ortiz-Barrios Zulmeira Herrera-Fontalvo

Javier Ruacutea-Muntildeoz and Saimon Ojeda-GutieacuterrezDepartment of Industrial Engineering

Universidad de la Costa CUC Barranquilla ColombiaFabio De Felice

Department of Civil and Mechanical EngineeringUniversity of Cassino and Southern Lazio Cassino Italy and

Antonella PetrilloDepartment of Engineering University of Napoli ldquoParthenoperdquo Napoli Italy

AbstractPurpose ndash The risk of adverse events in a hospital evaluation is an important process in healthcaremanagement It involves several technical social and economical aspects The purpose of this paper is topropose an integrated approach to evaluate the risk of adverse events in the hospital sectorDesignmethodologyapproach ndash This paper aims to provide a decision-making framework to evaluatehospital service Three well-known methods are applied More specifically are proposed the followingmethods analytic hierarchy process (AHP) a structured technique for organizing and analyzing complexdecisions based on mathematics and psychology developed by Thomas L Saaty in the 1970s decision-making trial and evaluation laboratory (DEMATEL) to construct interrelations between criteriafactors andVIKOR method a commonly used multiple-criteria decision analysis technique for determining a compromisesolution and improving the quality of decision makingFindings ndash The example provided has demonstrated that the proposed approach is an effective and usefultool to assess the risk of adverse events in the hospital sector The results could help the hospital identify itshigh performance level and take appropriate measures in advance to prevent adverse events The authors canconclude that the promising results obtained in applying the AHPndashDEMATELndashVIKOR method suggest thatthe hybrid method can be used to create decision aids that it simplifies the shared decision-making processOriginalityvalue ndash This paper presents a novel approach based on the integration of AHP DEMATEL andVIKOR methods The final aim is to propose a robust methodology to overcome disadvantages associatedwith each methodKeywords AHP DEMATEL VIKOR Public health Evidence-based medicinePaper type Research paper

1 IntroductionNowadays citizens pay a lot of attention to high-quality medical care and overall servicequality performed by the hospital (Lee et al 2008)

To manage a hospital successfully the important goals are to attract and then retain asmany patients as possible by meeting potential demands of various kinds of the patients(Yoo 2005) Patient safety is considered as a fundamental critical to satisfaction inhealthcare Nevertheless there could have errors that can cause injury or death Theseerrors can be detected before occurring in healthcare services but some of them are notdetected and might cause damage to a patientrsquos health If this error brings about damage itis called an adverse event Adverse events or in other words ldquoany unintended or unexpectedincident which could have or did lead to harm for one or more patientsrdquo in hospitals



0025-1747DOI 101108MD-09-2017-0917

Received 30 September 2017Revised 20 February 2018

29 April 20184 June 2018

Accepted 21 June 2018


2187

Risk of adverseevents in

hospital sector


constitute a serious problem with grave consequences Many studies have been conductedto gain an insight into this problem An interesting study carried out by Rafter et al (2015)highlights that between 4 and 17 percent of hospital admissions are associated with anadverse event and a significant proportion of these (one- to two-thirds) are preventableUnfortunately also in Colombia adverse events are frequent and cause death in some casesThe above considerations demonstrate ldquohow important is assessing the risk of adverseevents in hospitalsrdquo in order to manage the causes of adverse events A variety of methodsexist to gather an adverse event but these do not necessarily capture the same events andthere is variability in the definition of an adverse event

In our opinion in order to solve this ldquoproblemrdquo it is necessary to promote a standardizationof knowledge and practice in healthcare organizations However the complexity of healthcaredecision-making and evidence selection make this process problematic

Developing a decision-making framework for hospital adverse events considering that thequality of care delivered within a health system depends on how well the causes of adverseevents in hospital practice critical factors are managed could be an useful tool for shareddecision making and to benchmark hospital performance Traditionally performance inhospitals has been measured using routinely reported health data Nevertheless these datafailed to identify patient safety

Thus a systematic and multi-criteria approach helps to evaluate different factorssimultaneously and to weigh the importance and correlation among the factors

In fact using multiple-criteria decision-making (MCDM) methods a compromise solutionfor a problem with conflicting criteria can be determined and can help the decision-makersto improve the problems for achieving the final decision (Wang and Pang 2011) NumerousMCDM methods have been developed and there is no best method for the MCDM problemEach method has its strengths and weaknesses Therefore in recent years researchers haveattempted to combine different methods to select the best alternative The main advantageof MCDM methods is that they can help to manage many dimensions to consider relatedelements and evaluate all possible options under variable degrees (Wang and Pang 2011)

In this respect this study addresses the two main limitations of evidence-basedmanagement (EBMgt) First past contributions only provided a complete view of EBMgtidentifying potential shortcomings and limitations of data-driven methods (Holmes et al2006 Morrell and Learmonth 2015) ldquowhilst the second limitation refers to the fact thatEBMgt contributions focus more on the techniques to evidence generation rather than to theapplication of this kind of evidence by decision-makers and hospital managers to improveoperational performancesrdquo

In response to both statements our paper presents a case study where it is evidencedthat the policy-makers used an MCDM model to first define the patient safety performanceof hospitals from the public sector in order to then design particular and focusedimprovement strategies addressing their particular weaknesses

In particular this paper aims to provide a decision-making framework to evaluate therisk of adverse events in the hospital sector of Colombia Three well-known methods areapplied More specifically the following methods are proposed analytic hierarchy process(AHP) a structured technique for organizing and analyzing complex decisions based onmathematics and psychology developed by Thomas L Saaty (1982) in the 1970s decision-making trial and evaluation laboratory (DEMATEL) to construct interrelations betweencriteriafactors (Fontela and Gabus 1974) and VIKOR method a commonly used multiple-criteria decision analysis (MCDA) technique for determining a compromise solution andimproving the quality of decision making (Opricovic and Tzeng 2002)

In the remainder of this work the characterization of the decision-making scenario isprovided in Section 2 then a literature review on the reported studies in the field isanalyzed in Section 3 a description of method is provided in Section 4 In Section 5 the

2188

MD5610

scenario under study is analyzed Section 6 describes the model verification In thissection a discussion of results is presented Finally Section 7 presents a summary ofresearch contribution and findings

2 Characterization of the decision-making context from a multiple-criteriadecision aid approachMultiple-criteria decision aid is a research field within the Decision Analysis which assistsdecision-makers to achieve a suitable compromise solution considering the presence ofseveral and conflicting criteria (Dulmin and Mininno 2003 Sadok et al 2009) To thisregard four families have been created to categorize the MCDA methods (Guitoni andMartel 1998) single methods the single synthetizing criterion approach outrankingmethods and the mixed methods

According to our aim or rather to assess the risk of adverse events by identifying andranking the causes of adverse events the next step was to select the most appropriateMCDA approach in accordance with the decision-making scenario The general approach toidentifying the decision elements involved in this project is detailed below

bull Trade-off management in this case a mixed method has been adopted since itscapability of dealing with both qualitative and quantitative variables which areusually found in the healthcare domain Moreover it has been proved to beappropriate for providing more robust realistic and reliable results which isparticularly useful for hospital managers to make effective decisions (Zavadskaset al 2016) In addition hybrid methods may be used where a compensatory firstphase limits the choice followed by a non-compensatory second stage to finally makethe decision (Linkov et al 2011)

bull Incomparability of the options considering that health insurance companies need tochoose the hospitals with the lowest risk of adverse events incomparability is notadmitted In this regard indifference and preference relations have been linked to thedeviations observed between the values of predefined performance criteria andsub-criteria in order to rank the hospitals by taking into account their distance to theideal solution (eg VIKOR does)

bull Scaling effects in order to avoid the introduction of bias and inconsistency in theconclusions of the decision-making process the scaling effects has been eliminated assuggested by several authors such as Pecchia et al (2013) and Martins et al (2016)This is particularly relevant for hospital and healthcare managers when designingfocused strategies reducing the risk of adverse events

bull Rank reversal it is hard to tell if a particular decision-making method has derived thecorrect answer or not (Garciacutea-Cascales and Lamata 2012) Thus regarding thestability of results in our project a compensatory approach based on the use of AHPDEMATEL and VIKOR is proposed In addition a real case study is analyzed inorder to validate the general results

bull Uncertainty in input data in order to avoid uncertainty in input data an integrativeapproach has been adopted Data have been derived from three types of sources first-hand data expert knowledge or pre-existing (probabilistic or deterministic) modelsOf these approaches using field observation data is in many cases straightforwardand expert elicitation has been covered by excellent reviews (Saaty and Tran 2007Zhuuml 2014 Pecchia et al 2013)

bull Weights assessment one of the main activities in several performance evaluationtechniques is the assignment of relative contribution to the criteria and sub-criteria

2189


hospital sector

(Izquierdo et al 2016) In this respect the consistency index of each judgment hasbeen calculated Additionally healthcare managers (in this case the respondents) areusually unskilled in decision-making and it is therefore necessary to find a methodeasily guiding them to define the relative priorities of the criteria and sub-criteriawhen assessing the risk of adverse events in hospitals (eg AHPndashDEMATEL does)

Considering the aforementioned aspects a mixed method well matches with the decision-making context regarding the assessment of the patient safety level in hospitals

3 Literature reviewFrom the late 1990s onwards analysts began to consider applying an evidence-based approachto the management of healthcare organizations In particular evidence-based medicine rose toprominence in the 1990s and can be understood as a movement that sought to improve clinicaloutcomes across healthcare organizations by standardizing professional decision-making(Timmermans and Berg 2003 Diaby et al 2013) The use of MCDA has become the domainof medical assessment in order to help medical staff to make better decisions in criticalcircumstances (Dolan 2008) In detail some authors proposed the use of DEMATELmethod within healthcare fields For example Li et al (2014) adopt DEMATEL method to findout the total relation of the factors in emergency management and to figure out critical successfactors Supeekit et al (2016) propose a DEMATEL-modified analytic network process(ANP) to evaluate internal hospital supply chain performance Recently Si et al (2017)identify key performance indicators for holistic hospital management with a modifiedDEMATEL approach Some other authors such as Chang (2014) proposed the use of VIKORmethod that evaluates hospital service by employ fuzzy VIKOR Buumlyuumlkoumlzkan et al (2016)provide a new perspective for web service performance of healthcare institutions with differentquality evaluation criteria for ranking their web services based on fuzzy analytic hierarchyprocess (IF AHP) and intuitionistic fuzzy Višekriterijumsko kompromisno rangiranjeResenje (IF VIKOR)

The bibliographic research has shown interesting articles written about applyingdecision support systems to medical and healthcare decision making but little has beenpublished about the complex problem of patient safety and hospital services (Liberatore andNydick 2008 De Felice and Petrillo 2015) There are even few scientific papers that proposean integrated approach to identifying critical success factors in a hospitalrsquos managementservice Given the relevance of this theme and the lack of studies this research aims toevaluate the risk of adverse events in hospitalized patients in from Colombia through anMCDM method

However selecting an appropriate MCDM approach is a critical step for evaluating therisk of adverse events In this regard it is suggested to apply a hybrid approachcomprising of more than one MCDM method since the single techniques may providedifferent results (Royendegh and Erol 2009 Zavadskas et al 2016) Besides Zavadskaset al (2016) concluded that integrating both objective and subjective measures intothe utility function is an advantage for an integrated approach over the single methodSeveral authors have employed the hybrid approaches (two or more techniques) instead ofthe single methods (eg Tzeng and Huang 2012 Labib and Read 2015 Hosseini andAl Khaled 2016)

The combination of different methods allows overcoming the limitations of severaltechniques Particularly ldquoPreference Ranking Organization Methodrdquo and ldquotechnique fororder preference by similarity to ideal solutionrdquo (TOPSIS) do not provide an explicitprocedure to allocate the relative importance of criteria and sub-criteria (Anand and Kodali2008 Behzadian et al 2010 Behzadian et al 2012 Velasquez and Hester 2013) Thereforethere may be some imprecision arbitrariness and lack of consensus regarding the weights

2190

MD5610

used in the decision-making model Concerning AHP method several authors have highlyconcerned on the ldquorank reversalrdquo phenomenon relating to the preference order changes afteran alternative is added or deleted (Wijnmalen and Wedley 2008 Wang and Luo 2009Garciacutea-Cascales and Lamata 2012 Maleki and Zahir 2013) The same drawback wasobserved in TOPSIS (Shih et al 2007 Wang and Luo 2009 Huszak and Imre 2010 Garciacutea-Cascales and Lamata 2012) data envelopment analysis (DEA) (Wu et al 2010 Guo andWu2013 Soltanifar and Shahghobadi 2014) and the ldquosimple additive weightingrdquo (Huszak andImre 2010 Shin et al 2013 Shin 2017) techniques Another limitation of DEA method isthat all outputs and inputs are assumed to be known (Velasquez and Hester 2013)Regarding ANP it has been concluded as a highly complex and time-consumingmethodology requiring rigorous calculations when assessing composite priorities it thenincreases the effort (Percin 2008 Kumar and Haleem 2015)

The novelty of the present study is based on the integration of the AHP perhaps the mostwell-known and widely used multi-criteria method with DEMATEL and VIKOR methods toidentify key success factors of hospital service in order to avoid adverse events for patientsIn particular AHP was chosen due to its capability of calculating the relative importance ofdecision elements (Saaty and Vargas 2012 Vargas 2012) In this case equal weights of bothcriteria and sub-criteria cannot be assumed due to some bias may be introduced in the MCDMmodel and they must be then properly calculated (eg as AHP does) In detail in the presentresearch AHP method is used to define the global and the local weights of criteria andsub-criteria It is true that AHP method presents some disadvantages since it is not possible toanalyze interactions between elements But at the same time a decision-making approachshould have some characteristics satisfied by the AHP among which is being simple inconstruct and does not require any inordinate specialization In other words the mainadvantage of AHP compared to its generalization or ANP is its simplicity that allows it to beused also by not experts in mathematical applications that could be involved in the in thegovernance of their organizations as outlined and validated by Professor Saaty (2013) Thusto cover the gap to define interrelations between criteria and sub-criteria the DEMATELmethod is integrated to AHP Our choice to use DEMATELmethod and not ANP is motivatedalso by the consideration that ANP is unable to single out an element and identify itsstrengths and weaknesses On the other hand DEMATEL was selected since it helpshealthcare managers to discriminate the interdependencies between the decision elements bydeploying an impact-digraph map where the dispatchers and receivers can be clearlyidentified (Tseng 2011 Govindan et al 2015) Ultimately VIKORwas considered in this studysince it provides very precise ranking results (Anojkumar et al 2014) This method focuses onranking and selecting from a set of alternatives in the presence of conflicting criteria it canhelp the decision-makers to reach a final decision as stated by Sayadi et al (2009) Rankinghospitals in accordance with their risk of adverse events (eg VIKOR does) is very informativeand useful for patients searching for safe care and healthcare authorities who need toprioritize interventions and allocate resources effectively Even though rank reversal problemmay exist in VIKOR only a low impact can be expected in the top alternative of the ranking(Ceballos et al 2017) Nevertheless both criteria and sub-criteria preferences are not explicitlyelicited in VIKOR method (Zhang and Wei 2013) In addition correlations between decisionelements are not considered (Chauhan and Vaish 2014) In this regard some studies underpinthe fact that there may exist a correlation between factors predicting adverse events(Passarelli et al 2005 Pocar et al 2010) and it should be then incorporated into the model(eg as DEMATEL does)

4 Description of the proposed frameworkThe proposed framework aims to evaluate the risk of adverse events in public hospitalsThe methodology is comprised of four phases (refer to Figure 1) First a decision-making

2191


hospital sector

group is established to set up a decision hierarchy considering the personal opinion of theexpert decision-makers and the key indicators established by the Ministry of Health andSocial Protection Then AHP is applied to calculate the criteria and sub-criteria weightsAfter this DEMATEL is implemented to map out the interrelations between criteria andsub-criteria as well as identify the receivers and dispatchers Additionally it is used toassess the strength of each influence relation In both AHP and DEMATEL methods thedecision-makers are asked to perform pairwise comparisons between the decision elementsof the hierarchy To this end VIKOR is developed to rank the hospitals from highest tolowest measure of closeness coefficient The results from ideal and worst solution are alsoincorporated into this study Finally the hospital with the lowest risk category is identifiedand improvement opportunities are provided

Figure 1 summarizes the proposed framework

5 MCDM methodsIn this section AHP DEMATEL and VIKOR procedures are described in detail Eachmethod and their applications reveal pros and cons as analyzed by Mandic et al (2015) intheir research This is the main reason for which an integration of the three methods isproposed in the present research as explained in Section 3

51 Analytic hierarchy processCriteria and sub-criteria weights are obtained by applying AHP This method enables expertsto calculate these measures by constructing a hierarchy structure decomposing a complexdecision-making problem into different levels where the highest represents the goal the

Design of the proposedmulticriteria decision-

making model

Design of data collection tools for AHP and DEMATEL

Global and local weights ofcriteria and sub-criteria

Interrelations betweencriteriasub-criteria via

applying DEMATEL

VIKOR application

START

Establish an expert decision-making group

Set up the decision-makinghierarchy

Apply AHP to calculatecriteria and sub-criteria

weights

Use DEMTEL to map outthe interrelation betweencriteria and sub-criteria

Implement VIKOR to rankthe hospitals

Determine the hospital withthe HIGHEST risk category

END

1 Review the pertinent literature2 Identity the pertinent KPIs3 Survey design for AHP- DEMATEL

Figure 1Proposed frameworkfor evaluating the riskof adverse events inpublic hospitals

2192

MD5610

middle contains the assessment criteria and the lowest includes the alternatives(Cannavacciuolo et al 2012 Lee and Kozar 2006) A detailed description of this methodcan be found below

bull Collect the pairwise comparisons for both the criteria and the sub-criteria by using asurvey In this case in spite of the widely used fundamental scale ( Joshi et al 2011 Shaikand Abdul-Kader 2013) a three-point scale has been adopted to reduce inconsistenciesand facilitate a better comprehension of the decision-making process for the experts whoare not qualified in complex mathematics or with the AHP technique (eg Wang et al2009 Pecchia et al 2013 Barrios et al 2016 Meesariganda and Ishizaka 2017) In thisregard the scale has been defined as follows 1 as ldquoequal importancerdquo 3 as ldquomoderateimportancerdquo and 5 ldquostrong importancerdquo The reciprocal values were assigned to theremaining judgments 13 if ldquoless importancerdquo and 15 if ldquomuch less importancerdquo

bull Aggregate the comparisons by applying the geometric mean formula (Srdjevic 2007Saaty 2008 Jaskowski et al 2010 Ishizaka et al 2011) as described in Equation (1)Here nrsquo represents the number of experts and aij is represents the relative importanceof the ith criterionsub-criterion compared to the jth criterionsub-criterion

Yn0kfrac141

akij

1=n

(1)

bull Organize the judgments into an ntimesn pairwise comparison matrix A for criteria(Equation (2)) and matrix B for sub-criteria (Equation (3))

A frac14

1 a12 a1na21 1 a2n an1 an2 1

26664

37775 (2)

B frac14

1 b12 b1nb21 1 b2n bn1 bn2 1

26664

37775 (3)

In Equations (2) and (3) it can be appreciated that the diagonal values in the matrices A andB are equal to 1 since ifrac14 j In case of a decision-making group aij and bij are obtained byusing the geometric mean of all the judgments associated with the comparison

bull Obtain the criteria (Equation (5)) and sub-criteria (Equation (4)) weights In this respectthe relative importance degree of each sub-criterion i compared to each of the othersub-criteria in the same criterion c is called local weight (LWc

i ) In addition determinethe relative weight of each criteria c in relation to the hierarchy goal (FWc )

LWci frac14

Qnjfrac141 bij

1=nPn

ifrac141

Qnjfrac141 bij

1=n i j frac14 1 2 n (4)

2193


hospital sector

FWci frac14

Qnjfrac141 aij

1=nPn

ifrac141

Qnjfrac141 aij

1=n i j frac14 1 2 n (5)

bull To evaluate the suitability of the paired comparisons it is necessary to calculate theconsistency ratio (CR) by performing Equation (7) Here CI is defined asthe consistency index (refer to Equation (6)) In Equation (4) lmax represents theeigenvalue and n is the matrix size In order to evaluate how much the inconsistencyis acceptable AHP calculates a CR comparing the CI vs the consistency index of arandom-like matrix (RI) A random matrix is one where the judgments have beenentered randomly and therefore it is expected to be highly inconsistent Morespecifically RI is the average CI of 500 randomly filled in matrices which provide thecalculated RI value for matrices of different sizes as explained by Saaty (2012)If CR⩽10 percent is deemed as reasonable Otherwise the matrix is categorized asinconsistent and the comparisons should be then reviewed by the decision-makers

CI frac14 lmaxnn1

(6)

CR frac14 CIRI

(7)

bull Calculate the relative importance degree of each sub-criteria i in relation to thehierarchy goal which is called global weight (GWi) in accordance with the followingequation)

GWi frac14 LWci FWc (8)

52 Decision-making trial and evaluation laboratoryDEMATEL is a MCDM technique used to visualize the structure of complex causalrelationships through matrices and impact digraphs (Li and Tzeng 2009 Shieh et al 2010Chang and Cheng 2011 Ortiz-Barrios et al 2017) A typical digraph represents acommunication network where influencing and affected criteriasub-criteria can be clearlyappreciated (Yang and Tzeng 2011) In this respect the interdependence among decisionelements and influence levels can be determined (Amiri et al 2011) The DEMATELprocedure can be described as follows

bull Collect the pairwise comparisons and generate the group-direct influence matrix Zthe expert decision-makers are asked to make paired comparisons (zij) between thecriteria or sub-criteria aiming at evaluating their interdependence To perform thesejudgments a five-point scale is used no influence (0) low influence (1) mediuminfluence (2) high influence (3) and very high influence (4) The scores are collected bya data-gathering tool and introduced in matrix Z In this case if there is a decision-making group

bull Generate the group-direct influence matrix the experts are asked to evaluate thedependence and feedback between criteriasub-criteria aiming to identify meaningfulinterrelationships For this purpose the participants based on their personal opinionindicate the direct influence that each criterionsub-criterion i has on each othercriterionsub-criterion j via applying an integer four-point scale where 0

2194

MD5610

(no influence) 1 (low influence) 2 (medium influence) 3 (high influence) and 4 (veryhigh influence) After this zij values are grouped into the Zk frac14 frac12zkijnn calledldquoindividual direct influencerdquo matrix In this arrangement the diagonal elements areequal to 0 and the paired comparisons are aggregated by using the following equation

zij frac141l

Xlkfrac141

zkij i j frac14 1 2 n (9)

bull Normalize the direct influenced matrix Z the normalized direct-relation matrixXfrac14 [xij]ntimesn can be achieved via applying the following equations

X frac14 Zs (10)

s frac14 max max1p ipn

Xnjfrac141

zij max1p ipn

Xnifrac141

zij

( ) (11)

bull Obtain the total influence matrix T based on the normalized direct-relation matrix Xthe total relation matrix Tfrac14 [tij]ntimesn can be achieved by using Equation (12) whereI represents the identity matrix

T frac14 XthornX 2thornX 3thorn frac14X1ifrac141

Xi frac14 X IXeth THORN1 (12)

bull Develop the influential relation map (IRM) By calculating D+R (prominence) and DminusR(relation) values where Rj is the sum of the jth column in total influence matrix T (referto Equation (13)) andDi represents the sum of the ith row of matrixT (refer to Equation(14)) dispatcher and receiver criteriasub-criteria can be determined If (DminusR)W0 thecriterionsub-criterion has a net influence on the other criteriasub-criteria and can begrouped into the cause set (dispatchers) In turn if (DminusR)o0 then the element is beinginfluenced by the other elements on the whole and can be categorized into the effectgroup (receivers) On the other hand D+R values indicate the strength of influencesthat are given or received by a specific criterionsub-criterion i In this regard bothD+R and DminusR values provide meaningful outputs for any decision-making process

R frac14Xnjfrac141

tij (13)

D frac14Xnifrac141

tij (14)

bull Calculate the threshold value and obtain impact-digraph map (IRM) the thresholdvalue (θ) is used to identify the significant interrelations between criteria or sub-criteria(refer to equation (15)) and filter out negligible effects In this respect if the influencelevel of a criteriasub-criteria in matrix T is higher than θ then this criterionsub-criterion is selected and included in the IRM Otherwise the interrelation will beexcluded The IRM graph can be achieved by mapping the data set (D+R DminusR)

y frac14Pn

ifrac141

Pnjfrac141 tij

n2 (15)

2195


hospital sector

53 Vlsekriterijumska Optimizacija I Kompromisno Resenje (VIKOR)VIKOR is an outranking method that is implemented to solve a discrete decision-makingproblem with non-commensurable and decision criteria (Opricovic and Tzeng 2007 Sayadiet al 2009 San Cristoacutebal 2011) In this regard this technique ranks a set of alternativesbased on the closeness to the ideal scenario (compromise solution) which is represented bypredefined decision criteria (Tong et al 2007 Shemshadi et al 2011) To do this VIKORintroduces a multi-criteria ranking index describing the closeness of each alternative to theaspired solution (Ou Yang et al 2009) In this sense VIKOR is useful to select the mostprofitable alternatives for decision-makers (Bazzazi et al 2011) The procedure of VIKOR iscomprised of the following steps

(1) A set of m alternatives denoted as P1 P2hellip Pm is defined for the MCDM problemHere each alternative Pi is described by a number of decision criteria (n) The valueof each sub-criterion SCj is represented by fij and is computed in matrix A accordingto the following equation

A frac14

P1

P2

P3

Pm

SC1 SC2 SCn

f 11 f 12 f 1nf 21 f 22 f 2nf 31 f 32 f 3n

f m1 f m2 f mn

2666666666664

3777777777775 (16)

(2) Identify the best ethf nj THORN and the worst ethfj THORN values in each sub-criterion by using thefollowing equations correspondingly

f nj frac14maxi f ij for benefit criteria

mini f ij for cost criteria

( ) i frac14 1 2 m (17)

fj frac14mini f ij for benefit criteria

maxi f ij for cost criteria

( ) i frac14 1 2 m (18)

(3) Calculate the Si and Ri values via applying Equation (19) and (20) respectively Herewj denotes the weight of the sub-criteria SCj This measure is provided by thecombined technique AHPndashDEMATEL

Si frac14Xnjfrac141

wj f nj f ij f nj fj

(19)

Ri frac14 maxjwj f nj f ij f nj fj

0

1A (20)

(4) Determine the Qi values by using Equations (21) (22) and (23) Here v (usually 05)represents the weight for the strategy of the maximum group utility whereas 1minusv

2196

MD5610

denotes the contribution of the individual regret

Qi frac14 vSiSn

SSnthorn 1veth THORNRiRn

RRn (21)

Sn frac14 miniSi S frac14 maxjSj (22)

Rn frac14 miniRi R frac14 maxjRj (23)

(5) Rank the alternatives (ie hospitals) based on Si Qi and Ri values (set an increasingorder for each value)

(6) Provide a compromise solution (P (1)) by selecting the best-ranked alternativeaccording to Qi ranking list and fulfilling the conditions below

bull Acceptable advantage (Equations (24) and (25))

Q P 2eth THORN

Q P 1eth THORN

XDQ (24)

DQ frac14 1= m1eth THORN (25)

Here Q(P (2)) is the hospital with the second position in the Qi ranking list

bull Acceptable stability in decision making the alternative (P (1)) must be also thebest in Si and Ri ranking lists

In case of one the conditions is not satisfied select one of these solutions

bull (P (1))y(P (2)) if there is no acceptable stability in decision making

bull (P (1)) (P (2))hellip (Pm) if there is no acceptable advantage Here (P (m)) is subject tothe following equation with the purpose of establishing the maximum m

Q P meth THORN

Q P 1eth THORN

oDQ (26)

6 Application of the proposed approach61 Evaluating the risk of adverse events in Colombian hospitalsStep 1 design of the multi-criteria decision-making model Considering that approximately84 percent of all the medication-related adverse events resulted in severe reactions in80 percent of all the hospitals an adverse event occurs every three to four weeksapproximately the most frequent adverse events are inpatient fall (4545 percent) andintravenous fluid infiltration (3636 percent) all the hospitals are focused on implementingonly corrective actions which implies that few efforts have been made to deploy preventionprograms diminishing the occurrence and impact of adverse events it is necessary to satisfythe Colombian regulations on patient safety eg Decree No 1011 of 2016 (this legislationestablishes the mandatory quality-assurance system for general healthcare system inColombia) Resolution No 2003 of 2014 (it defines the registration procedures and conditionsof healthcare providers in addition to the condition for the approval of healthcare services)Decree No 903 of 2014 (this normativity reads the provisions and make adjustments to thesingle system of accreditation in healthcare as a component of the mandatory

2197


hospital sector

quality-assurance system for healthcare services and defines rules for its operation ingeneral systems of social security in healthcare and occupations hazards) ResolutionNo 256 of 2016 (it reads the provisions related to the quality information system which is acomponent of the mandatory quality-assurance system for healthcare servicesAdditionally it sets performance indicators to monitor healthcare quality) DecreeNo 3518 of 2006 (this normativity creates and regulates the Public Health MonitoringSystem to provide information on the dynamic of the facts that may affect the populationhealth) Resolution No 1445 of 1996 (this legislation lays down the rules for the complianceof sanitary conditions at hospitals) Administrative Manual for Emergency services(it contains the guidelines for the effective management of healthcare services) and LondonProtocol (a document covering the research analysis and recommendation process aimingto minuciously study any adverse event) a multi-criteria decision model was developed toaddress the problem of assessing the risk of adverse events in hospitals and subsequentlyhelp healthcare managers to design and promote prevention programs for patient safety

This project was presented to the ethics committee of each participant hospital The chiefexecutive of each entity gave informed consent for participation Nonetheless as this studywas performed through interviews and patient participation was not queried no formalapproval from the committees was necessary Then the expert team was selected Theselection process of these participants began with the identification of decision-makerprofiles In this respect four types of experts were found to be meaningful for thedecision-making process physicians healthcare managers head nurses and representativesof academic sector linked to the healthcare industry

The team of experts was comprised of

bull One head nurse with a masterrsquos degree on healthcare quality and wide experience(11 years in the management of patient safety programs and committees in bothprivate and public hospital sectors) in the management and implementation ofpatient safety programs

bull One healthcare manager with a specialization in healthcare services and more thaneight years of experience in hospital managerial positions related to both public andprivate healthcare industry

bull One general physician with a masterrsquos degree in healthcare management and13 years of experience in public hospital management

bull One industrial engineer from the academic sector with extensive experience andknowledge in healthcare logistics and multi-criteria models for performance evaluationThe industrial engineer acted as a facilitator to take over the judgment process

A head nurse was considered to be a part of the expert decision-making team since she hasdesigned implemented and managed patient safety programs in different hospitals of thepublic sector hence she has significant experience to judge about the relevance andinterrelations of different criteria and sub-criteria that converge in adverse events On theother hand a healthcare manager was invited to participate in this group due to his wideknowledge and expertise regarding the metrics established by the Ministry of Health andSocial Protection to monitor and control patient safety activities Additionally a generalphysician was asked to participate as an expert due to his wide experience when addressingadverse events during the healthcare activities This is relevant to accurately identify themost influential factors in the decision-making hierarchy while setting improvementstrategies to reduce adverse events

Finally industrial engineer established the hierarchy with the support of the expert groupand gathered the paired judgments for both AHP and DEMATEL methods Each participant

2198

MD5610

had to demonstrate a wide experience on addressing adverse events in hospitals (W15 years)Furthermore the potential decision-maker had to be involved in the public healthcare sectorTo finally select the participants an analysis on ldquocurriculum vitaerdquo data was carried out withthe aid of the healthcare cluster representatives and the predefined profiles

The decision-making group identified a total of six criteria (C1 C2 C3 C4 C5 C6) and 27sub-criteria (S1 S2hellip S27) to evaluate the risk of adverse events in a hospital from the publicsector The criteria and sub-criteria were established based on the personal experience of expertsthe aforementioned regulations and considerations of the London Protocol (Cronin 2006)The experts took into account all the aforementioned patient safety regulations in order toprovide a MCDM model responding to the current needs of Colombian healthcare system

The multi-criteria hierarchy was then verified and discussed during multiple sessionswith the expert decision-making team to establish if it was accurate and comprehensibleThe final decision model is presented in Figure 2

Particularly the aforementioned criteria were labeled and described as stated in Table IAfterwards a detailed description of the sub-criteria is provided for each criterion

In ldquopatientrdquo dimension (C1) ldquoagerdquo (S1) represents the length of patientrsquos life In this regardelderly neonate and children are the patients with the highest risk of adverse events On theother hand ldquobackgroundrdquo (S2) sub-criterion refers to the set of patientsrsquo clinical historiesthat may predispose hospitals to incidents ldquoDisease complexityrdquo (S3) is also deemed inldquopatientrdquo criterion This sub-criterion considers the number of underlying diseases ofpatients treated in a particular hospital Additionally ldquopatient clinical conditionrdquo (S4) takesinto account the severity of patientsrsquo clinical conditions as a potential contributor of clinicalerrors Another matter of concern is ldquosocial and cultural aspectsrdquo (S5) where both limitingsocial and cultural beliefs can be identified and their affectations measured in order todevelop more precise improvement strategies Finally ldquopatient personalityrdquo (S6) is includedto represent the effects of emotional and mental patientsrsquo status on activating latent failures

In ldquotechnologyrdquo criterion (C2) ldquostate of medical equipmentrdquo (S7) is defined as thepercentage of medical equipment that is operating at good condition ldquoAvailability ofmedical equipmentrdquo (S8) refers to the percentage of medical equipment that are available forimmediate use Finally ldquouse of medical equipmentrdquo (S9) is described as the percentage offailures produced by an incorrect manipulation of medical devices A high contributionof these sub-criteria to the risk of adverse events may generate the need of implementingtraining programs supported by the providers and continuous monitoring in charge ofmaintenance departments

Another criterion of importance is ldquoenvironmentrdquo (C3) Herein ldquostate of theinfrastructurerdquo (S10) refers to the physical conditions of the furniture utensils andaccessories used by the hospital during the healthcare services On the other hand ldquoworkoverloadrdquo (S11) represents the times of peak demand which may increase the rates ofadverse events ldquoSpace conditionsrdquo (S12) is also deemed in this dimension In this respectS12 encompasses the lighting ventilation and noise conditions of hospitals to be evaluatedas potential root causes of patient safety incidents Another aspect of concern is ldquoshiftpatternrdquo (S13) This criterion determines how the distribution of work shifts may affect thestaff performance and consequently generate incidents Lastly ldquolabor atmosphererdquo (S14)describes the employeesrsquo perceptions regarding the work environment strongly activatingtheir errors and violations producing conditions in the workplace

Regarding ldquowork forcerdquo (C4) criterion ldquofatiguerdquo (S15) may represent a significantsource of stress among doctors nurses and support staff In this respect both mental andphysical exhaustion may affect them to perform normally and consequently generateerrors during healthcare services On the other side drowsiness (S16) determines whetherthe hospital demands are excessive and make employees experience reduced quality and

2199


hospital sector

quantity of sleep Technical and non-technical competences (S17) is another aspect ofinterest in this dimension S17 encompasses a set of generic skills ndash non-technical ndash thatare outside the formal education syllabus (Sahandri and Abdullah 2009) and thosespecific ndash technical ndash for a particular hospital job position and workplace environment

Patient (C1)

Age (S1)

Background(S2)

Diseasecomplexity (S3)

Patient clinicalcondition (S4)

Social andcultural

aspects (S5)

Patientpersonality

(S6)

State ofmedical

equipment (S7)

Availability ofmedical

equipment (S8)

Use of medicalequipment (S9)

State of theinfrastructure

(S10)

Work overload(S11)

Spaceconditions (S12)

Shift pattern(S13)

Laboratmosphere

(S14)

Fatigue (S15)

Drowsiness(S16)

Technical andnon-technicalcompetencies

(S17)

Mental andphysical state

(S18)

Attitude andmotivation (S19)

Adherence tohealthcare

protocols (S20)

Presence ofhealthcare

protocols (S21)

Clarity in theProcedures

(S22)

Informationquality (S23)

Proceduresdissemination

(S24)

Lack ofcommunication (S25)

Lack ofleadership (S26)

Lack ofmonitoring (S27)

Team work (C6)

Work methods(C5)

Work force (C4)

Environment(C3)

Technology (C2)

H1

H2

Hm

Goa

l E

valu

ate

the

risk

of a

dver

se e

vent

s in

the

hosp

ital s

ecto

r

Figure 2Multi-criteria decision-making model toevaluate the risk ofadverse events in thehospital sector

2200

MD5610

(Awang et al 2006) In this respect outdated staff with little work experience might causeactive failures during the healthcare operations On the other hand ldquomental and physicalstaterdquo (S18) measures how the contributory factors (eg stressors) may lead to a range ofphysical diseases (eg hypertension diabetes and cardiovascular conditions) and poormental health This is increasingly determinant since it negatively influences onabsenteeism and profits in addition to leading to human errors loss of concentration andpoor decision-making (World Health Organization and Funk 2005 Rajgopal 2010)Furthermore ldquoattitude and motivationrdquo (S19) represents the motivation level andcommitment of healthcare staff when treating patients In this regard significant positiveassociations have been found between staff satisfaction levels and measures of qualityimprovement and patient safety (Agyepong et al 2004 Alhassan et al 2013) Hence itcould be considered as a contributing factor to poor service quality increased labor strikeactions and patient dissatisfaction Finally ldquoadherence to service protocolsrdquo (S20) wasincluded to identify the gap between patient safety guidelines and clinical practice In thisrespect a significant difference may result in patients not receiving appropriate care andhigh risk of adverse events

The ldquoworking methodsrdquo criteria (C5) is underpinned by four sub-criteria ldquopresence ofhealthcare protocolsrdquo (S21) ldquoclarity in the proceduresrdquo (S22) ldquoinformation qualityrdquo (S23) andldquoprocedure communicationrdquo (S24) ldquoPresence of healthcare protocolsrdquo indicates whether thehospital adopts healthcare guidelines for specific patient safety circumstances This is

Criterion Sub-criteria General description of the criterion

Patient (C1) Age (S1)Background (S2)Disease complexity (S3)Patient clinical condition (S4)Social and cultural aspects (S5)Patient personality (S6)

This criterion considers the physical socialemotional and mental conditions of patients thatmay predispose hospitals to generate adverseevents during healthcare services

Technology(C2)

State of medical equipment (S7)Availability of medical equipment (S8)Use of medical equipment (S9)

It represents the status and availability of medicalequipment and information management systemssupporting the healthcare services in a publichospital

Environment(C3)

State of the infrastructure (S10)Work overload (S11)Space conditions (S12)Shift pattern (S13)Labor atmosphere (S14)

This factor involves a set of infrastructure spaceand working conditions under which the operationsof the hospital take place It is also deemed apotential cause of adverse events and must be thenminuciously analyzed to avoid future difficulties

Work force(C4)

Fatigue (S15)Drowsiness (S16)Technical and non-technicalcompetencies (S17)Mental and physical state (S18)Attitude and motivation (S19)Adherence to healthcare protocols (S20)

This criterion represents the professionalemotional physical and mental state of doctorsnurses and support staff that may increase theseverity and frequency of adverse events

Workmethods (C5)

Presence of healthcare protocols (S21)Clarity in the procedures (S22)Information quality (S23)Procedures dissemination (S24)

It evaluates how the healthcare procedures arecreated disseminated and deployed to diminishandor eliminate the risk of adverse events

Team work(C6)

Lack of communication (S25)Lack of leadership (S26)Lack of monitoring (S27)

This dimension assesses how the interdependenceand feedback flows between departments mayaffect the rates of adverse events In this regardconflicts of interests may appear and team worksshould be able to overcome obstacles

Table IDescription of criteria

2201


hospital sector

relevant to assist healthcare professionals how to act and which steps to follow for effectivepatient care (Ebben et al 2013) Another criterion of particular interest is ldquoclarity in theproceduresrdquo which involves measuring the level of understanding and comprehensionexpressed by the physicians regarding the correct implementation of medical proceduresOn the other hand ldquoinformation qualityrdquo is described as the quality of the content providedby healthcare information systems in terms of timeliness appropriateness reliabilityaccuracy and completeness Finally procedure communication is defined as the percentageof processes that are explained to the stakeholders aiming at achieving their commitmentduring the implementation period

The ldquowork teamrdquo criteria (C6) is evaluated by three sub-criteria ldquomiscommunicationrdquo(S25) ldquolack of leadershiprdquo (S26) and ldquolack of supervisionrdquo (S27) The first sub-criterionmeasures the effectiveness of communication flows into the work teams of hospitals This isrelevant when considering that miscommunication may lead to employee conflict a drop inmorale and turnover ldquoLack of leadershiprdquo considers the strength and capability ofthe supervisors and directors to make hospitals operate effectively in relation to theorganizational goals In this regard the healthcare leaders should be encouraged to guidethe workers to perform satisfactorily in order to avoid adverse events and detect potentialrisk sources Finally ldquolack of supervisionrdquo represents the ability of healthcare leaders toidentify potential adverse events aiming at diminishing the occurrence probability

Step 2 design of data collection tools for AHP and DEMATEL To efficiently make thepairwise judgments this section illustrates the data-gathering tools used for both AHP andDEMATEL techniques The main goal is to expose a simple and understandable way topresent the above-mentioned MCDM methods to the participants who are not expert inmathematical applications (eg doctors and nurses) In this regard a survey (Figure 3) wasinitially created to collect the AHP judgments between criteriasub-criteria For eachcomparison it was asked ldquoAccording to your experience how important is each criterionsub-criterion on the left with respect to the criterionsub-criterion on the right whenevaluating the risk of adverse events in hospitalsrdquo The respondents answered by using theaforementioned three-level AHP scale (as described in Sub-section 31) during a half-hourmeeting organized by the industrial engineer The scale is defined as follows 1 is assumedas ldquoequally importantrdquo 3 as ldquomoderately importantrdquo 5 ldquostrongly importantrdquo 13 ldquolessimportantrdquo and 15 ldquomuch less importantrdquo The survey scheme diminishes the inconsistencylevel and eliminates intransitive comparisons After this the resulting priorities wereaggregated by using the geometric mean (Equation (1))

Another data collection instrument was designed for DEMATEL comparisons (Figure 4)With this information both criteria and sub-criteria can be categorized as dispatchers orreceivers In this regard for each pairwise judgment it was asked ldquoAccording to yourexperience how much each criterionsub-criterion on the left affects the criterion

According to your experience how important is each criterionsub-criterion on the left with respect to the criterionsub-criterion on theright when evaluating the risk of adverse events in hospitals

Age

Age

Age

Age

Age

Is

Is

Is

Is

Is

Much less

Much less

Much less

Much less

Much less

Less

Less

Less

Less

Less

Equally

Equally

Equally

Equally

Equally

Moderately

Moderately

Moderately

Moderately

Moderately

Strongly

Strongly

Strongly

Strongly

Strongly

Important than

Important than

Important than

Important than

Important than

Background

Disease Complexity

Patient clinicalcondition

Social and culturalaspects

Patient personality

Figure 3Survey layout forAHP (patient cluster)

2202

MD5610

sub-criterion on the rightrdquo The participants from the decision-making team used the five-level scale established in Sub-section 32 to evaluate interdependence and feedback Thisprocess was then repeated until finalizing all the comparisons

Step 3 global and local weights of criteria and sub-criteria The next phase of the proposedapproach is the application of the combined AHPndashDEMATEL hybrid method As aconsequence the global (GW) and local weights (LW) of criteria and sub-criteria can bedetermined Herein the GW represents the contribution of a criterionsub-criterion to thedecision-making aim (assess the risk of adverse events in a hospital) On the other side the LWis the relative relevance of each decision element within each cluster Both weights willunderpin the definition of general policies that should be deemed by the policy-makers andhospital managers in order to improve the performance regarding patient safety Also thisinformation will be later used as input of VIKOR method where the three hospitals underanalysis as a supplement of this study will be finally ranked in accordance with their risk ofadverse events Additionally the consistency values of AHP matrices are presented todetermine whether the judgments are completely trustworthy for the decision-making process

Initially the collected pairwise comparisons in AHP technique (refer to Step 1) wereaggregated and organized into A (criteria) and B (sub-criteria) matrices correspondinglyAn illustration of AHP comparison matrix is presented in Table II

The judgments were introduced in Superdecisions reg software and the limit matrix wasachieved to obtain the GW and LW values (without interdependence) as shown in Table IIIfor both criteria and sub-criteria

The consistency values were then obtained (Table IV) to validate the reliability of thecomparisons The results demonstrated that all matrices achieved acceptable consistencyvalues (CR⩽10 percent) In this respect the data-gathering process can be considered assatisfactory and survey layout is therefore useful to reduce misunderstandings and

According to your experience how much each criterionsub-criterion on the left affects the criterionsub-criterion on the right

State ofmedical

equipment

State of medicalequipment


State ofmedical

equipment


equipment

Availability of medicalequipment



equipment

Use of medicalequipment




Has

Has

Has

Has

Has

Has

No influence

No influence

No influence

No influence

No influence

No influence

Low influence

Low influence

Low influence

Low influence

Low influence

Low influence Mediuminfluence

Mediuminfluence

Mediuminfluence

Mediuminfluence

Mediuminfluence

Mediuminfluence

High influence

High influence

High influence

High influence

High influence

High influence Very highinfluence

Very highinfluence

Very highinfluence

Very highinfluence

Very highinfluence

Very highinfluence

on

on

on

on

on

on

Figure 4Survey layout forDEMATEL (work

force cluster)

SI S2 S3 S4 S5 S6

SI 1 214 214 5 3 5S2 047 1 1 228 5 341S3 047 1 1 1 387 451S4 020 044 1 1 451 368S5 033 02 026 022 1 13256 020 029 022 027 076 1

Table IIAHP comparison

matrix for ldquopatientrdquocluster

2203


hospital sector

judgment errors On the other hand it is fully appreciated that some complex matrices(eg environment criteria and patient) presented very low CRs so that the above-mentioneddeclaration can be strongly confirmed

Even though AHP can calculate both criteria and sub-criteria weights (Saaty and Shang2011) it does not consider dependence and feedback Therefore a hybrid AHPndashDEMATELtechnique is proposed to additionally analyze influences among different factors and understandcomplex cause-and-effect relationships in the decision-making problem (Wu and Tsai 2012)

Cluster GW LW

Patient (C1) 0368Age (S1) 0130 0353Background (S2) 0076 0207Disease complexity (S3) 0068 0184Patient clinical condition (S4) 0054 0147Social and cultural aspects (S5) 0022 0060Patient personality (S6) 0018 0049Technology (C2) 0071State of medical equipment (S7) 0025 0357Availability of medical equipment (S8) 0029 0405Use of medical equipment (S9) 0017 0239Environment (C3) 012State of the infrastructure (S10) 0029 0239Work overload (S11) 0028 0231Space conditions (S12) 0023 0192Shift pattern (S13) 0026 0219Labor atmosphere (S14) 0014 0118Work force (C4) 0176Fatigue (S15) 0043 0246Drowsiness (S16) 0025 0144Technical and non-technical competences (S17) 0033 0188Mental and physical state (S18) 0020 0116Attitude and motivation (S19) 0026 0149Adherence of healthcare protocols (S20) 0028 0157Work methods (C5) 0116Presence of healthcare protocols (S21) 0022 0190Clarity in the procedures (S22) 0038 0333Information quality (S23) 0035 0309Procedures dissemination (S24) 0019 0167Team work (C6) 0149Lack of communication (S25) 0049 0332Lack of leadership (S26) 0055 0374Lack of monitoring (S27) 0043 0294

Table IIILW and GW valuesfor criteria and sub-criteria (AHP method)

Cluster CR

Criteria 0035Patient 0059Work methods 0067Work force 0066Work team 0014Environment 0014Technology 0015

Table IVConsistency valuesfor AHP matrices

2204

MD5610

This approach provides a more robust framework to create long-term improvement strategiesfor both healthcare professionals and decision-makers The ANP can simultaneously deal withlinear dependence and feedback however the assumption of equal weight for each cluster whenobtaining the weighted supermatrix is not acceptable in practical applications (Liu et al 2014Kou et al 2014)

To implement AHPndashDEMATEL the relative weights of criteria and sub-criteria on thebasis of interdependence (WFc andWGc respectively) are calculated by using Equation (27)(criteria) and Equation (28) (sub-criteria) Herein the weights derived from AHP applicationare multiplied by the normalized matrix of DEMATEL X

WGc frac14

P1

P2

P3

Py

SC1 SC2 SCz

r11 r12 r1zr21 r22 r2zr31 r32 r3z

ry1 ry2 ryz

2666666666664

3777777777775

GW 1

GW 2

GW 3

GWm

26666666664

37777777775 (27)

WFc frac14

P1

P2

P3

Py

SC1 SC2 SCz

r11 r12 r1zr21 r22 r2zr31 r32 r3z

ry1 ry2 ryz

2666666666664

3777777777775

FW 1

FW 2

FW 3

FWm

26666666664

37777777775 (28)

The normalized DEMATELmatrices are derived from the direct influenced matrix Z as statedin Equations (10) and (11) An illustration of a matrix Z is shown (refer to Table V) and itsnormalized version is presented in Table VI After thisWFc andWGc values were obtained byapplying Equation (27) and (28) respectively Table VII condenses the relative contributions ofcriteria and sub-criteria considering linear dependence and feedback relationships

To provide a deeper understanding of the decision-making hierarchy the globalcontributions of criteria have been illustrated in Figure 5

Age BackgroundDisease

complexityPatient clinical

conditionSocial and

cultural aspectsPatient

personality

Age 0 38 46 34 16 24Background 34 0 34 36 14 16Disease complexity 3 42 0 42 14 18Patient clinical condition 42 46 42 0 26 2Social and cultural aspects 1 2 18 2 0 14Patient personality 14 22 16 26 14 0

Table VDirect influencedmatrix (Patient

cluster)

2205


hospital sector

In accordance with AHPndashDEMATEL results ldquowork methodsrdquo was the criterion with thehighest relative contribution (FWfrac14 198 percent) However the difference between ldquoworkmethodsrdquo (first place) and patient (seventh place) is not significant (78 percent) whichevidences that all the factors should be simultaneously considered to develop clinicalimprovement strategies preventing injuries or reducing their severity It will be thereforenecessary to create an integrated clinical risk management program involving theaforementioned factors In this regard the system surrounding patients should provide a



conditionSocial and


personality


Table VINormalized directinfluenced matrix(patient cluster)

Cluster GW LW


Table VIILW and GW values ofcriteria and sub-criteria (AHPndashDEMATEL method)

2206

MD5610

safety net for potential complications resulting in prolonged hospital stay disability at thetime of discharge or death

Regarding ldquopatientrdquo cluster (Figure 6(a)) the most relevant sub-criteria was ldquopatient clinicalconditionrdquo (249 percent) Hence risk managers have to properly explore the patient statuswhen accessing healthcare services This knowledge may lead to determining whether anadverse event may occur due to patient incidence Based on this statement patients with verycomplex clinical condition have substantial risks of both poor outcomes and adverse events(Hayward and Hofer 2001 Forster et al 2008) In this regard patients play an increasinglyimportant role in the prevention of clinical incidents and the reduction of non-quality costs

In ldquotechnologyrdquo cluster (Figure 6(b)) the most significant element was ldquouse of medicalequipmentrdquo (414 percent) From this result it can be said that the contributions ofinappropriate use of technology to increasing error rates are high Particularly this is evensharper in surgical specialties of vascular surgery cardiac surgery and neurosurgery(Donaldson et al 2000) This evidences that while technology has the potential to improvemedical care it is not without risks Furthermore some experts warned of the introductionof yet-to-be errors after the adoption of new medical equipment (Hughes 2008) In thisrespect difficulties may emerge considering the poor attention paid by nurses to theimplementation of new technology settings and its role in healthcare services

Considering ldquoenvironmentrdquo dimension (Figure 7(a)) ldquostate of infrastructurerdquo represented248 percent of this criterion Nevertheless the gap between this sub-criterion and ldquoshiftpatternrdquo (163 percent) is just 85 percent which demonstrates that all the environment-related elements should be concurrently taken into consideration to avoid the fact that asubstantial number of patients experience adverse events in hospitals In this respect the

Global weights of criteria when assessing the risk adverse events in thehospital sector

2001801601401201008060402000

Workmethods

Technology Team work Environment Work force Patient

Criterion

Glo

bal w

eigh

t

198179

171 167 165

120

Figure 5GW values of criteriato evaluate the risk ofadverse events in the

hospital sector

Patient personality118

Disease complexity192 Age

157

Background184


414


270


317

Clinical conditionof the patient

249Social and

cultural aspects99

Notes (a) Patient (b) technology

(a) (b)

Figure 6LW values

2207


hospital sector

work environment has been recognized as a contributor to the occurrence of adverse eventsand medical errors (Rasmussen et al 2014) and work-related stress has been found as highlyassociated with this problem (Wrenn et al 2010) Hence work environmental conditionsmust be monitored by risk managers who should verify the unpredictable and shiftingworking conditions in healthcare departments Furthermore special attention must be paidto specialists who have been reported as the cause with the highest risk of adverse eventsSumming up a transformation of the medical environment is highly required with basis onan organizational wide-approach where all healthcare professionals are committed toachieving the desired results of maximum safety

Regarding ldquowork forcerdquo criterion (Figure 7(b)) ldquoattitude and motivationrdquo (202 percent) andldquomental and physical staterdquo (200 percent) were the most crucial sub-criteria Hereinnon-significant differences were also found and therefore it is suggested considering all thedecision elements to create multi-criteria improvement strategies for better performancerelated to both physicians and medical staff In this regard special focus must be given todistractions and interruptions which may precede skill-based errors especially divertingattention and forgetfulness (Barton 2009) Additionally it should be noted that the decisionsmade by both doctors and nurses are associated with the availability of essential informationworkload and barriers to information Hence these aspects have to be rigorously reviewed toavoid adverse events On the other hand mistakes violations and incompetence may evidenceinsufficient training and inadequate experience therefore human resources departments mustdesign appropriate competence schemes to reduce the effects of whatever human error occursThis is even more relevant when considering this factor as the most representative for thisparticular study

In ldquowork methodsrdquo cluster (Figure 8(a)) procedures dissemination (269 percent) was themost representative element However no significant difference was found between this sub-criterion and ldquoclarity in the proceduresrdquo (214 percent) which was considered as the least

Laboratmosphere

205

Spaceconditions

219

Shift pattern163

Work overload165


200

State ofinfrastructure

248

Attitude andmotivation

202

Adherence tohealthcare protocols

132

Fatigue150

Drowsiness177

Technical andNon-technicalCompetencies

139

Notes (a) Environment (b) work force

(a) (b)

Figure 7LW values

ProceduresDissemination

263

Informationquality255

Clarity in theprocedures

214

Presence ofhealthcareprotocols

268

Lack ofmonitoring

359

Lack ofLeadership

329

Lack ofCommunication

312

Notes (a) Work methods (b) team work

(a) (b)

Figure 8LW values

2208

MD5610

significant aspect From these results it is evident the need of providing a completemulti-criteria framework to ameliorate the gap between healthcare protocols and clinicalpractice which might result in patients not receiving safe care In this respect it is useful tooffer concise and clear instructions on how to provide consistent medical services effectivelyAdditionally in an effort to take a lead in promoting patient safety it will be essential toenable clinicians to be aware of protocols and checklists through improved standardizationand communication In this respect healthcare managers will also have to designate a safetychampion in every departmentcare unit so that organizationrsquos commitment can be furtherevidenced and patient safety policies deployed and efficiently disseminated in clinical practiceThereby conditions for safe medical care can be greatly enhanced

On the other hand in ldquoteam workrdquo criterion (Figure 8(b)) a similar behavior can beobserved with little differences between ldquolack of monitoringrdquo (359 percent) and ldquolack ofcommunicationrdquo (312 percent) Therefore the risk managers will have to focus onimproving both team collaboration and professional communication channels to diminishpotential medical errors and the subsequent implications on patientsrsquo safety (eg severeinjury and unexpected death) Particularly when clinicians are not communicatingeffectively medical errors may occur due to the lack of critical information and unclearorders (OrsquoDaniel and Rosenstein 2008) Thus healthcare leaders play a key role to promote acommon aim (eg reduce adverse events) and carry out plans for patient safety With this inmind the decision-makers will have to monitor the progress of these strategies in order toensure their correct deployment in healthcare services To do this process-of-care measuresshould be incorporated and process-improvement techniques adapted aiming to identifyinefficiencies and preventable errors so that team work can effectively act in accordancewith the organization goals and international standards of patient safety

As next step a comparative analysis between AHP and AHPndashDEMATEL was carriedout to identify changes in the GW values of criteria (Figure 9) and sub-criteria (Figure 10)

Regarding the overall importance of the criteria the most significant change wasobserved in C1 (patient) with a difference value of minus02468 The result is largely explainedby the DminusR (minus03350) and D+R measures (66763) through which this factor was stronglycategorized as a receiver Other meaningful differences can be appreciated in C2

C6

C5

C4

C3

C2

C1

0 005 01 015 02 025 03 035 04

Global weight (GW)

01710149

01980116

01650176

0167012

01790071

0120368

Crit

erio

n

GW_AHP-DEMATEL GW_AHP

Figure 9Comparison between

AHP and AHPndashDEMATEL (GWvalues of criteria)

2209


hospital sector

(technology) and C5 (work methods) criteria with 0108 and 0082 respectively Both criteriawere qualified as dispatchers with DminusRfrac14 03675 D+Rfrac14 64093 in C2 and DminusRfrac14 00831D+Rfrac14 72216 for C5 criterion From these results a substantial impact on other decisionelements could be further evidenced which underpins the increase in the relativecontribution of these criteria with respect to the goal

In accordance with the results provided in Figure 10 all the GW scores were concluded tobe different when incorporating DEMATEL method Particularly a substantial decreasewas found in the sub-criteria weights S1 (age) S2 (background) S3 (disease complexity) andS4 (patient clinical condition) Herein it is important to consider the fact that the GW ofldquopatientrdquo criterion changed dramatically (as indicated above) which ended up affecting theoverall importance of these elements in the decision-making model On the contrary ameaningful increase was observed in S7 (state of medical equipment) S8 (availability ofmedical equipment) S9 (use of medical equipment) S10 (state of the infrastructure) S12(space conditions) S14 (labor atmosphere) S18 (mental and physical state) S19 (attitude andmotivation) S21 (presence of healthcare protocols) S23 (information quality) S24(procedures dissemination) and S27 (lack of monitoring) These results confirm thepresence of interrelations in the decision-making model and therefore the application ofAHPndashDEMATEL method can be considered as useful to also identify dependence andfeedback Another aspect of interest is the fact that risk managers can properly design andimplement long-term strategies to eliminate or diminish the risk of adverse events inhospitals This is a meaningful advantage of the AHPndashDEMATEL hybrid technique overthe AHP method and then is recommended for similar applications For this particularcase the safety patient managers should primarily focus on improving work methodstechnology team work environment and work force which evidences what the regulationssets (refer to Section 4) the safety patient systems must be ready to address potentialadverse events and diminish avoidable latent failures and affectations in patients

Step 4 Interrelations between criteriasub-criteria via applying DEMATEL The third stepof the proposed approach evaluates the interrelations between criteria or sub-criteria byimplementing DEMATEL technique For this purpose IRMs and influence strengthcalculations are provided to show which factors and sub-factors can be categorized into thecause (dispatcher) and effect (receiver) groups when assessing the risk of adverse events inhospitals This information offers valuable insights for healthcare decision-making andguides risk managers to the development of strategic frameworks emphasizing on reducingavoidable failures in the long term Aside from this it is fully appreciated by the healthcarecluster managers in order to define future prospects and intersectoral projects addressingpatient safety difficulties That is where external healthcare institutions may provide anopportunity to alleviate the burden faced as a result of this problem

014

012

01

008

006

004

002

0S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14 S15 S16 S17 S18 S19 S20 S21 S22 S23 S24 S25 S26 S27

Sub-criterion

Glo

bal w

eigh

t (G

W)

GW_AHP GW_AHP-DEMATEL

Figure 10Comparative analysisbetween AHP andAHPndashDEMATEL(GW values ofsub-criteria)

2210

MD5610

In order to analyze the interrelations IRMs were developed (Figures 11ndash13) First the IRMfor ldquopatientrdquo is illustrated (refer to Figure 11(a)) The threshold value for this cluster wasdefined as θfrac14 (20762562)frac14 05767 Based on this reference number age (S1) patientclinical condition (S4) and patient personality (S6) are the dispatchers on the other handbackground (S2) disease complexity (S3) and social and cultural aspects (S5) are thereceivers Based on the graph particular attention must be given to patient clinical condition(S4) since it has a strong influence (D+Rfrac14 85260) must be therefore highly considered asthe focus of improvement strategies regarding patient criterion In this regard effectiveprevention and promotion plans should be created to ensure better health status of thepopulation and consequently reduce the failures caused by patients

15

05

ndash05315 32 325 33 335 34 345 35

ndash15

ndash2

ndash1

0

15

1

0

ndash1

ndash15

ndash05

05

1

S21

S23

S24

S22

D +R D +R

DndashR

DndashR

204 2045 205 2055 206 2065 207 2075 208 2085 209

S26

S25

S27

(a) (b)


Figure 13Influential relation

map for criteria

1

08

06

04

02

0

ndash0255 6 65 7 75 8

ndash04

ndash06

DndashR

DndashR

D +RD +R

25

2

15

1

05

0

ndash05

ndash1

ndash15

12 125 13 135 14 145 15 155 16 165

S13

S14

S12

S11

S10 S16

S17

S20

S15

S18 S19

(a) (b)


Figure 12Influential-relation

map for criteria

06

04

02

ndash02

ndash04

ndash06

ndash08

04 45 5 55 6 65 7 75 8 85 9

D+R

DndashR

06

03

04

05

02

0

ndash02

ndash01

ndash04

ndash03

01

DndashR

S6

S1

S4

S3

S2

S5

214 216 218 22 222 224 226 228 23 232 234

D+R

S8

S7

S9

(a) (b)



map for criteria

2211


hospital sector

The IRM for technology is presented (Figure 11(b)) The threshold was calculated asθfrac14 (33389432)frac14 37099 by the industrial engineer with expertise on decision-makingtechniques Herein state of medical equipment (S7) is the dispatcher whilst availability ofmedical equipment (S8) and Use of medical equipment are the receivers The graph specifiesthat S7 exerts a meaningful influence on both receivers (D+Rfrac14 114019) thus maintenancedepartments must implement predictive and preventive models to ensure medicalequipment functioning according to the standards and greatly diminish the risk of adverseevents considering that technology is the factor with the highest contribution

An impact diagram was also defined for environment criterion (Figure 12(a)) Theestimated reference value was θfrac14 (16717152)frac14 06687 Thus state of the infrastructure(S10) and work overload (S11) were concluded as dispatchers meanwhile space conditions(S12) shift pattern (S13) and labor atmosphere (S14) were categorized as receivers Based onthese insights it was found that state of the infrastructure (S10) has a strong effect on mostof the sub-criteria in this cluster Hence the tasks associated with this sub-factor should beeffectively deployed through continuous investment flows and optimized maintenanceplans Additionally risk managers should incorporate knowledge from reported literature toproduce solutions which will provide a safer environment for patients

An impact map was also drawn for work force criterion (Figure 12(b)) The establishedthreshold value for this cluster was computed to be θfrac14 (43819562)frac14 12172 From this graphit can be assumed that drowsiness (S16) is the only dispatcher and the rest was qualified asreceivers This can be further explained with the map where S16 influences the rest ofsub-criteria In this respect the cornerstone of this finding lies on the fact that drowsiness hasbeen recognized as a relevant contributing factor to the active failures of patient safetysystems In this context it is important to continuously evaluate the working load and healthstatus of physicians nurses and support staff so that skills can be implemented properly

Another criterion of concern (work methods) described in Figure 13(a) was also mappedsearching for prolific areas of intervention For this purpose the threshold value wascalculated as θfrac14 (66816042)frac14 41760 The main outcomes of this analysis refer to the factthat Presence of healthcare protocols (S21) and clarity in the procedures (S22) werecategorized as dispatchers On the other side procedures dissemination (S24) andinformation quality (S23) were classified as receivers However the most relevant findingwas on S21 sub-criterion since it affects all the decision elements in ldquowork methodsrdquo clusterConsequently the healthcare managers should be able to exploit the international standardsand regulations on patient safety through better clinical management In addition it isnecessary to look for scenarios facilitating the correct deployment of these protocols so thatimplementation errors and the learning curve can be meaningfully slackened

An IRM was also constructed for ldquoteam workrdquo factor (Figure 13(b)) The adopted referencenumber for this cluster was determined as θfrac14 (31052532)frac14 34503 Consequently lack ofleadership (S26) and lack of monitoring (S27) were classified into the cause group and lack ofcommunication (S25) was categorized as part of the effect group In accordance with thediagram a special attention must be paid to S26 since it affects the others significantly This ismainly related to the effort required from healthcare supervisors to support the technicaldeployments derived from patient safety management In this regard effective solutions willbe founded on efficient team work where the leaders should guide people to gain a betterunderstanding of the system Once this happens it is possible to monitor the sources ofpotential failures and subsequently reduce the occurrence and severity of adverse events

As the primary focus of this study is to provide meaningful insights in the decision-making framework Table VIII specifies the total influence matrix T for criteria The cellshighlighted in gray indicate the significant correlations The adopted threshold value forthis matrix was θfrac14 (44811862)frac14 12448 From these results it can be noted that

2212

MD5610

meaningful correlations are concentrated in technology (C2) Work force (C4) work methods(C5) and team work (C6) Herein C2 C5 and C6 are of particular interest since they wereclassified into the cause group and should be therefore considered to reduce the risk ofadverse events in hospitals On the other hand no affectation was detected on C6 and onlyone can be seen over C2 reason why these criteria obtained the highest relation valuesFinally prominence and relation values of the criteria and sub-criteria have been enlisted inTable IX where a summary of dispatchers and receivers are also provided

Criterionsub-criterion Prominence (D+R) Relation (DminusR) Dispatcher Receiver

Patient (C1) 146653 minus08084 XAge (S1) 76611 05324 XBackground (S2) 79435 minus07073 XDisease complexity (S3) 79655 minus01931 XPatient clinical condition (S4) 85260 03946 XSocial and cultural aspects (S5) 44784 minus00542 XPatient personality (S6) 49505 00276 XTechnology (C2) 141880 08102 XState of medical equipment (S7) 114019 05487 XAvailability of medical equipment (S8) 107951 minus01947 XUse of medical equipment (S9) 114425 minus03540 XEnvironment (C3) 146616 minus02985 XState of the infrastructure (S10) 75604 07567 XWork overload (S11) 60027 01234 XSpace conditions (S12) 70939 minus01102 XShift pattern (S13) 61967 minus04867 XLabor atmosphere (S14) 65806 minus02833 XWork force (C4) 156579 minus04390 XFatigue (S15) 152161 minus02577 XDrowsiness (S16) 142838 18735 XTechnical and non-technical competences (S17) 135566 minus04473 XMental and physical state (S18) 158878 minus01126 XAttitude and motivation (S19) 160831 minus00024 XAdherence of healthcare protocols (S20) 126116 minus10534 XWork methods (C5) 161732 01012 XPresence of healthcare protocols (S21) 336785 03633 XClarity in the procedures (S22) 318680 13152 XInformation quality (S23) 346884 minus01765 XProcedures dissemination (S24) 333971 minus15020 XTeam work (C6) 142777 06345 XLack of communication (S25) 204428 minus12123 XLack of leadership (S26) 208469 12017 XLack of monitoring (S27) 208153 00105 XNote ldquoXrdquo indicates whether or not Dispatcher and Receiver have those parameters

Table IXRelation (DndashR) andprominence (D+R)

values of criteria andsub-criteria

C1 C2 C3 C4 C5 C6 D R D+R DminusR

C1 10880 10456 11905 12555 12610 10878 69285 77369 146653 minus08084C2 13097 10217 12538 13662 13883 11595 74991 66889 141880 08102C3 12509 10745 10916 13299 13192 11154 71815 74801 146616 minus02985C4 13376 11758 12737 12503 13863 11857 76094 80484 156579 minus04390C5 14411 12611 13826 14791 13325 12408 81372 80360 161732 01012C6 13096 11103 12878 13675 13486 10324 74561 68216 142777 06345R 77369 66889 74801 80484 80360 68216

Table VIIITotal influence matrix

T for criteria

2213


hospital sector

62 Ranking three Colombian hospitals according to the risk of adverse eventsStep 5 VIKOR application Complementary to this analysis VIKOR method is applied torank the three hospitals under analysis according to the risk of adverse events in order toinform patients searching for safe care (best-ranked hospitals) and healthcare authoritieswho need to prioritize interventions and allocate resources effectively The adoption ofVIKOR method extends the usability of the results (practical implications) emanating fromAHP and DEMATEL techniques and it hence contributes to the still scant evidence base onEBMgt VIKOR ranks a set of alternatives based on the proximity to the ideal scenario(compromise solution) taking into account the formulas and conditions described in theSub-section 33 For the project development three hospitals (P1 P2 and P3) from Colombianhealthcare system were selected These institutions are administrative entities with financialsustainability whose primary aim is to provide a defined set of medical services seeking forpreventing diseases and promoting healthcare Particularly P1 is a first-level hospital withsecond-level specialized healthcare with a focus on patient needs and family expectationsFurthermore it has remodeled facilities with a satisfactory layout and high-tech medicalequipment On the other hand P2 is also a first-level medical institution comprised ofqualified and service-minded human resource with a sense of belonging However it has alimited space and old-fashioned medical technology In turn P3 can be defined as a hospitalwith basic medical services provided with quality efficiency and a patient safety policyNonetheless its facilities are very old and its layout is inefficient The medical equipment isalso antiquated and failures on adverse events monitoring system can be appreciated

For the VIKOR implementation a group of indicators or key performance indexes (KPI)was defined one for each sub-criterion (refer to Table X) based on the regulationsestablished by the Ministry of Health and Social Protection The mathematical formulationfor the calculation of each KPI is also provided in Table X

After organizing the KPIs in the A matrix of VIKOR method (refer to Table XI) the besteth f nj THORN and worst eth fj THORN values for each sub-criterion were determined The sub-criteria weightswere provided by the combined AHPndashDEMATEL method

Then Si and Ri values were calculated by using Equation (19) and (20) respectively (referto Table XII) After this by applying Equations (21) (22) and (23) Qi measures weredetermined Herein Sfrac14 0148 Sminusfrac14 0581 Rfrac14 0033 Rminusfrac14 0074 and vfrac14 05 Thereby thehospitals were ranked in accordance with Si Ri and Qi values (refer to Table XIII)

Each ranking of hospitals (alternatives) is made in increasing order and the best-rankedalternative (compromise solution) is determined by corroborating two conditions(Sub-section 33) acceptable advantage and acceptable stability in decision-makingA summary of this validation is provided in Table XIV Both conditions are satisfied andtherefore P1 is the hospital with the least risk of adverse events

In order to facilitate continuous improvement on patient safety management of thehospitals under assessment the separations from the ideal scenario were illustrated inFigure 14 This is to easily identify how close each alternative is to this performance andwhich sub-criteria must be improved to reduce the overall gap (Si) In this sense it is evidentthat P1 is the closest to the ideal solution even though it is recommendable to improve inS19 (attitude and motivation) and S5 (social and cultural aspects) On the other handparticular attention must be paid to P2 since the major deviations are given in sub-criterionS7 (state of medical equipment) S8 (availability of medical equipment) S10 (state of theinfrastructure) and S12 (space conditions) where contributions to adverse events aresignificant In this regard a diagnosis should be firstly performed to determine the causes ofthese poor measures and then establish effective solutions to the problem with basison the dispatchers Finally the worst-ranked hospital (P3) presents serious difficulties inS9 (use of medical equipment) S16 (drowsiness) S18 (mental and physical state) S22 (clarity

2214

MD5610

Sub-criteria KPI Formula

Age (S1) Average age of patients Sum of the ages of the patientsTotal ofattended patients

Background (S2) of patients with one or more ofthe following clinical conditionsDiabetesHypertension

(Number of patients with diabetes andorhypertensionTotal number of attendedpatients)times100

Disease complexity (S3) of patients with complexdiseases

(Number of patients with complex diseasesTotal number of attended patients)times100

Patient clinical condition(S4)

Average stay in ICU (days) Sum of the individual stay periods in ICUTotal number of attended patients

Social and culturalaspects (S5)

Weighted average of the socialstrata

A value is assigned to each social strataLow (1)Medium (2)High (3)n1 Proportion of population in low social stratan2 Proportion of population in medium socialstratan3 Proportion of population in high socialstrataN Total populationP

n1 1eth THORNthorn n2 2eth THORNthorn n3 3eth THORN=N Patient personality (S6) of patients with psychological

intervention(Number of patients with psychologicalinterventionTotal of attended patients)times100

State of medicalequipment (S7)

of medical equipment in goodcondition

(Number of medical equipment in goodconditionNumber of medicalequipment)times100

Availability of medicalequipment (S8)

of medical equipment available (Number of medical equipment in operationNumber of medical equipment)times100

Use of medical equipment(S9)

Average month number of medicalequipment failures due to misuse

(Number of annual medical equipmentfailures due to misuse12)

State of the infrastructure(S10)

of adequate rooms (Number of adequate roomsTotal number ofrooms)times100

Work overload (S11) of workers who exceed theirworking time when performinghospital activities

(Number of workers who exceed theirworking time when performing hospitalactivitiesTotal number of workers)times100

Space conditions (S12) of failures due to lack oflighting ventilation reduced spaceor excessive noise

(Number of failures due to lack of lightingventilation reduced space or excessivenoiseTotal number of failures)times100

Shift pattern (S13) Risk level of hospital workers A 5-point scale was defined as followsClass 1 Minimum riskClass 2 Low riskClass 3 Medium riskClass 4 High riskClass 5 Maximum risk

Labor atmosphere (S14) of satisfied workers (Number of satisfied workersTotal numberof workers)times100

Fatigue (S15) Average overtime worked byemployees in a week

(Sum of overtime worked in a hospital perweekTotal number of workers)

Drowsiness (S16) Average number of employeesworking at night shift

(Sum of employees working at night timeTotal number of night shifts)

Technical and non-technical competencies(S17)

of qualified personnel (Number of professionals workersTotalnumber of workers)times100

(continued )

Table XKey performance

indexes forsub-criteria

2215


hospital sector


Mental and physical state(S18)

of workers with good physicaland mental state

(Number of workers with good physical andmental stateTotal number of workers)times100

Attitude and motivation(S19)

of workers with good attitudeand motivation level

(Number of workers with good attitude andmotivation levelTotal number ofworkers)times100

Adherence to healthcareprotocols (S20)

Proportion of monitored adverseevents

(Number of adverse events undersupervisionTotal number of adverseevents)times100

Presence of healthcareprotocols (S21)

Presence of healthcare protocols Yes (1)No (0)

Clarity in the procedures(S22)

Average medical errors per month (Sum of annual medical errors12)

Information quality (S23) of information requests met (Number of information requests metTotalnumber of received requests)times100

Procedures dissemination(S24)

of disseminated procedures (Number of disseminated proceduresTotalnumber of procedures)times100

Lack of communication(S25)

Average monthly number oferrors due to lack ofcommunication

(Sum of annual number of errors due to lackof communication12)

Lack of leadership (S26) of supervisors with leadershiptraining

(Number of supervisors with leadershiptrainingTotal number of supervisors)times100

Lack of monitoring (S27) Existence of security rounds Yes (1)No (0)Table X

Sub-criterion S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14GW 0019 0022 0023 003 0012 0014 0057 0048 0074 0041 0028 0037 0027 0034P1 386 44 60 0 15 36 95 93 1 90 17 3 3 96P2 414 56 80 0 154 43 80 85 1 60 14 5 3 97P3 448 52 70 0 154 19 91 89 2 80 10 5 3 93Best value 386 44 60 0 154 19 95 93 1 90 10 3 3 97Worst value 448 56 80 0 15 43 80 85 2 60 17 5 3 93Sub-criterion S15 S16 S17 S18 S19 S20 S21 S22 S23 S24 S25 S26 S27GW 0025 0029 0023 0033 0033 0022 0053 0042 005 0052 0053 0056 0061P1 9 7 90 85 89 100 1 3 100 70 2 92 1P2 6 9 88 90 92 100 1 2 100 62 2 90 1P3 5 11 86 75 95 100 1 4 91 54 3 88 1Best value 5 7 90 90 95 100 1 2 100 70 2 92 1Worst value 9 11 86 75 89 100 1 4 91 54 3 88 1

Table XIInitial matrix A forhospital alternatives

Sub-criterion S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14GW 0019 0022 0023 003 0012 0014 0057 0048 0074 0041 0028 0037 0027 0034P1 0 0 0 0 0012 001 0 0 0 0 0028 0 0 0009P2 0009 0022 0023 0 0 0014 0057 0048 0 0041 0016 0037 0 0P3 0019 0015 0012 0 0 0 0015 0024 0074 0014 0 0037 0 0034Sub-criterion S15 S16 S17 S18 S19 S20 S21 S22 S23 S24 S25 S26 S27 SjGW 0025 0029 0023 0033 0033 0022 0053 0042 005 0052 0053 0056 0061P1 0025 0 0 0011 0033 0 0 0021 0 0 0 0 0 0148P2 0006 0015 0012 0 0017 0 0 0 0 0026 0 0028 0 0369P3 0 0029 0023 0033 0 0 0 0042 005 0052 0053 0056 0 0581

Table XIISi and Ri values

2216

MD5610

in the procedures) S23 (information quality) S24 (procedures dissemination) S25 (lack ofcommunication) and S26 (lack of leadership) which evidences a fairly catastrophicperformance regarding the elements from the cause group (technology team work workforce and work methods) To address this problem P3 should create training programs forboth nurses and physicians in collaboration with the providers Additionally it isrecommended to monitor the effectiveness of these programs aiming to evidence theachieved results in terms of reduced number of adverse events and potential failures On the

Alternatives Si Si rank Ri Ri rank Qi (vfrac14 05) Qi rank

P1 0148 1 0033 1 0000 1P2 0369 2 0057 2 0548 2P3 0581 3 0074 3 1000 3

Table XIIISi Ri and Qi ranking

for hospitals inaccordance with theirrisk of adverse events

Condition Conclusion

C1 Acceptable advantage (0548⩾ 05) SatisfiedC2 Acceptable stability in decision making (1st place in ranking for both Si and Ri) Satisfied

Table XIVEvaluation ofconditions for

compromise solution

S26S27

S1S2

S3

S4

S5

S6

S7

S8

S9

S10

S11

S12

S13S14S15

S16

S17

S18

S19

S20

S21

S22

S23

S24

S25

006

005

004

003

002

001

0

P1 P2 P3

Figure 14Spider diagram for

separations from theideal scenario

2217


hospital sector

other side the human resources department of this P3 should evaluate the physical status ofemployees and determine whether the work load is adequate for the purpose of designingfocused improvement plans Regarding the difficulties with work methods its patientsafety department ought to carefully revise how the protocols are being documenteddeployed and disseminated since the system evidences symptoms of poor understandingand comprehension reason which dramatically increases the risk of adverse events andaffectations on patients Finally it is proposed to verify the accurateness of information flowsin work teams and the roles played by its supervisors In this case the human resourcesdepartment should work on designing coaching programswhere these details can be analyzedand improved Furthermore it is relevant to determine whether the information system ispertinent and useful for P3 hospital With these strategies communication and leadershipproblems can be effectively addressed The above-mentioned recommendations can be furtherreplicated by other hospitals with similar performance on patient safety

7 ConclusionsIn the context of healthcare the evaluation of any outcome measure involves several technicalsocial and economic aspects Thus it is necessary to take into account the relationships betweenthem At this aim the multi-criteria decision methods concur MCDM clearly may help in thematter although the large literature on the topic does not allow determining easily whichprocedure is the more appropriate Each method contains strengths and weaknesses Forexample AHP hierarchy can have as many levels as needed to fully characterize a particulardecision situation Furthermore AHP can efficiently deal with tangible as well as non-tangibleattributes But at the same time perfect consistency is very difficult to obtain with AHP or it doesnot allow to evaluating interrelations and influences between the elements that compose thedecision-making process Hence to overcome disadvantages associated with AHP an integrationusing DEMATEL method is proposed DEMATEL is used for researching and solvingcomplicated and intertwined problem groups In particular it is useful to investigateinterrelationships between the criteria for evaluating effects Finally VIKOR method is proposedto calculate the ratio of positive and negative ideal solution It proposes a compromisesolution with an advantage rate Therefore the hybrid and integrated approachAHPndashDEMATELndashVIKOR was found to provide robust realistic and reliable results whenassessing hospital patient safety level This increases the likelihood of a favorable outcomederived from the decision-making process Additionally it responds to the following facts equalweights of decision element cannot be assumed since some bias may be incorporated into theMCDMmodel and theymust be then properly estimated some studies support the fact that theremay exist correlation between criteria predicting adverse events it is relevant to inform patientssearching for safe healthcare and authorities who need to prioritize sectorial interventions andproperly allocate resources and to overcome the limitations of single MCDM methods

The example provided has demonstrated that the proposed approach is an effective anduseful tool to assess the risk of adverse events in the hospital sector The results could help thehospital identify its performance level and respond appropriately in advance to preventadverse events We can conclude that the promising results obtained in applying theAHPndashDEMATELndashVIKOR method suggest that the hybrid method can be used to createdecision aids that it simplifies the shared decision-making process Furthermore the decisionhere formulated (assessing the risk of adverse events in hospitals) has been madeconscientiously explicitly and judiciously (even searching for the best MCM methods) usedwith basis on the best available evidence (findings from literature review pairwise judgmentsfrom experts and key performance indicators) as stated by Morrell and Learmonth (2015)

It is important to acknowledge that the findings may be related to the characteristics of thestudy design Importantly the study was limited to three hospitals in Colombia which could

2218

MD5610

partially explain the VIKOR results Future research will take into account two new aspects agreater number of hospitals and different countries On the other hand a sensitivity analysisbased on Monte Carlo approach and three simulation models (random weights rank-orderweights and response distribution weights) will be developed in order to test the influence ofboth criteria and sub-criteria weights on the final ranking (Butler et al 1997)

References

Agyepong IA Anafi P Asiamah E Ansah EK Ashon DA and Narh‐Dometey C (2004) ldquoHealthworker (internal customer) satisfaction and motivation in the public sector in Ghanardquo TheInternational Journal of Health Planning and Management Vol 19 No 4 pp 319-336

Alhassan RK Spieker N van Ostenberg P Ogink A Nketiah-Amponsah E and de Wit TFR(2013) ldquoAssociation between health worker motivation and healthcare quality efforts in GhanardquoHuman Resources for Health Vol 11 No 1 pp 11-37

Amiri M Sadaghiyani J Payani N and Shafieezadeh M (2011) ldquoDeveloping a DEMATELmethod toprioritize distribution centers in supply chainrdquo Management Science Letters Vol 1 No 3pp 279-288

Anand G and Kodali R (2008) ldquoSelection of lean manufacturing systems using the PROMETHEErdquoJournal of Modelling in Management Vol 3 No 1 pp 40-70

Anojkumar L Ilangkumaran M and Sasirekha V (2014) ldquoComparative analysis of MCDM methodsfor pipe material selection in sugar industryrdquo Expert Systems with Applications Vol 41 No 6pp 2964-2980

Awang Z Abidin HZ Arshad MR Habil H and Yahya AS (2006) ldquoNon-technical skills forengineers in the 21st century a basis for developing a guidelinerdquo Faculty of Management andHuman Resource Development Universiti Teknologi Malaysia Skudai Johor

Barrios MAO De Felice F Negrete KP Romero BA Arenas AY and Petrillo A (2016)ldquoAn AHPndashTOPSIS integrated model for selecting the most appropriate tomographyequipmentrdquo International Journal of Information Technology amp Decision Making Vol 15No 4 pp 861-885

Barton A (2009) ldquoPatient safety and quality an evidence-based handbook for nursesrdquo AORN JournalVol 90 No 4 pp 601-602

Bazzazi AA Osanloo M and Karimi B (2011) ldquoDeriving preference order of open pit minesequipment through MADM methods application of modified VIKOR methodrdquo Expert Systemswith Applications Vol 38 No 3 pp 2550-2556

Behzadian M Kazemzadeh RB Albadvi A and Aghdasi M (2010) ldquoPROMETHEE acomprehensive literature review on methodologies and applicationsrdquo European Journal ofOperational Research Vol 200 No 1 pp 198-215

Behzadian M Otaghsara SK Yazdani M and Ignatius J (2012) ldquoA state-of the-art survey ofTOPSIS applicationsrdquo Expert Systems with Applications Vol 39 No 17 pp 13051-13069

Butler J Jia J and Dyer J (1997) ldquoSimulation techniques for the sensitivity analysis of multi-criteriadecision modelsrdquo European Journal of Operational Research Vol 103 No 3 pp 531-546

Buumlyuumlkoumlzkan G Feyzioglu O and Gocer F (2016) ldquoEvaluation of hospital web services usingintuitionistic fuzzy AHP and intuitionistic fuzzy VIKORrdquo IEEE International Conference onIndustrial Engineering and Engineering Management December pp 607-611

Cannavacciuolo L Iandoli L Ponsiglione C and Zollo G (2012) ldquoAn analytical framework based onAHP and activity-based costing to assess the value of competencies in production processesrdquoInternational Journal of Production Research Vol 50 No 17 pp 4877-4888

Ceballos B Pelta DA and Lamata MT (2017) ldquoRank reversal and the VIKOR method an empiricalevaluationrdquo International Journal of Information Technology amp Decision Making Vol 17 No 2pp 1-13

2219


hospital sector

Chang KH and Cheng CH (2011) ldquoEvaluating the risk of failure using the fuzzy OWA andDEMATEL methodrdquo Journal of Intelligent Manufacturing Vol 22 No 2 pp 113-129

Chang TH (2014) ldquoFuzzy VIKOR method a case study of the hospital service evaluation in TaiwanrdquoInformation Sciences Vol 271 pp 196-212

Chauhan A and Vaish R (2014) ldquoA comparative study on decision making methods with intervaldatardquo Journal of Computational Engineering Vol 1 No 1 pp 1-10

Cronin CM (2006) ldquoFive years of learning from analysis of clinical occurrences in pediatric care usingthe London Protocolrdquo Healthcare Quarterly Vol 9 pp 16-21

De Felice F and Petrillo A (2015) ldquoImproving Italian healthcare service quality using analytichierarchy process methodologyrdquo 6th European Conference of the International Federation forMedical and Biological Engineering MBEC Vol 45 Dubrovnik September 7ndash11 pp 981-984

Diaby V Campbell K and Goeree R (2013) ldquoMulti-criteria decision analysis (MCDA) in health care abibliometric analysisrdquo Operations Research for Health Care Vol 2 pp 20-24

Dolan JG (2008) ldquoShared decision-making ndash transferring research into practice the analytic hierarchyprocess (AHP)rdquo Patient Education and Counseling Vol 73 pp 418-425

Donaldson MS Corrigan JM and Kohn LT (Eds) (2000) To Err is Human Building a Safer HealthSystem Vol 6 National Academies Press Washington DC

Dulmin R and Mininno V (2003) ldquoSupplier selection using a multi-criteria decision aid methodrdquoJournal of Purchasing and Supply Management Vol 9 No 4 pp 177-187

Ebben RH Vloet LC Verhofstad MH Meijer S Mintjes-de Groot JA and van Achterberg T(2013) ldquoAdherence to guidelines and protocols in the prehospital and emergency care setting asystematic reviewrdquo Scandinavian Journal of Trauma Resuscitation and Emergency MedicineVol 21 No 1 pp 1-9

Fontela E and Gabus A (1974) ldquoDEMATEL innovative methodsrdquo Report No 2 Structural analysisof the world problematique Battelle Geneva Research Institute Geneva

Forster AJ Kyeremanteng K Hooper J Shojania KG and van Walraven C (2008) ldquoThe impact ofadverse events in the intensive care unit on hospital mortality and length of stayrdquo BMC HealthServices Research Vol 8 No 1 pp 251-259

Garciacutea-Cascales MS and Lamata MT (2012) ldquoOn rank reversal and TOPSIS methodrdquoMathematicaland Computer Modelling Vol 56 Nos 56 pp 123-132

Govindan K Khodaverdi R and Vafadarnikjoo A (2015) ldquoIntuitionistic fuzzy based DEMATELmethod for developing green practices and performances in a green supply chainrdquo ExpertSystems with Applications Vol 42 No 20 pp 7207-7220

Guitoni A and Martel JM (1998) ldquoTentative guidelines to help choosing an appropriate MCDAmethodrdquo European Journal of Operation Research Vol 109 No 2 pp 501-521

Guo D and Wu J (2013) ldquoA complete ranking of DMUs with undesirable outputs using restrictions inDEA modelsrdquo Mathematical and Computer Modelling Vol 58 Nos 56 pp 1102-1109

Hayward RA and Hofer TP (2001) ldquoEstimating hospital deaths due to medical errors preventabilityis in the eye of the reviewerrdquo JAMA Vol 286 No 4 pp 415-420

Holmes D Murray SJ Perron A and Rail G (2006) ldquoDeconstructing the evidence-based discourse inhealth sciences truth power and fascismrdquo International Journal of Evidence-Based HealthcareVol 4 No 3 pp 180-186

Hosseini S and Al Khaled A (2016) ldquoA hybrid ensemble and AHP approach for resilient supplierselectionrdquo Journal of Intelligent Manufacturing Vol 1 No 1 pp 1-22

Hughes R (Ed) (2008) Patient Safety and Quality An Evidence-Based Handbook for Nurses Vol 3Agency for Healthcare Research and Quality Rockville MD

Huszak A and Imre S (2010) ldquoEliminating rank reversal phenomenon in GRA-based networkselection methodrdquo IEEE International Conference on Communications (ICC) pp 1-6

2220

MD5610

Ishizaka A Balkenborg D and Kaplan T (2011) ldquoInfluence of aggregation and measurement scale onranking a compromise alternative in AHPrdquo Journal of the Operational Research Society Vol 62No 4 pp 700-710

Izquierdo NV Viloria A Gaitaacuten-Angulo M Bonerg O Lezama P Erase JJC and Gutieacuterrez AS(2016) ldquoMethodology of application of diffuse mathematics to performance evaluationrdquoInternational Journal of Control Theory and Applications Vol 1 No 1 pp 1-6

Jaskowski P Biruk S and Bucon R (2010) ldquoAssessing contractor selection criteria weights withfuzzy AHP method application in group decision environmentrdquo Automation in ConstructionVol 19 No 2 pp 120-126

Joshi R Banwet DK and Shankar R (2011) ldquoA DelphindashAHPndashTOPSIS based benchmarkingframework for performance improvement of a cold chainrdquo Expert Systems with ApplicationsVol 38 No 8 pp 10170-10182

Kou G Ergu D and Shang J (2014) ldquoEnhancing data consistency in decision matrix adaptingHadamard model to mitigate judgment contradictionrdquo European Journal of OperationalResearch Vol 236 No 1 pp 261-271

Kumar S and Haleem A (2015) ldquoEvaluating bullwhip effect mitigation an analytical network process(ANP) applicationrdquo International Journal of Advanced Research in Engineering Science andManagement Vol 2 No 1 pp 1-14

Labib A and Read M (2015) ldquoA hybrid model for learning from failures the Hurricane Katrinadisasterrdquo Expert Systems with Applications Vol 42 No 21 pp 7869-7881

Lee Y and Kozar KA (2006) ldquoInvestigating the effect of website quality on e-business successan analytic hierarchy process (AHP) approachrdquo Decision Support Systems Vol 42 No 3pp 1383-1401

Lee YC Yen TM and Tsai CH (2008) ldquoUsing importancendashperformance analysis and decisionmaking trial and evaluation laboratory to enhance order-winner criteria ndash a study of computerindustryrdquo Information Technology Journal Vol 7 No 3 pp 396-408

Li CW and Tzeng GH (2009) ldquoIdentification of a threshold value for the DEMATEL method usingthe maximum mean de-entropy algorithm to find critical services provided by a semiconductorintellectual property mallrdquo Expert Systems with Applications Vol 36 No 6 pp 9891-9898

Li Y Hu Y Zhang X Deng Y and Mahadevan S (2014) ldquoAn evidential DEMATEL method toidentify critical success factors in emergency managementrdquo Applied Soft Computing JournalVol 22 pp 504-510

Liberatore MJ and Nydick RL (2008) ldquoThe analytic hierarchy process in medical and health caredecision making a literature reviewrdquo European Journal of Operational Research Vol 189pp 194-207

Linkov I Bates ME Canis LJ Seager TP and Keisler JM (2011) ldquoA decision-directed approachfor prioritizing research into the impact of nanomaterials on the environment and humanhealthrdquo Nature Nanotechnology Vol 6 No 12 pp 777-784

Liu HC You JX Zhen L and Fan XJ (2014) ldquoA novel hybrid multiple criteria decision makingmodel for material selection with target-based criteriardquoMaterials amp Design Vol 60 pp 380-390

Maleki H and Zahir S (2013) ldquoA comprehensive literature review of the rank reversal phenomenon inthe analytic hierarchy processrdquo Journal of Multi-Criteria Decision Analysis Vol 20 Nos 34pp 141-155

Mandic K Bobar V and Delibašic B (2015) ldquoModeling interactions among criteria in MCDM methodsa reviewrdquo in Liu S Delibašić B and Oderanti F (Eds) First International Conference ICDSSTProceedings Conference Paper in Lecture Notes in Business Information Processing ProceedingsBelgrade doi 101007978-3-319-18533-0_9

Martins CL de Almeida JA de Oliveira Bortoluzzi MB and de Almeida AT (2016) ldquoScaling issuesin MCDM portfolio analysis with additive aggregationrdquo in Liu S Delibašić B and Oderanti F(Eds) International Conference on Decision Support System Technology Springer Champp 100-110

2221


hospital sector

Meesariganda BR and Ishizaka A (2017) ldquoMapping verbal AHP scale to numerical scale for cloudcomputing strategy selectionrdquo Applied Soft Computing Vol 53 April pp 111-118


OrsquoDaniel M and Rosenstein AH (2008) ldquoProfessional communication and team collaborationrdquoin Hughes RG (Ed) Patient Safety and Quality An Evidence-Based Handbook for NursesAgency for Healthcare Research and Quality Advances in Patient Safety Rockville MD April

Opricovic S and Tzeng GH (2007) ldquoExtended VIKOR method in comparison with outrankingmethodsrdquo European Journal of Operational Research Vol 178 No 2 pp 514-529

Opricovic S and Tzeng GH (2002) ldquoMulticriteria planning of post-earthquake sustainablereconstructionrdquo Computer-Aided Civil Infrastructure Engineering Vol 17 No 3 pp 211-220

Ortiz-Barrios MA Aleman-Romero BA Rebolledo-Rudas J Montes-Villa L De Felice F andPetrillo A (2017) ldquoThe analytic decision-making preference model to evaluate the disasterreadiness in emergency departments the ADT modelrdquo Journal of Multi-Criteria DecisionAnalysis Vol 24 Nos 56 pp 204-226

Ou Yang YP Shieh HM Leu JD and Tzeng GH (2009) ldquoA VIKOR-based multiple criteria decisionmethod for improving information security riskrdquo International Journal of InformationTechnology amp Decision Making Vol 8 No 2 pp 267-287

Passarelli MCG Jacob-Filho W and Figueras A (2005) ldquoAdverse drug reactions in an elderlyhospitalised populationrdquo Drugs amp Aging Vol 22 No 9 pp 767-777

Pecchia L Martin JL Ragozzino A Vanzanella C Scognamiglio A Mirarchi L and Morgan SP(2013) ldquoUser needs elicitation via analytic hierarchy process (AHP) A case study on a computedtomography (CT) scannerrdquo BMC Medical Informatics and Decision Making Vol 13 No 2pp 1-11

Percin S (2008) ldquoUsing the ANP approach in selecting and benchmarking ERP systemsrdquoBenchmarking An International Journal Vol 15 No 5 pp 630-649

Pocar M Di Mauro A Passolunghi D Moneta A Alsheraei AMTA Bregasi A Mattioli R andDonatelli F (2010) ldquoPredictors of adverse events after surgical ventricular restoration foradvanced ischaemic cardiomyopathyrdquo European Journal of Cardio-Thoracic Surgery Vol 37No 5 pp 1093-1100

Rafter N Hickey A Condell S Conroy R OrsquoConnor P Vaughan D and Williams D (2015)ldquoAdverse events in healthcare learning from mistakesrdquo QJM An International Journal ofMedicine Vol 108 No 4 pp 273-277

Rajgopal T (2010) ldquoMental well-being at the workplacerdquo Indian Journal of Occupational andEnvironmental Medicine Vol 14 No 3 p 63

Rasmussen K Pedersen AHM Pape L Mikkelsen KL Madsen MD and Nielsen KJ (2014)ldquoWork environment influences adverse events in an emergency departmentrdquo TRIAL Vol 7No 1 pp 10-0949

Royendegh BD and Erol S (2009) ldquoA DEAndashANP hybrid algorithm approach to evaluate auniversityrsquos performancerdquo International Journal of Basic amp Applied Sciences Vol 9 No 10pp 115-129

Saaty TL (1982) Decision Making for Leaders The Analytical Hierarchy Process for Decisions in aComplex World Wadsworth Belmont CA ISBN 0-534-97959-9 Paperback Pittsburgh RWSISBN 0-9620317-0-4 ldquoFocuses on practical application of the AHP briefly covers theoryrdquo

Saaty TL (2008) ldquoDecision making with the analytic hierarchy processrdquo International Journal ofServices Sciences Vol 1 No 1 pp 83-98

Saaty TL (2012) Decision Making for Leaders The Analytic Hierarchy Process for Decisions in aComplex World 3rd rev ed RWS Publications Pittsburgh PA

Saaty TL (2013) ldquoThe modern science of multicriteria decision making and its practical applicationsthe AHPANP approachrdquo Operations Research Vol 61 No 5 pp 1101-1118

2222

MD5610

Saaty TL and Shang JS (2011) ldquoAn innovative orders-of-magnitude approach to AHP-based multi-criteria decision making prioritizing divergent intangible humane actsrdquo European Journal ofOperational Research Vol 214 No 3 pp 703-715

Saaty TL and Tran LT (2007) ldquoOn the invalidity of fuzzifying numerical judgments in the analytichierarchy processrdquo Mathematical and Computer Modelling Vol 46 Nos 78 pp 962-975

Saaty TL and Vargas LG (2012) ldquoHow to make a decisionrdquo Models Methods Concepts ampApplications of the Analytic Hierarchy Process Springer Boston MA pp 1-21

Sadok W Angevin F Bergez JEacute Bockstaller C Colomb B Guichard L Reau R and Doreacute T(2009) ldquoEx ante assessment of the sustainability of alternative cropping systems implicationsfor using multi-criteria decision-aid methods ndash a reviewrdquo Sustainable Agriculture SpringerHolland pp 753-767 doi 101007978-90-481-2666-8_46

Sahandri MGH and Abdullah SK (2009) ldquoGeneric skills in personnel developmentrdquo EuropeanJournal of Social Sciences Vol 11 No 4 pp 484-492

San Cristoacutebal JR (2011) ldquoMulti-criteria decision-making in the selection of a renewable energy projectin Spain the VIKOR methodrdquo Renewable Energy Vol 36 No 2 pp 498-502

Sayadi MK Heydari M and Shahanaghi K (2009) ldquoExtension of VIKOR method for decision-making problem with interval numbersrdquo Applied Mathematical Modelling Vol 33pp 2257-2262

Shaik MN and Abdul-Kader W (2013) ldquoTransportation in reverse logistics enterprise acomprehensive performance measurement methodologyrdquo Production Planning amp ControlVol 24 No 6 pp 495-510

Shemshadi A Shirazi H Toreihi M and Tarokh MJ (2011) ldquoA fuzzy VIKOR method for supplierselection based on entropy measure for objective weightingrdquo Expert Systems with ApplicationsVol 38 No 10 pp 12160-12167

Shieh JI Wu HH and Huang KK (2010) ldquoA DEMATEL method in identifying key success factorsof hospital service qualityrdquo Knowledge-Based Systems Vol 23 No 3 pp 277-282

Shih HS Shyur HJ and Lee ES (2007) ldquoAn extension of TOPSIS for group decision makingrdquoMathematical and Computer Modelling Vol 45 Nos 78 pp 801-813

Shin YB (2017) ldquoRank reversal phenomenon in cross-efficiency evaluation of data envelopmentanalysisrdquo International Journal of Business and Economic Development Vol 5 No 1 pp 1-6

Shin YB Lee S Chun SG and Chung D (2013) ldquoA critical review of popular multi-criteria decisionmaking methodologiesrdquo Issues in Information Systems Vol 14 No 1 pp 358-365

Si S-L You X-Y Liu H-C and Huang J (2017) ldquoIdentifying key performance indicators for holistichospital management with a modified DEMATEL approachrdquo International Journal ofEnvironmental Research and Public Health Vol 14 No 8 pp 976-934

Soltanifar M and Shahghobadi S (2014) ldquoSurvey on rank preservation and rank reversal in dataenvelopment analysisrdquo Knowledge-Based Systems Vol 60 pp 10-19

Srdjevic B (2007) ldquoLinking analytic hierarchy process and social choice methods to support groupdecision-making in water managementrdquo Decision Support Systems Vol 42 No 4 pp 2261-2273

Supeekit T Somboonwiwat T and Kritchanchai D (2016) ldquoDEMATEL-modified ANP to evaluateinternal hospital supply chain performancerdquo Computers and Industrial Engineering Vol 102pp 318-330

Timmermans S and Berg M (2003) The Gold Standard The Challenge of Evidence-Based Medicineand Standardization in Health Care Temple University Press Philadelphia PA

Tong LI Chen CC and Wang CH (2007) ldquoOptimization of multi-response processes using theVIKOR methodrdquo The International Journal of Advanced Manufacturing Technology Vol 31No 11 pp 1049-1057

Tseng ML (2011) ldquoUsing a hybrid MCDM model to evaluate firm environmental knowledgemanagement in uncertaintyrdquo Applied Soft Computing Vol 11 No 1 pp 1340-1352

2223


hospital sector

Tzeng GH and Huang CY (2012) ldquoCombined DEMATEL technique with hybrid MCDMmethods forcreating the aspired intelligent global manufacturing amp logistics systemsrdquo Annals of OperationsResearch Vol 197 No 1 pp 159-190

Vargas LG (2012)Models Methods Concepts amp Applications of the Analytic Hierarchy Process SpringerNew York NY

Velasquez M and Hester PT (2013) ldquoAn analysis of multi-criteria decision making methodsrdquoInternational Journal of Operations Research Vol 10 No 2 pp 56-66

Wang CH and Pang CT (2011) ldquoUsing VIKOR method for evaluating service quality of onlineauction under fuzzy environmentrdquo International Journal of Computer Science amp EngineeringTechnology Vol 1 No 6 pp 307-314

Wang G Qin L Li G and Chen L (2009) ldquoLandfill site selection using spatial informationtechnologies and AHP a case study in Beijing Chinardquo Journal of Environmental ManagementVol 90 No 8 pp 2414-2421

Wang YM and Luo Y (2009) ldquoOn rank reversal in decision analysisrdquo Mathematical and ComputerModelling Vol 49 Nos 56 pp 1221-1229

Wijnmalen DJ and Wedley WC (2008) ldquoNon-discriminating criteria in the AHP removal and rankreversalrdquo Journal of Multi-Criteria Decision Analysis Vol 15 Nos 56 pp 143-149

World Health Organization and Funk M (2005) Mental Health Policies and Programmes in theWorkplace World Health Organization Geneva

Wrenn K Lorenzen B Jones I Zhou C and Aronsky D (2010) ldquoFactors affecting stress inemergency medicine residents while working in the EDrdquo The American Journal of EmergencyMedicine Vol 28 No 8 pp 897-902

Wu HH and Tsai YN (2012) ldquoAn integrated approach of AHP and DEMATEL methods inevaluating the criteria of auto spare parts industryrdquo International Journal of Systems ScienceVol 43 No 11 pp 2114-2124

Wu J Yang F and Liang L (2010) ldquoAmodified complete ranking of DMUs using restrictions in DEAmodelsrdquo Applied Mathematics and Computation Vol 217 No 2 pp 745-751

Yang JL and Tzeng GH (2011) ldquoAn integrated MCDM technique combined with DEMATEL for anovel cluster-weighted with ANP methodrdquo Expert Systems with Applications Vol 38 No 3pp 1417-1424

Yoo S (2005) ldquoService quality at hospitalsrdquo in Ha YU and Yi Y (Eds) Asia Pacific Advances inConsumer Research Vol 6 Association for Consumer Research Duluth MN pp 188-193

Zavadskas EK Govindan K Antucheviciene J and Turskis Z (2016) ldquoHybrid multiple criteriadecision-making methods a review of applications for sustainability issuesrdquo Economic Research(Ekonomska istraživanja) Vol 29 No 1 pp 857-887

Zhang N and Wei G (2013) ldquoExtension of VIKOR method for decision making problem based onhesitant fuzzy setrdquo Applied Mathematical Modelling Vol 37 No 7 pp 4938-4947

Zhuuml K (2014) ldquoFuzzy analytic hierarchy process fallacy of the popular methodsrdquo European Journal ofOperational Research Vol 236 No 1 pp 209-217

Further reading

Colombo F and Tapay N (2004) OECD Directorate for Employment Labour and Social AffairsOECD Health Working Papers Private Health Insurance in OECD Countries the Benefits andCosts for Individuals and Health Systems doi 101787527211067757

Corresponding authorAntonella Petrillo can be contacted at antonellapetrillounicasit


2224

MD5610

Cost drivers for managingdialysis facilities in a large

chain in TaiwanChia-Ching Cho

Department of Accounting and Information TechnologyCollege of Management National Chung Cheng University Min-Hsiung Taiwan

AnAn ChiuDepartment of International Business College of Business

Feng Chia University Taichung TaiwanShaio Yan Huang

Department of Accounting and Information Technology College of ManagementNational Chung Cheng University Min-Hsiung Taiwan and

Shuen-Zen LiuDepartment of Accounting College of Management

National Taiwan University Taipei Taiwan

AbstractPurpose ndash As the rise in expenditures will be even faster when the baby-boom generation soon reacheshealthcare-dependent ages healthcare providers are facing cost management decision of achieving superiorperformance Taiwan provides a unique environment that the dialysis service providers face only one medicalbuyer The purpose of this paper is to discuss cost factors of dialysis facilitiesDesignmethodologyapproach ndash This study provides a comprehensive analysis of factors influencing thedialysis costs using the data collected from a large renal clinic chain at Taiwan The multiple linear regressionanalysis is employed to examine the factors influencing dialysis costs The research sample composed of1255 patients is collected from 16 dialysis centers in TaiwanFindings ndash The results indicate that the treatment costs of dialysis are influenced by managerial factorsincluding capacity utilization rate (CUR) the percentage of shares held by the owners and the geographicallocation of clinics (LC) The findings assist renal clinics to identify the parts critical to the cost controlOur results indicate that medical variable costs for performing the dialysis treatments are significantlyinfluenced by such managerial factors as CUR the percentage of ownersrsquo shares holding and LCPractical implications ndash By identifying a comprehensive set of costs drivers for dialysis services thisstudy provides useful information for both health providers and policy makers In specific the result assiststhese providers to consider the utilization of better mechanismsinstruments to control costs by increasing theoperational efficiency and achieving the economies of scaleOriginalityvalue ndash This paper contributes to exploring costs drivers that are generally absent from theextant literature The result suggests that the regulators should be aware that the dialysis providers mayreject costly patients Hence to establish the appropriate monitoring mechanisms to prevent such incidence isimportant Finally many other countries in addition to Taiwan also have a similar practice as national healthinsurances or services (eg Medicare in the USA or National Health Service in the UK) Those health systemsmay all face a similar cost control issues for handling end-stage renal disease patients The analysis can helphealth systems worldwide to better design the reimbursement rates to account for the differences existed indealing with the dialysis treatment costsKeywords Healthcare Cost management Cost driver Dialysis servicePaper type Research paper

1 IntroductionRising healthcare expenditures is one of the most contentious issues and a matter of greatconcern for policy makers around the world (Stoltzfus 2012 Strope et al 2009 Ziebarth 2014)Managing costs by utilizing resources effectively is regarded as fundamental to success in



0025-1747DOI 101108MD-06-2017-0550

Received 3 June 2017Revised 8 February 2018

14 February 2018Accepted 13 April 2018


2225

Managingdialysisfacilities


todayrsquos competitive environment M-shaped society and aging of population are two mainphenomena in Taiwan M-shaped society is a polarized society with the extreme rich and theextreme poor The disappearance of the middle class will result in a decline in health careincome affecting the stability of national health insurance The aging of population is anothersignificant issue in Taiwan According the estimate of Council for Economic Planning andDevelopment (CEPD) in Taiwan the elderly population will be over 14 percent in 2017 and by2060 the population aged 65 or older from the current 107 to 416 percent becoming an agingsociety This trend is bound to make the rising number of chronic diseases as well as to improvethe relevance of drug demand meaning that the government will also increase the budget forhealthcare related indirectly promote industrial development

End-stage renal disease (ESRD) is a complete or near complete failure of the kidneys tofunction to excrete wastes concentrate urine and regulate electrolytes The disease usuallyoccurs as chronic renal failure progresses to the point where kidney function is less than10 percent of baseline At this point the kidney function is so low that without dialysis orkidney transplantation complications are multiple and severe and death will occur fromaccumulation of fluids and waste products in the body Treating patients with an ESRD is animportant healthcare problem worldwide The dialysis market has seen a robust growth in thepast five years The total expenditures of ESRD treatments in 2012 is around USD997 millionwith annual growth rate of 27 percent The US Renal Data System (2014) indicates that thecountry of highest ESRD prevalence rate in 2012 is Taiwan followed by Japan and USA

Hemodialysis which relies mainly on medical equipment hemodialysis centers or clinicsis a capital-intensive undertaking In the supply chain of the global dialysis industry thetwo most profitable businesses are the upstream hemodialysis machine and the downstreamdialysis center First of all the hemodialysis machine market is an absolutely oligopolymarket with very high barriers to entry Taking Taiwan as an example there are still nomanufacturers that have the ability to manufacture and all the machines need to beimported abroad At present over 50 percent of the global market share is FreseniusMedical Care (FMC) the German medical device manufacturer The downstream dialysiscenter in all countries has a major group of medical institutions to provide As in the case ofthe USA FMC and DaVita the two largest chain companies have over 1000 kidney dialysiscenters throughout the USA and Buffettrsquos now overcoded DaVita is the most profitabledownstream in the industry chain Dialysis center In Taiwan according to the informationprovided by the Health Protection Bureau there are a total of 562 hemodialysis medicalinstitutes of which 21 are medical centers 239 are related hospitals and 302 are primary-levelclinics in addition to medical centers mostly for the chain or strategic alliance type

Taiwan provides a unique environment that the dialysis service providers only face asingle medical buyer the Bureau of National Health Insurance (BNHI) This model describesthe Taiwan health system which is also called the single payer system and has elements ofboth Beveridge and Bismarck models The single payer tends to have considerable marketpower to negotiate for lower prices National health insurance plans also control costs bylimiting the medical services they will pay for or by making patients wait to be treatedThe criteria to fulfill to get accredited by the system are the number of times The paymentpolicy of the National Health Insurance Agency is fee-for-service-based payment whichmeans that clinics receive fixed reimbursement every time patients have dialysis treatmentThe reimbursement includes technical fees general materials fees special materials feespharmaceutical fees testing fees special drug fee (including EPO) and renal anemia bloodtransfusion cost The maximum number of monthly dialysis treatment is 14 timesThe patient only needs to pay the drug fees not related to the dialysis Due to the feefor a service-based payment system the main factor affecting the reimbursement isthe number of times patients who can receive stable renal dialysis in medical institutionsAccording to Industrial Technology Research Institute research data at present there are

2226

MD5610

50000 patients who have received long-term renal dialysis treatment in Taiwan up to75 million times in one year The BNHI sets a fixed reimbursement rate for the dialysistreatment in per patientrsquos visit However the dialysis medical resources required by patientsare not exactly the same With this challenging environment dialysis providers are pressedto engage in the non-price competition such as purchasing sophisticated equipmentemploying better physicians and enhancing medical services A renal physician can takecare of no more than 20 sickbeds by law On the other hand the providers have to considerthe balance between the service quality and cost control since dialysis service providers areunder the pressure to generate income

The cost analysis and management is always a hot issue in healthcare literature Priorstudies focus on discussing factors of national health care expenditure (Levy et al 2006Fowles et al 1996 Van Vliet and Van de Ven 1993) and patientsrsquo characteristics for medicalneeds However there is no sufficient understanding on what factors influence the dialysiscosts in previous researches These factors are known as costs drivers Therefore this studyaims to analyze costs drivers of dialysis facilities

The data are collected from the large renal clinic data in Taiwan In this study multiplelinear regression analysis is employed to examine cost factors Independent variablescomprise five managerial factors and other control variables including medical treatmentspatient characteristics and medical qualities The sample consists of 1255 observationscollected from 16 dialysis centers from 2007 to 2008 Our results indicate that the treatmentcosts of dialysis are directly influenced by the capacity utilization rate (CUR) percentage ofshares held by owners and location of clinics (LC)

2 Research data and methodsData sourceThe data are collected from the large renal clinic company in Taiwan which is amultinational corporation and operates around 60 dialysis facilities across Asia treatingnearly 4000 patients annually The company also has a strategic alliance with SatelliteHealthcare which is one of the big-six dialysis providers in the USA In October 1997 therenal clinic chain had acquired its first dialysis facility in Taiwan As the numbers of clinicsgrew the company developed a comprehensive country management infrastructureincluding but not limiting to accounting management and clinical reporting systems

The dialysis service market in Taiwan can be divided into three parts which areclinic chains hospital groups and independent units The clinic chains participate around40 percent of the market Our sample chain is the third largest provider The hospitalgroups including public healthcare organizations and private medical foundations thatown more than two hospitals occupy 30 percent of the market Finally the independentunits share the left 30 percent of the market but have been losing steadily their marketshares over the years

We focus on the drivers of variable costs for the following two reasons First healthcareproviders cannot control most of fixed costs because of the regulations discussed earlierFurther the salary levels of renal physicians and nurses are determined by the market andare also quite stable over time Thus the variable costs of renal clinics are much morepossible to be managed Second clinics have different variable costs even in the same clinicchain It is interesting to investigate the factors cause the differences

The case company provided us with the monthly operation data and other relatedinformation The sample consists of 1255 patients from 16 dialysis centers which onlyprovide the hemodialysis services These data are mainly drawn from the operations in2007-2008 We use 2007 and 2008 annual data and then randomly selected individualmonthly data We compare annual and monthly data to make sure that there is nosignificant difference in the relationship between income and cost

2227


Research methodMultiple linear regression model is utilized to examine the factors influencing the costs ofdialysis in this paper (Ullmann 1984 Menke 1997 Kyne et al 2002) In the analysis thetotal medical variable costs per treatment are calculated as a dependent variable The costsinclude hemodialysis concentrate physiological saline dialyzer EPO Calcujex RocaltrolFerrumin blood transfusion extra medicines and other medicines and supplies In the casecompany total medical variable costs are approximately 2313 percent of the total costsPatients have 12-14 visits (treatments) in a month according to their own health conditions

This paper focuses on managerial factors to analyze the cost factors This paper selectsfive managerial factors for further discussion including CUR LC shareholding ratebusiness model length of time clinics managed by the case company The independentvariables consist of five managerial factors and other control variables which are brieflydiscussed as follows

CUR The dialysis costs of capacity are largely fixed such as personnel salaries andequipment depreciation Hence the average capacity cost decreases as service volumeincreases To this end good management of capacity is critical to the productivity andoperating performance for the renal clinics (Hertenstein et al 2006) As per abovediscussion the price of dialysis in Taiwan is fixed Thus physicians can only engage in anon-price competition such as using better medicines nutriments andor high-qualitydialyzer (Dranove and Satterthwaite 2000) This paper investigates whether the physicianswould incur higher variable costs to attract or retain patients when capacity utilizationdecreases physicianrsquos shareholding rates business model and length of time clinicsmanaged by the case company The CUR is the percentage of a clinicrsquos production capacityused over In the healthcare setting the number of hospital beds (eg dialysis usuallyperformed in a bed at Taiwan) employed for dialysis is divided by the total beds clinicsowned to represent the CUR (eg occupancy rate)

LC If the clinic is located in the three big cities namely Taipei Taichung andKaohsiung in Taiwan LC equals 1 otherwise it equals 0 LC is critical to the operating costsincluding rental costs wages and marketing costs and the degree of competition Moreintensive competition tends to result in the higher operation costs for clinics (Dranove andSatterthwaite 2000) LC is critical to the operating costs including rental costs wages andmarketing costs and the degree of competition The intensive competition results in higheroperation costs for clinics (Dranove and Satterthwaite 2000) In Taiwan the density ofmedical resources is much higher in large cities than in rural areas implying that thecompetition is more intensive in these big cities This paper examines whether clinicslocated in the bigger cities incur more medical variable costs than those in the rural areasWe use a dummy variable LC as a proxy of location

Shares holding rate (SHR) The SHR is defined as the percentage of equity shares of clinicsheld by the case company for examining the agency problem and that is the relationshipbetween a principal (eg the renal chain company) and an agent (eg the physicians in theclinics) Ang et al (2000) document that the management ownership is negatively associatedwith operating expenses Strope et al (2009) argue that physiciansrsquo ownership is associatedwith an increasing use of ambulatory surgical centers representing the efforts of costsreduction In specific physicians who do not have ownership and receive the fixed salary havelow incentive to control the operating costs To reduce the agency problem the moststraightforward way is to increase physiciansrsquo ownership ( Jensen and Meckling 1976)

Business model (BS) The case company has two different types of business modelsThe first one is to establish a clinic inside an affiliated hospital The other one is to have anindependent clinic outside the hospital A clinic affiliated with hospitals possess two kindsof advantages having a more stable pool of patients and providing patients with more

2228

MD5610

flexibility to visit other medical departments located in the same hospital during the samevisit (Chen 2004) However clinics affiliated with hospitals need to pay the monthly feeranging from USD12000 to 43000 to the contracted hospitals In contrast an independentclinic does not have to pay such fees The disadvantage is that the independent clinic needsto develop its own patient base and this task could be quite costly Thus it is important toinvestigate how different business models will affect the operating costs We use a dummyvariable ndash BS ndash as proxy of the business model If the clinic is affiliated with hospitalsBS equals 1 otherwise BS equals 0

Length of time clinics managed by the case company (time) Mitchell et al (2000) indicatethat the transfer of learning and experiences from chain organizations improve thecapabilities and performance of individual units In addition Rogers (1995) suggests thatone of the key elements for new technology spread is time That is physicians and relatedstaffs need time to accumulate enough operating experiences to master a new managementsystem and its related technologies Effective learning will increase the operating efficiencyand thus reduce the operational costs In this paper the case company is a multinationaldialysis service provider The main strength is to offer professional administrative supportsof ESRD care such as training physicians and nurses purchasing dialysis medicines andsupplies and management consulting The clinic joins this renal chain company at differenttimings and is expected to improve its own management skills through the infrastructureand framework provided by this chain company Thus this paper investigates whether theclinic has higher medical variable costs when length of time clinics managed by the casecompany is shorter The length of time is calculated by howmany years that the clinics havebeen managed by the case company

Other control variables This paper investigates four control variables including reusingdialyzer (RAK) erythropoietin (EPO) the hours of renal dialysis per treatment (HR) offeringother medicine (OTHM) and blood transfusion (BT) RAK OTHM and BT are dummyvariables and EPO is expressed in terms of international unit

Prior studies focus on risk-adjustment factors of national healthcare expenditure(Levy et al 2006 Fowles et al 1996 Van Vliet and Van de Ven 1993) Specifically theseaforementioned studies investigate such factors as patientsrsquo characteristics and clinicalconditions This paper adds these control variables whenif data permit These factors arenamely age (AGE) gender (GEN) the existence of hepatitis B (BHE) and C (CHE) diabetes(DM) hypertension (HTM) cardiovascular diseases (VC) arteriovenostomy type (TY) yearfor dialysis (TYTD) albumin (ALB) and hematocrit (HCT) AGE ALB and HCT are definedbased on their appropriate measures and the rest factors are dummy variables Finallymortality (MOR) and transfer rate (TR) are employed to control for medical quality of theclinics Both variables are also closely monitored by the BNHI in Taiwan The model isconstructed as follows

Costterm frac14 athornb1CURthornb2SHRthornb3LCthornb4BSthornb5TIMEthornb6RAKthornb7HRthornb8EPO

thornb9OTHMthornb10BTthornb11GENthornb12AGEthornb13BHEthornb14CHE

thornb15DMthornb16HTMthornb17VCthornb18TYthornb19TYTDthornb20ALBthornb21HCT

thornb22MORthornb23TRthorne

where cost termfrac14 total medical variable costs CURfrac14 the capacity utilization rateSHRfrac14 the percentage of clinic ownership held by the company LCfrac14 1 if the clinic is locatedat one of big cities (including Taipei Taichung and Kaohsiung) in Taiwan otherwise 0BSfrac14 1 if the clinic is an affiliate of hospital otherwise 0 TIMEfrac14 the length of timemanaged by the case company RAKfrac14 1 if the clinic does not reuse dialyzer otherwise 0

2229


HRfrac14 the hours of dialysis per treatment EPOfrac14 erythropoietin OTHMfrac14 1 if the renalclinic offers other medicine otherwise 0 BTfrac14 1 if the treatment needs blood transfusionotherwise 0 GENfrac14 1 if the patient is male otherwise 0 AGEfrac14 the patientrsquos age BHEfrac14 1 ifthe patient suffers from hepatitis B otherwise 0 CHEfrac14 1 if the patient suffers fromhepatitis C otherwise 0 DMfrac14 1 if the patient suffers from diabetes otherwise 0 HTMfrac14 1 ifthe patient suffers from hypertension otherwise 0 VCfrac14 1 if the patient suffers fromcardiovascular diseases otherwise 0 TYfrac14 1 if the patient uses the fistula of dialysis portalotherwise 0 TYTDfrac14 total years of ESRD patients starting dialysis to date ALBfrac14 theindex of albumin HCTfrac14 the index of hematocrit MORfrac14mortality (the ratio of patientdeaths that occurred in the specific clinic during the time period from 2007 to 2008)TRfrac14 transfer rate (the ratio of patient transferring to other clinics or hospitals that occurredin the specific clinic during the time period from 2007 to 2008)

3 Empirical resultsThe sample has three special cost characteristics First the dialysis clinic has a very highproportion of fixed costs which is 74 percent of the total costs In contrast the fixed costs inother specialty clinics such as dental clinics usually range from 56 to 62 percent Secondover 50 percent of the fixed costs are the personnel costs of physicians and nurses which areway higher than the costs of long-term assets Last a high volatility of variable costs existsamong the clinics due to the different patient and clinic characteristics

Descriptive statisticsTable I presents the samples composition and the percentage of average medical variablecosts for these 16 dialysis clinics In Table I the highest medical variable cost rate is381 percent the mean is around 158 percent and the lowest is about 63 percent

Descriptive statistics of the independent variables in the regression model are presentedin Table II The average of a clinicrsquos CUR is 5801 percent and the highest is 7955 percentand the lowest is 3704 percent It shows that the competition is so intense that the clinics failto operate in a full capacity in Taiwan The maximal SHR is 100 percent and the lowest is

Renal clinic code Sample numbers Max () Min () Mean () SD ()

1 209 310 69 163 412 53 208 85 138 323 54 276 69 152 484 64 290 69 156 455 72 224 85 144 336 57 321 88 165 427 154 381 63 115 348 57 284 93 173 419 117 360 89 190 4710 122 298 91 169 3911 57 219 73 146 2812 89 296 127 167 2513 39 200 87 147 3114 36 328 129 181 5215 56 324 123 177 3216 19 216 149 175 18Total 1255 63 63 158 44Notes The percentage of average medical variable costs is equal to average medical variable costs dividedby average unit revenue In this table we show the percentage to substitute original costs on account ofkeeping confidential for the sample company but we use the dollar value of costs in the regression model

Table ISample numbers andthe medical variablecost rate

2230

MD5610

26 percent which shows that the controlling power of the case company is very different inits clinics chain The result presents that 62 percent of the observations are in large urbanareas and is consistent with the high urbanization development in Taiwan Further thispaper finds that 59 percent of the patients receive dialysis therapies in an affiliate of aspecific hospital It shows that the hospital is an important patientrsquos source for the dialysisclinics The average time that renal clinics join the sample company is about 63 years andthe longest time is 983 years The sample consists of 47 percent of men and 53 percent ofwomen respectively The percentage is very close to that of Taiwanrsquos population

Multiple linear regression model This paper uses the multiple linear regression analysis toexamine the factors influencing the costs of dialysis Multicollinearity does not appear to be asignificant problem here since the Pearson correlations for all independent variables are lessthan 06 Moreover the VIFs variance inflation factors (VIF) of independent variables in theregression are actually smaller than 10 In specific the VIF of CUR is 385 which is the highestone Since the White (1980) test indicates the existence of heteroskedasticity problem thispaper uses the heteroskedasticity-consistent standard errors (HCSEs) introduced in the studyto correct the problem Main results of the multiple linear regression are presented in Table III

Table III reports the results of regression This study pays the attention to the role ofmanagerial cost drivers It first compares the result of regression with the five managerialfactors using model 2 and results of regression without those variables using model 1The explanatory power (adjusted R2) changes from 0575 to 0654 which indicates asignificant increase (ΔR2 is 0079 F-value is 56213 p-valueo001) This result implies thatwithout including managerial factors there will be a serious omitted variable problem inanalyzing the costs drivers

In model 2 it is noted that three of the managerial factors significantly affect the totalmedical variable costs The result suggests that non-medical factors may changephysiciansrsquo behavior and thus adjust their medical expenditures accordingly First the CURis negatively (minus404 po001) associated with the total medical variable costs In other

Variable Mean Medium Minimum Maximum SD

CUR 5801 5725 3704 7955 1134SHR 8339 10000 2600 10000 2094LC 062 100 000 100 049BS 059 100 000 100 049Time 630 627 050 983 247RAK 072 100 000 100 045HR 400 400 300 600 028EPO 1615996 1500000 000 14000000 1196779OTHM 051 100 000 100 050BT 005 000 000 100 022GEN 047 000 000 100 050AGE 6011 6000 1500 9100 1334BHE 012 000 000 100 033CHE 030 000 000 100 046DM 022 000 000 100 041HTM 039 000 000 100 049VC 028 000 000 100 045TY 077 100 000 100 042TYTD 668 610 000 2800 572ALB 391 390 190 540 047HCT 3045 3010 1600 4660 415MOR 124 118 000 380 105TR 139 125 000 811 212

Table IIDescriptive Statistics

of independentvariables

2231


words when capacity utilization is low physicians tend to incur the higher costs in handlingtreatments A further investigation indicates that three elements of medical variable costswhich are dialyzer EPO and other medicines are significantly higher in these facilities witha lower CUR Second the SHR is positively (287 po001) associated with the variable costsThat is when physicians own a smaller percentage of the clinics they have less incentive tocontrol the variable costs Thus the ownership structure does concern the operation of renalclinics Finally clinics located (LC) in the larger cities tend to incur higher total medicalvariable costs per patient (3112 po001) than those located in the rural areas As expectedintense competition may impose the significant costs for the renal clinics in the bigger citiesThe remaining two managerial factors types of business model (BS) and the length of timejoining the case company (time) are not related to the medical variable costs of the clinicsThe insignificance of BS implies that a dialysis process is similarly provided despite ofbusiness models In addition given that the chain company makes great efforts inincreasing the operating efficiency for their clinics the insignificance of time may suggestthat the company should review its present management policy and make some neededimprovements accordingly The control variables are found to be positively associated withvariable treatment costs Reusing dialyzer (RAK) is the most important costs driverfollowed by the amount of erythropoietin EPO as ranked by their standardized coefficients

Variables Estimated coefficient p-value

Constant 23725 o001

Managerial factorsCUR minus427 o001SHR 285 o001LC 3469 o001BS minus 191 087Time 235 029

Clinical factorsRAK 21008 o001HR 5113 o001EPO 001 o001OTHM 3110 o001BT 4858 o001

Patient characteristicsGEN 1827 o001AGE_Q4 minus1101 o001BHE 3439 o001CHE minus425 057DM minus1363 007HTM 1326 003VC minus1903 001TY 809 025TYTD 240 o001ALB minus4073 o001HCT 063 047

Medical qualitiesMOR minus531 011TR minus048 083Adj R2 0655Notes See p 14 for definitions of variables po005 po001

Table IIIThe result of linearityregression model

2232

MD5610

With regard to the patientsrsquo characteristics male and younger patients significantlyaffect variable costs Patients diagnosed with hepatitis B or hypertensions consume moremedical variable costs but those diagnosed with cardiovascular diseases on the other handconsume less We find that the albumin index is negatively related with the medical variablecosts This result implies that patients with the sufficient nourishment consume less medicalresources Finally the total number of years that patients receive the dialysis treatment(TYTD) is positively associated with the medical variable costs That is the longertreatment periods not older ages generally result in the higher medical variable costsFurthermore the correlation between patientsrsquo age and treatment periods is negativeIn practice the dialysis provider generally notices that experienced patients tend torequestdemand more provided services For example they may ask for additional medicineandor other nutritional supplement

In summary it is interesting to note that medical variable costs are driven by factors farmore complex than what have been shown in the prior literature and managerial factorsappear to be more critical than normally expected

Sensitivity testsThe multiple linear regression model applies the ordinary least square (OLS) method toexamine the factors that influence the costs of dialysis in this study We conduct threeresidual tests to examine whether applying the OLS is adequate or not First we employ thenormal probability plot to test the normality The result shows that the generality of pointsin the probability plot falls on the 45deg line Second we use Durbin-Watson test for theindependence of errors The D-W statistics is 1752 suggesting that the independence issueis not a concern at all Thirdly the White test shows that heteroskedasticity problem mayexist with respect to the error term We use the HCSEs proposed in the study of White (1980)to correct the problem In addition to HCSEs we apply the weighted least squares estimationsuggested by Barber and Thompson (2004) to re-test our regression Table IV shows thatthe main result is still consistent

In the linear regression model the independent variable TYTD is the total number ofyears that ESRD patients receive the dialysis services An alternative of TYTD is thenumber of years of receiving dialysis services from this case company (YTD) We reanalyzethe regression model by changing from the TYTD to YTD The results are generallyconsistent with the original model but the adjusted R2 is lower Detailed statistics arepresented in the Table V

4 Discussion and research implicationsManagers are under increasing pressure to control and justify the cost of sales (Kumar et al2014 Skiba et al 2016)This study uses the data obtained from a large renal clinic chain atTaiwan to investigate the relationships between the dialysis costs and their correspondingcost drivers A special attention is paid to these managerial factors that are absent in theextant literature In addition the factors associated with medical treatments patientsrsquocharacteristics and clinical quality are controlled in this study Our results indicate thatmedical variable costs for performing the dialysis treatments are significantly influenced bysuch managerial factors as CUR percentage of ownersrsquo shares holding and LC

Our findings provide some useful implications for both healthcare providers and policymakers The dialysis providers can better control the associated costs by increasing theoperational efficiency such as CUR In other words there are four dimensions which need tobe improved ndash first a higher utilization rate cannot only bring in more revenues but alsotend to reduce the variable dialysis costs Good information technology and informationsystems (ITIS) will thus improve operations such as increasing the bed utilization rate by

2233


providing the current complete and relevant information in a timely manner (Turan andPalvia 2014) Otherwise the case company is a clinic chain with 16 dialysis centers Utilizingelectronic medical record exchange while adjusting patients among different clinics candecrease the transaction costs (Chang et al 2009) Second the agency problem thatcommonly existed in the profit-seeking settings may also affect the operation of the dialysisclinics and the clinic chain company may sell some portion of its shares to physicians torelief this aforementioned problem Third the company has to design the appropriateperformance schemes to better motivate physicians to reduce the involved costs Fourth abetter cost control mechanisminstrument becomes more important if a clinic locates in thebigger cities as its competition is rather intense Furthermore to correctly identify costfactors is based on the high-quality operating data Woodall et al (2013) also point out thatthe quality of an organizationrsquos data is paramount to its success Dialysis providers canassure that data are suitable for use by performing the appropriate quality assessment(Batini et al 2009 Loshin 2011) Lin et al (2014) indicate that healthcare providers mayhave a higher IT maturity stronger intention to implement IT assessment better ITISresource allocation capabilities and more IT benefits than firms in other industries Dialysisproviders can review their current ITIS and integrate official (eg NHI) and internal IS toincrease the resulting operational performance

With regard to the patientsrsquo characteristics male and younger patients significantlyaffect variable costs The result however is not consistent with other research usingmedical claims data (Cheng et al 2005) One thing should be mentioned is that youngerpatients needing more costs may be confined to clinical dialysis procedures The olderpatients are supposed to be the most costly in the entire care process because of their having

WLS modelVariables Estimated coefficient p-value

Constant 17310 o001CUR minus282 o001SHR 261 o001LC 1790 005BS minus268 073TIME 106 053RAK 22505 o001HR 4881 o001EPO 001 o001OTHM 2789 o001BT 4538 o001GEN 1935 o001AGE minus073 o001BHE 2155 001CHE minus1160 004DM minus413 045HTM 413 039VC minus1433 001TY 543 027TYTD 285 o001ALB minus2085 o001HCT 082 022MOR minus619 012TR minus086 052Adj R2 0740Notes po005 po001

Table IVThe result ofWLS model

2234

MD5610

multiple comorbidities and needing more medical sources (Knauf and Aronson 2009)Patients diagnosed with hepatitis B or hypertensions consume more medical variable costsbut those diagnosed with cardiovascular diseases on the other hand consume less We findthat the albumin index is negatively related with the medical variable costs This resultimplies that patients with the sufficient nourishment consume less medical resources Thesefindings indicate that ESRD with different complications may significantly affect themedical variable costs In addition these medical costs can be decreased by implementing acentral purchasing mechanismpolicy which is based on quantity discount or othereconomic purchasing methods Klein (2012) indicates that internet-based purchasingapplications had a positive contribution on both claims management and operationalperformance outcomes for handling medical practices

The main results on the control variables related to medical treatments and patientsrsquocharacteristics are generally consistent with findings obtained from the previous researchHowever it does have certain differences between this study and the prior ones Forexample prior studies find that older patients consume more medical resources in thedialysis process based on insurance claims data (Howland et al 1987 Schauffler andHowland 1992) which is contrary to our finding based on the actual costs data In oursamples it is interesting to uncover that the older patients take other medicines less andtheir dialyzers do have a higher proportion of reuse which make the average costs of carefor older patients lower than their younger counterparts in the dialysis procedure In themeantime the Taiwan BNHI pays a fixed amount per dialysis regardless of the case if the

TYTD model YTD modelVariables Estimated coefficient p-value Estimated coefficient p-value

Constant 29666 o001 28339 o001CUR minus404 o001 minus405 o001SHR 287 o001 292 o001LC 3112 o001 2991 o001BS minus24 082 minus33 075TIME 202 036 24 027RAK 21229 o001 21174 o001HR 4713 o001 5035 o001EPO 001 o001 001 o001OTHM 3151 o001 3221 o001BT 4756 o001 4847 o001GEN 1854 o001 1856 o001AGE minus109 o001 minus115 o001BHE 3104 o001 3099 o001CHE minus79 029 minus091 090DM minus1288 008 minus1578 003HTM 1342 003 1217 005VC minus1758 001 minus1734 001TY 67 034 708 031TYTD 242 o001TYD 252 o001ALB minus4479 o001 minus4463 o001HCT 087 031 099 025MOR minus599 016 minus62 015TR minus013 094 005 098Adj R2 0654 0652Notes Original model uses the year for dialysis (TYTD) as a control variable new model uses the year fordialysis within sample company (YTD) as a control variable See p 14 for definitions of variables po005po001

Table VThe result of

sensitivity tests

2235


dialyzer is reused or not Our finding suggests that using administrative data to analyze thecosts drivers could provide a more accurate finding than the claims data More researchusing actual costs data in this subject area is thus highly encouraged

The BNHI in Taiwan has set a fixed payment rate for the dialysis treatments and thispolicy is similar to the fee-for-service payment system of Medicare used in USA To this endit is rather easy to implement and estimate the budget However the consumption of medicalresources for dialysis is not uniform among the patients and clinics Specifically in oursample medical variable costs range from 381 to 63 percent of the average revenue Thusit might be inappropriate to use a simple payment scheme to determine the healthcarepolicy Furthermore dialysis providers might consciously select less costly patients whilerejecting these patients who are more costly to treat as they are operating under thefinancial incentive to reduce the associated costs The NHI should pay more attention tomonitor this potential cherry-picking behavior of dialysis providers and strive its best tomaintain a satisfactory quality under the fixed payment scheme In addition other countrieswhich have national health services or insurances (eg National Health Service in UK orMedicare in USA) are also interested in control their relative payments for caring ESRDpatients A refined analysis of costs drivers for dialysis as shown in the paper may offer avaluable help to these healthcare systems to design and develop the reasonablereimbursement rates to account for the existing differences in treatment costs

By identifying a comprehensive set of costs drivers for dialysis services this studyprovides useful information for both healthcare providers and policy makers The maincontribution of this research is to explore costs drivers that are generally absent from theextant literature In specific our analysis assists these providers to consider the utilizationof better mechanismsinstruments to control costs by increasing the operational efficiencyand achieving the economies of scale Furthermore given the incentive to reduce costsdialysis providers might consciously select less costly patients for a treatment whilerejecting these patients that are more costly to treat To remedy this unfortunateconsequence the BNHI should carefully assess the potential cherry-picking behavior ofdialysis providers and strive its best to maintain the quality with the fixed payment schemeFinally many other countries in addition to Taiwan also have a similar practice as nationalhealth insurances or services (eg Medicare in the USA or National Health Service in theUK) Those health systems may all face a similar cost control issues for handling ESRDpatients Our analysis can help health systems worldwide to better design thereimbursement rates to account for the differences existed in dealing with the dialysistreatment costs

Nevertheless our study could be enriched by taking several possible extensions intoconsideration First our study bridges the literature gap by conducting a comprehensiveanalysis of factors influencing dialysis costs using with cross-sectional data from casecompanyrsquos operation But one-year data provided by the case company may pose alimitation of a lack of validation If collecting the time-series data to check how changes interms of different health policies ( from fee-for-serves to global budget payment system)affect the dialysis costs is possible it is expected that more interesting and distinctiveresults and implications can be located This kind of analysis however may require morerefined data provided by the company to conduct additional research and investigationSecond some non-medical cost factors are considered but the process of dialysis servicemay be much more complex to study to determine if there will be a concern onto the omittedvariable problem Additional managerial factors such as customersrsquo (eg patientsrsquo) andemployeersquos satisfaction and different incentive schemes of physicians might also influencethe dialysis costs This line of refinement can be analyzed further to clarify the underlyingcosts structure of renal clinics in addition to clinical factors Third the chain operations inTaiwan or other countries are more popular now than in the past If different types of clinic

2236

MD5610

chains are subsumed into a study various characteristics of clinics or diseases may enablethe analysis of costs drivers more complete Fourth comparing Taiwanrsquos data with datafrom renal clinics in other countries such as the USA Asian and European countries willprovide a better insight to improve the external validity of our results Finally the costsmanagement issues are critical to the most health service providers and having a goodquality of costs data is a base requirement

References

Ang JS Cole RA and Lin JW (2000) ldquoAgency costs and ownership structurerdquo The Journal ofFinance Vol 55 No 1 pp 81-106

Barber JA and Thompson SG (2004) ldquoMultiple regression of cost data use of generalized linearmodelsrdquo Journal of Health Services Research and Policy Vol 9 No 4 pp 197-204

Batini C Cappiello C Francalanci C and Maurino A (2009) ldquoMethodologies for data qualityassessment and improvementrdquo ACM Computing Surveys Vol 41 No 3 pp 1-52

Chang IC Hwang HG Hung MC Kuo KM and Yen DC (2009) ldquoFactors affecting cross-hospitalexchange of electronic medical recordsrdquo Information and Management Vol 46 No 2 pp 109-115

Chen CT (2004) ldquoA study of strategic management and performance of district hospitals in Taiwanafter the implementation of national health insurancerdquo Kaohsiung Medical UniversityDepartment of Public Health master thesis Kaohsiung

Cheng CT Hou HP and Chien CW (2005) ldquoFactors associated with resource utilization of end stagerenal dialysis patientsrdquo Journal of Healthcare Management Vol 6 No 3 pp 291-308

Dranove D and Satterthwaite MA (2000) ldquoThe industrial organization of health care marketsrdquo inCulyer AJ and Newhouse JP (Eds) Handbook of Health Economics Elsevier ScienceNorth Holland pp 1093-1139

Fowles JB Weiner JP and Knutson D (1996) ldquoTaking health status into account when settingcapitation ratesrdquo The Journal of the American Medical Association Vol 276 No 16 pp 1316-1321

Hertenstein JH Polutnik L and McNair CJ (2006) ldquoCapacity cost measures and decisions two fieldstudiesrdquo Journal of Corporate Accounting and Finance Vol 17 No 3 pp 63-78

Howland J Stokes J 3rd and Crane SC (1987) ldquoAdjusting capitation using chronic disease riskfactors a preliminary reportrdquo Health Care Financing Review Vol 9 No 2 pp 15-23

Jensen MC and Meckling WH (1976) ldquoTheory of the firm managerial behavior agency costs andownership structurerdquo Journal of Financial Economics Vol 3 No 4 pp 305-360

Klein R (2012) ldquoAssimilation of internet-based purchasing applications within medical practicesrdquoInformation amp Management Vol 49 No 3 pp 135-141

Knauf F and Aronson PS (2009) ldquoESRD as a window into Americarsquos cost crisis in health carerdquoJournal of the American Society of Nephrology Vol 20 No 10 pp 2093-2097

Kumar V Sunder S and Leone RP (2014) ldquoMeasuring and managing a salespersonrsquos future value tothe firmrdquo Journal of Marketing Research Vol 51 No 5 pp 591-608

Kyne L Hamel MB Polavaram R and Kelly CP (2002) ldquoHealth care costs and mortality associatedwith nosocomial diarrhea due to Clostridium difficilerdquo Clinical Infectious Diseases Vol 34 No 3pp 346-353

Levy JM Robst J and Ingber MJ (2006) ldquoRisk-adjustment system for the Medicare capitated ESRDprogramrdquo Health Care Financing Review Vol 27 No 4 pp 53-69

Lin HCK Chuang TY Lin IL and Chen HY (2014) ldquoElucidating the role of ITIS assessment andresource allocation in ITIS performance in hospitalsrdquo Information amp Management Vol 51No 1 pp 104-112

Loshin D (2011) The Practitionerrsquos Guide to Data Quality Improvement Morgan KaufmannBurlington MA

2237


Menke T (1997) ldquoThe effect of chain membership on hospital costsrdquo Health Services Research Vol 32No 2 pp 177-196

Mitchell W Baum J Berta W Banaszak-Holl J and Bowman D (2000) ldquoOpportunity andconstraint chain-to-component transfer learning in multiunit chains of US nursing homes1991-1997rdquo in Bontis N and Choo CW (Eds) The Strategic Management of Intellectual Capitaland Organizational Knowledge Oxford University Press New York NY pp 555-573

Rogers E (1995) The Diffusion of Innovation 4th ed Free Press New York NY pp 11-20Schauffler HH and Howland J (1992) ldquoUsing chronic disease risk factors to adjust Medicare

capitation paymentsrdquo Health Care Financing Review Vol 14 No 1 pp 79-91Skiba J Saini A and Friend SB (2016) ldquoThe effect of managerial cost prioritization on sales force

turnoverrdquo Journal of Business Research Vol 69 No 12 pp 5917-5924Stoltzfus JT (2012) ldquoEight decades of discouragement the history of health care cost containment in

the USArdquo Forum for Health Economics amp Policy Vol 15 No 3 pp 53-82Strope SA Daignault S Hollingsworth JM Ye Z Wei JT and Hollenbeck BK (2009) ldquoPhysician

ownership of ambulatory surgery centers and practice patterns for urological surgery evidencefrom the state of Floridardquo Medical Care Vol 47 No 4 pp 403-410

Turan AH and Palvia PC (2014) ldquoCritical information technology issues in Turkish healthcarerdquoInformation and Management Vol 51 No 1 pp 57-68

Ullmann SG (1984) ldquoCost analysis and facility reimbursement in the long-term health care industryrdquoHealth Services Research Vol 19 No 1 pp 83-102

US Renal Data System (2014) ldquoUSRDS 2014 Annual data report ESRD in the United States ndash anoverview of USRDSrdquo National Institutes of Health National Institute of Diabetes and Digestiveand Kidney Diseases Bethesda MD pp 183-210

Van Vliet RC and Van de Ven WP (1993) ldquoCapitation payments based on prior hospitalizationsrdquoHealth Economics Vol 2 No 2 pp 177-188

White H (1980) ldquoA heteroskedasticity-consistent covariance matrix estimator and a direct test forheteroscedasticityrdquo Econometrica Vol 48 No 4 pp 817-838

Woodall P Borek A and Parlikad AK (2013) ldquoData quality assessment the hybrid approachrdquoInformation amp Management Vol 50 No 7 pp 369-382

Ziebarth NR (2014) ldquoAssessing the effectiveness of health care cost containment measures evidencefrom the market for rehabilitation carerdquo International Journal of Health Care Finance andEconomics Vol 14 No 1 pp 41-67

Corresponding authorAnAn Chiu can be contacted at ananchiu2009gmailcom


2238

MD5610

Measuring information exchangeand brokerage capacity of

healthcare teamsFrancesca Grippa

College of Professional Studies Northeastern University BostonMassachusetts USA

John Bucuvalas Andrea Booth and Evaline AlessandriniCincinnati Childrenrsquos Hospital Medical Center Cincinnati Ohio USA

Andrea Fronzetti ColladonDepartment of Enterprise Engineering

University of Rome Tor Vergata Rome Italy andLisa M Wade

Cincinnati Childrenrsquos Hospital Medical Center Cincinnati Ohio USA

AbstractPurpose ndash The purpose of this paper is to explore possible factors impacting team performance inhealthcare by focusing on information exchange within and across hospitalrsquos boundariesDesignmethodologyapproach ndash Through a web-survey and group interviews the authors collected dataon the communication networks of 31 members of four interdisciplinary healthcare teams involved in asystem redesign initiative within a large US childrenrsquos hospital The authors mapped their internal andexternal social networks based on management advice technical support and knowledge disseminationwithin and across departments studying interaction patterns that involved more than 700 actorsThe authors then compared team performance and social network metrics such as degree closeness andbetweenness centrality and computed cross ties and constraint levels for each teamFindings ndash The results indicate that highly effective teams were more inwardly focused and less connectedto outside members Moreover highly recognized teams communicated frequently but overall less intenselythan the othersOriginalityvalue ndash Mapping knowledge flows and balancing internal focus and outward connectivity ofinterdisciplinary teams may help healthcare decision makers in their attempt to achieve high value forpatients families and employeesKeywords Healthcare Communication processes Knowledge creation Work teams Social networksPaper type Case study

IntroductionAs recently highlighted in literature the healthcare sector is an environment that isrich in isolated silos and professional ldquotribesrdquo in need of connectivity (Long et al 2013Sexton et al 2017) The healthcare community is increasingly recognizing the need to findnew approaches to improve both outcomes and the overall experience for patients andhealthcare workers It has been widely demonstrated that the majority of the avoidableadverse events are due to the lack of effective communication and collaboration with anestimated 80 percent of serious medical errors involving miscommunication during thehand-off between medical providers (Solet et al 2005) Defining clear handoff practicesreducing interruptions and distractions ensuring a common understanding about thepatient and clarification of transition of responsibility are all key factors to reduce errorsand improve patient safety (Palmieri et al 2008) As reported by several studies over thepast two decades (Kohn et al 2000 Landrigan et al 2010 Makary and Daniel 2016)



0025-1747DOI 101108MD-10-2017-1001

Received 15 October 2017Revised 20 February 2018

9 May 2018Accepted 17 May 2018


2239

Brokeragecapacity ofhealthcare

teams


medical error is the third leading cause of death in the USA A recent literaturereview ( James 2013) described an incidence range of 210000ndash400000 deaths a yearassociated with medical errors among hospital patients Most of the errors are oftenrelated to issues found in the healthcare system rather than problems attributable toindividual errors

One of the approaches to solve this problem is to improve the communication processeswithin and across hospital units building interdisciplinary teams to help reduce themultiple gaps that exist among professions departments and specialties including theclinician-patient divide (Awad et al 2005 Long et al 2013) Working in teams has beendemonstrated to reduce errors as medical staff rely on each otherrsquos expertiseand specialized knowledge (Dutton et al 2003 Chin et al 2004 Lemieux-Charles andMcGuire 2006)

By creating interdisciplinary and cross-functional healthcare teams hospitals have theopportunity to balance the trade-off between exploitation (internal focus) and exploration(external focus) creating the foundations for a true ambidextrous organization(Orsquoreilly and Tushman 2004) This requires selecting individuals with the rightcombination of skills hierarchical position status and external connections which canaffect the exploration and exploitation of new knowledge and impact the trade-off in teamcomposition (Perretti and Negro 2006)

This case study describes how healthcare teams exchange information within andacross boundaries search for new knowledge in order to create a completely new caredelivery system and in doing so rely on internal ties and knowledge of the processThe healthcare teams involved in this study were composed of professionals involved inmaking patientmedical decisions (eg nurses physicians) as well as by others whosedecisions impact health outcomes and safety (eg director of patientfamily experiencehead of the ER unit)

Literature reviewThe present study is based on the recognition that teams are an essential component forbridging the gaps between isolated units within hospitals Our case study relies on adefinition of teams as complex systems made of individuals ldquowho are interdependent intheir tasks who share responsibility for outcomes who see themselves and who are seenby others as an intact social entity embedded in one or more larger social systemsrdquoand who manage their relationships across organizational boundaries (Cohen andBailey 1997 p 241) Our focus is on the task-related team defined as a group ofindividuals whose task requires members to work together to produce something forwhich they are collectively accountable and whose acceptability is potentially assessable(Hackman 2004)

Healthcare teams are usually described based on the type of tasks they perform(Lemieux-Charles and McGuire 2006) Project teams management teams and caredelivery teams might be distinguished based on their daily activities which can involve acombination of direct care of patients and designing new health delivery modesNevertheless their actions have a similar impact on patient safety Each member bringshis or her special knowledge and capabilities but also interpersonal relationshipswith the members inside and outside of the team (Ancona et al 2009) Yet even thoughindividual team members may have distinct and complementary expertise effectiveteams require close ties among the members ability to effectively communicate andorganizational support

In their literature review of healthcare team effectiveness from 1985 to 2004 Lemieux-Charlesand McGuire (2006) linked outcomes to team effectiveness and to processes like effective teamcommunication and cohesion They observed that increased team autonomy correlated with

2240

MD5610

decreased hospital readmissions and with higher levels of staff satisfaction and retention High-functioning teams have been characterized by positive communication patterns and high levelsof collaboration and participation (Shortell 2004 Temkin-Greener et al 2004) Other studiesfound that increased team diversity and interdependence are associated with decreased lengthof stay and hospital charges (Dutton et al 2003) Further evidence indicated that teamcommunication and training in the use of quality improvement methods was linked withimproved patient outcomes

Studies conducted in other industries found that structurally diverse work groups arecharacterized by members who use their different organizational affiliations roles orpositions to expose the team to unique sources of knowledge which is beneficialfor performance (Burt 2004 Cummings 2004) In particular Cummings (2004) foundthat effective work groups engage in external knowledge sharing through theexchange of information know-how and feedback with important stakeholders outsideof the group

In a study that investigated the association between team constraint and teamperformance of 15 process improvement teams Rosenthal (1997) noticed how differencesin social networks explain performance variation teams composed of members with moreentrepreneurial networks were more likely to be recognized for improving the quality ofplant operations In a study of 120 new-product development projects undertaken by41 divisions Hansen (1999) found evidence that weak inter-department ties help aproject team search for knowledge in other departments but impede the transfer ofcomplex knowledge which relies on strong ties between the two parties to a transferBurt (1992 2005) used social network indicators and performance data frommanagerial networks across industries (not including healthcare) to demonstrate thatnetworks that span structural holes are associated with creativity and learning morepositive evaluations and more successful teams Burt (2004) also found that densenetworks do not necessarily enhance performance and could be associated withsubstandard results

The analysis of collaboration and communication among healthcare staff is a keycomponent of any system redesign initiative that aims at improving quality of care(Wagner et al 2001 Shanafelt et al 2010 2015 Bodenheimer and Sinsky 2014) While bestoutcomes depend on productive interactions and communication among members ofinterdisciplinary healthcare teams coordination becomes difficult as teams grow in size Inthe setting of complex care teams must gather information from multiple subspecialistssynthesize the information acquired come to decisions and execute a plan (Delva et al 2008Harrod et al 2016)

MethodComplex care often involves input from and coordination with other departments soinformation must flow beyond unit divisional and departmental boundaries Reportingrelationships increase complexity since team members may belong to distinctdepartments and many individuals belong to multiple teams The visualization of theserelationships is the first step to recognize interdependencies and bottlenecks For thisreason in this study we use social network analysis to build social maps and extractcentrality indicators that can reveal blockages in the information flows and offer ideas onhow to improve team effectiveness

ParticipantsThe study participants were 42 employees of a large childrenrsquos hospital in the USA[1](20 women and 22 men) There was an equal representation of different roles acrosshospitalrsquos units including physicians nurses business directors AVP and VP of finance

2241


teams

directors of quality improvement initiatives clinical pharmacists anesthesiologists andprogram managers Participation in this study was on a voluntary basis Almost all thehospital units were represented in our sample with at least one representative for eachdepartment The hospital units that had more than three members participating in the studywere Anesthesia unit Health Improvement Center Patient Services and PatientFamilyExperience Gastroenterology and the Heart Institute Participants were not workingtogether at the time of the study They represented different units and departments thatwere also located in geographically distant hospital campuses Each member was assignedto a team based on work experience functional unit and tenure within the organizationFor example the director of finance and the AVP of finance were both assigned to the teamin charge of learning how the hospital costs were affecting value creation for patientsfamilies and employees We recognize that individual differences tenure within theorganization and knowledge of the topic could impact the team outcomes Each team wascomposed of both senior and junior employees with a tracked record of expertise in theirrespective area Members with experience in other service industries were also included

InstrumentsThrough a web-questionnaire sent via e-mail we asked participants to report up to 25 peoplewithin and outside the hospital they would go to when looking for advice based on subjectmatter expertise seeking support for their career development seeking technical support orsharing new ideas Out of the 42 team members involved in the project 31 responded to thesurvey (72 percent response rate)

Using the name generator technique (Burt et al 2012) we created a list of 700 uniquecontacts with whom respondents communicated more frequently within and outside thehospital This allowed the creation of four different social networks based on connectionsamong individuals seeking managerial advice sharing new ideas looking for an expertopinion during complex cases and for solving technical problems both within and acrossthe hospitalrsquos boundaries To map and measure the internal and external social networkswe used metrics of social network analysis (Burt 1992 Wasserman and Faust 1994Cross et al 2002) which helped to identify brokers boundary spanners and centralconnectors who can transfer knowledge between departments and increase collaboration

ProcedureParticipants were assigned to five teams whose goal was to conduct a preliminaryinventory of strengths weaknesses and opportunities to improve the current system of caredelivery at the hospital and learn how the organization impacted the experience of theirpatients families and staff The teams were charged with finding exemplars in valuedelivery both in healthcare and other industries They were prompted to look outsidethe hospital boundaries at organizations in other industry that had excelled in quality ofservice and personalization of the experience (eg The Walt Disney Company) The finaldeliverable was an assessment of the current situation of the hospital with a proposal ofimprovements with regards to five areas safety patient and family experience (PFE)employee engagement and team function (EETF) healthcare outcomes and costs The fiveteams were charged with exploring challenges and opportunities of a new care deliverysystem that could result in a quantum leap in improvement of outcomes patientfamily andprovider experience The team members worked together over a period of six months andpresented their findings during a two-day synthesis session They generated extensivereports to describe the status quo for the five subjects and offered recommendations forimprovements Members of the teams met face-to-face during bi-weekly meetings to engagein design prototyping testing and implementation of a new healthcare delivery system

2242

MD5610

The team focused on ldquoSafetyrdquowas excluded from the analysis since their reportdeliverablewas missing at the time of the observation The analysis included four teams plus anoperational team whose members coordinated their work to guarantee a seamless process

In order to understand the mechanisms that could lead to effective teamwork inhealthcare we collected different variables Team performance was assessed by teamleaders at the end of the six month-period and was operationalized based on the numberand quality of insights as well as their impact on the project Team leaders were asked toassess the teams on three criteria originality of the findings number of findings and impactof findings on the overall project in terms of quality and usefulness Scores spanned from0 (frac14 very low) to 5 (frac14 extremely high)

To understand the degree of connectivity of teams (Wasserman and Faust 1994 Everettand Borgatti 2005) we used social network analysis metrics that can offer insights on theinternal dynamics and existing ties among members (Cummings 2004) These metrics aredescribed in Table I and include degree centrality in-degree centrality out-degree centralitycloseness centrality To identify the ability of team to be outwardly connected we selectedbetweenness centrality network constraints and cross ties (see also Figure 1) These metricsoffer the opportunity to measure the brokerage capacity of team members to establishconnections with other units and teams Indicators of brokerage capacity measure the averageability of team members to serve as bridges within or outside their team while connectivityfocuses on the direct contacts of team members in terms of number of incoming and outgoingties as well as the degree to which a team member is near all other members and thereforemore embedded at the network core Prior studies have highlighted the benefits of key socialnetwork positions in networks such as advice and trust (Battistoni and Fronzetti Colladon2014) By observing both connectivity and brokerage capacity we aim at measuring themembersrsquo ability to explore new radical ideas coming from other industry and otherdepartments and to exploit the already existing knowledge within the organization (Orsquoreillyand Tushman 2004) As Hargadon (2005 p 17) suggested by holding a central position intheir informal social networks individuals are more likely ldquoto acquire knowledge withoutacquiring the ties that typically bind such knowledge to particular worldsrdquo

Metric Definition

Degree centrality The total number of ties a node has to other nodesIn-degreecentrality

Number of incoming ties representing received requests of advice knowledge sharingand technical support

Out-degreecentrality

Number of outgoing ties representing requests of advice knowledge sharing andtechnical support made by each individual

Closenesscentrality

The average length of the paths linking a node to all others This measure can sometimesbe seen as a proxy of the speed with which a node can be reached or can reach the others(Wasserman and Faust 1994)

Betweennesscentrality

The extent to which a node is connected to other nodes that are not connected to eachother It is a measure of the degree to which a node serves as a bridge mediating forinstance a request of advice

Networkconstraint

Measures the extent to which an actorrsquos network is a limitation around himher limitinghis or her vision of alternative ideas and sources of support Network constraint is anindex that measures the extent to which a personrsquos contacts are also linked amongthemselves closing the triads (ie if A is connected to B and C there is also a link betweenB and C) A social actor who can mediate a connection between unlinked peers can takeadvantage of hisher social position and choose for example a ldquodivide et imperardquostrategy or be the broker of good ideas Burt (2004) In this example we have aldquostructural holerdquo which is the missing link between B and C and therefore a lower valueof network constraint for A

Cross ties Number of links towards actors belonging to social clusters different from theirs

Table IMetrics of social

network analysis usedin the study

2243


teams

ResultsTo visually represent how frequently members cross the organizational boundaries toaccess critical information we mapped information flows among the departments andamong team members Figure 2 identifies the teams whose members potentially acted asknowledge brokers showing a lower ldquonetwork constraintrdquo score who were in a position tobetter facilitate the exchange of information across hospital units and teams Actors withlow constraints have more opportunities for brokering as well as an advantage with respectto information access (Burt 1992) Most of the knowledge brokers were members of theOutcomes team spanning connections across different departments and outsidestakeholders Operational Team members who coordinated the entire improvementproject and members of the PFE and EETF teams were deeply embedded in multiple workgroups playing various roles across departments and acting as ambassadors of the project

Figure 3 illustrates the variation in out-group communication for each teamThe Outcome team and the PFE team had more ties with external stakeholders than theCost and EETF teams

In general teams had more external contacts with other hospitals or universitydepartments Other external links were with people working in the healthcare industry(private companies) personal contacts or employees of the government or of the Institute forHealthcare Improvement The Outcome team had more heterogeneous contacts The PFEteam also had a significant amount of communication which cross the organizationalboundaries Cost and EETF teams on the other hand had significantly lower interactionswith potential external knowledge sources

Figure 4 reports the metrics of social interaction for each team by differentiating betweenout-group and in-group metrics as well as between their brokerage capacity and networkconnectivity For the in-group and out-group communications the PFE and Outcome teamshave more cross ties and higher betweenness centrality which indicate a stronger effort to

Team Performance

Connectivity

Degree Centrality

In-degree Centrality

Out-degree Centrality

Closeness Centrality

Betweenness Centrality

Brokerage Capacity

Network Constraints

Cross TiesFigure 1Variables representedin the study

Distribution of Top 20 Knowledge Brokers6

4

3

2

1

0Operational

Num

ber

of T

op B

roke

rs

Cost

Team Name

EETF OutcomesPFE

5

Figure 2Top knowledgebrokers across teams(EETF stands foremployee engagementand team functionand PFE stands forpatient-familyexperience)

2244

MD5610

connect across boundaries and tap into other unitsrsquo expertise (ie a higher brokerage capacity)With respect to connectivity we see that Outcome and PFE teams have more outgoing tiesand are closer to the network core PFE also shows high values of in-degree centrality provingits significant amount of communication also within the hospital boundaries In addition PFEand Outcome teams are more central with respect to closeness we see how overall theyoutperform the Cost and EETF teams in terms of connectivity

Figure 5 illustrates the scores associated to the work of the four teams based on anassessment of number of findings originalityquality of findings and impact The EETFand the Cost teams received the highest scores in all the three criteria

In summary our findings indicate that teams that perform better have an inverserelationships with brokerage capacity and connectivity They have less frequentinteractions with external knowledge sources and are less embedded also in the internalnetwork which translates in a lower closeness and less direct connections On the otherhand teams whose report received lower scores (PFE and Outcomes) were highly connectedwith other hospital units government agencies and industry professionals The findingsseem to suggest that highly ranked teams are focused more inwardly and their membersare less central with respect to the full network (lower values of average betweenness andcloseness centrality) Members of the Cost team and EETF team ( for the out-group) hadhigher network constraint scores (Burt 2004) indicating a higher closure of theirego-networks which indicates that each of the memberrsquos contacts is connected to hisherother contacts This means that closer relationships with their team membersmdashwith alower number of direct contacts to manage and less brokerage tiesmdashproduced moreeffective knowledge sharing and efficient communication processes

DiscussionThis study confirms previous research on team effectiveness (Gupta et al 2006 Siggelkowand Rivkin 2006) describing the relationship between team performance andcommunication as having an inversely u-shaped form team effectiveness can be pursuedby balancing exploitation (internal focus) and exploration (external focus) and by avoiding

Outcome

Hospital

University

Industry

Personal

GovernmentIHI

PFE

EEFT

Cost

Figure 3Out-group

communications

2245


teams

12

10

6

4

2

0Outcome

Number of Findings Originality Impact

PFE EETF Cost

8 36

28

3638 38

34 36

4 43632

3

Team Performance

Figure 5Team performancebased on numberquality and impactof findings

Brokerage Capacity Ingroup

Outgroup

Connectivity

200

PFE

PFE

PFE

PFE030

025

020

015

010

005

000

035030

025020

015

005010

0000605

0403

0201

00

EETF

EETF

EETF

EETF

Outcome

Outcome

Outcome

Outcome

Cost

Cost

Cost

Cost

Avg Constraint

Avg Constraint

Avg Betweenness

Avg Betweenness

Cro

ss T

ies

Cro

ss T

ies

Avg

Clo

sene

ssA

vg C

lose

ness

Indegree

Indegree

Outdegree

Outdegree

150

100

50

0

000005

010015

020025

030035

040000 005 010

015 020 025030 035

040

07

06

05

04

03

02

01

00000

005

010

015

0000

100

80

60

40

20

0010

008

004006

00206 000

0504

0302

0100

0002 000400060008001000120014

0016

Figure 4Team position basedon connectivityand brokeragecapacity metrics

2246

MD5610

excessive or inadequate communication In this study we found that highly effective teamswere more inwardly focused and less connected to outside members Members of theout-group were both employees working in other hospital units and individuals outside thehospital connected to the participants

The results indicate that teams who scored the highest in terms of quality originalityand impact of findings (EETF and Cost) communicated frequently but overall lessintensely than others Having less scattered communication seems to be associated withhigher team effectiveness as teams may focus on the immediate deliverable and have moreefficient conversations We find that acting as broker and facilitating information flowsmight not always conduce to higher recognition Consistently our results seem to suggestthat a large number of cross-ties between team members and people outside the hospital isnot necessarily associated to increased team performance While innovation has beenassociated in the past with the ability of teams and organizations to cross institutionalboundaries and tap into new ideas and different perspectives (Ancona et al 2009) there is afine balance between excessive communication and inadequate interaction with variousstakeholders A not too high level of inter-group connections is more likely to lead to thehighest performance ldquoby enabling superior ideas to diffuse across groups without reducingorganizational diversity too quicklyrdquo (Fang et al 2010 p 625)

Our exploratory study found that higher network constraint levels are possiblyconducive to higher team performance Brokers on the other hand have a very importantrole in the long term especially when the project becomes increasingly oriented to outsidestakeholders rather than toward internal team operations The brokers identified in eachteam might become strategic partners or champions when the project enters the nextphase where their network position will facilitate the creation of interfaces with otherexternal organizations and outside members An explanation for our finding is that theteams had a limited time to get to know each other understand how every member couldcontribute to the overall goal and leverage each otherrsquos knowledge both tacit and explicitTeams who produced a more impactful deliverable had a more focused communicationdispersed over a lower number of connections and with fewer cross ties with externalstakeholders This might have helped members to stay focused and leverage each otherrsquosinformal connections within the team It is important to remember that the final deliverablewas a report containing information and suggestions on the current state of the hospitalwith regard to healthcare cost healthcare outcomes employee engagement and patientexperience It could be that teams with fewer external ties had a better chance to focus oncollecting relevant institutional knowledge while others who had higher external ties mighthave been pulled into different directions and could have been less focused on their task Inparticular the EETF team was composed of members who had immediate access to internalknowledge repository and a direct formal ties to the HR department which helped locate theright information in the most efficient way Because of the strong ties of the EETFrsquosmembers the team had access to internal documents and built a deliverable that resonatedimmediately with the hospital leadership Future research should verify if our findings arereplicable when healthcare teams have different goals or when they are long established(with a long history of interaction among all members) Our teams had the same goal (ieexplore challenges and opportunities of a new care delivery system that could result in aquantum leap in improvement of outcomes patientfamily and provider experience) thoughthey varied in team compositions and ties to other units

Another possible reason for our result on brokerage and performance is connected to arecent research study on social contagion Centola (2015) built a model of social networkformation and demonstrated how breaking down group boundaries to increase the diffusionof knowledge may result in less effective knowledge sharing Centolarsquos research suggeststhat complex ideas are more freely integrated across groups if some degree of group

2247


teams

boundaries is preserved This is aligned with the idea that social ties are constrained byindividualsrsquo location in social spaces and that their social identities are defined by theirparticipation in social groups (McPherson 2004 Kossinets and Watts 2009)

In our study teams seemed more effective and efficient when fewer cross-ties existedsignaling an increased focus on internal team operation We also found that thehighest-scoring teams used communication media in a parsimonious way Instead ofswitching from one communication medium to the other they chose one or two channels tointeract with each other The most effective teams were able to reduce ambiguity andincrease team effectiveness by using only a limited number of channels instead ofdispersing time and energy on multiple media (Dennis et al 2008)

Conclusions and limitationsThis study offers healthcare leaders practical insights on strategies for building teamsthat are interdisciplinary in nature and have a good balance of external and internalconnections Healthcare leaders would benefit from providing teams with the opportunityto work closely with each other establishing strong internal connections In an initialphase interdisciplinary teams with members representing several medical disciplines androles need time to brainstorm learn about individual differences and expertise The teamsin our study were still in the initial stage of the Tuckmanrsquos model of team development(Tuckman and Jensen 1977) After being formed (stage 1) they experience a stormingphase (stage 2) where members are more internally focused and are spending time andenergy getting to know each other as well as their potential contribution That explainswhy highly performing teams at the time of our study were mostly connected internallyrather that with outside members It would be interesting to explore in another studywhether external connections are more prominent in highly effective teams during thefinal stages of norming and performing when the team results are planned to bebroadcasted to a larger external audience

Our study provides some insights to support healthcare decision makers in their attemptto achieve high value for patients families and employees (Porter 2010) We offer empiricalevidence to support clinicians and healthcare providers in their attempt to measureoutcomes at the institutional and team level using new metrics of knowledge flow andteam function Clinicians are trained to rely only on the ldquogold standardrdquo of researchmethodologies which favor quantitative data and empiricism (Walshe and Rundall 2001)In this paper we adopted observational methods and qualitative research to inform decisionmaking providing actionable insights easy to understand an immediately applicable

While this study seems to confirm the need to favor internal focus which could later bebalanced with outward connectivity it still leaves several questions unanswered includingsome raised by Gupta et al (2006) What is the impact on performance when ideas are beingexploited by other individuals or teams How does organizational politics come into playwhen its members have to decide what information to share and when they may feelexploited by others

Our results are based on a sample of teams involved in a specific system redesign projectand were composed of a variety of roles including nurses physicians and medical directorsbut also program directors and others holding administrative roles While their dailyactivities vary with reference to the immediate impact on patientsrsquo health and safety (directand indirect care) the teams in our study were a good sample of the three types of healthcareteams found in literature project management and care delivery (Lemieux-Charles andMcGuire 2006) A fruitful next step in this research stream would be to replicate the studyfocusing only on care delivery teams whose coordination mechanisms and communicationprocesses could be different as they directly involve patients and their families Futurestudies could compare teams over a longer period of time as well as teams that are more

2248

MD5610

similar in task context and composition (eg only nurses and physicians) In that scenarioteam performance could be multifaceted to include clinical outcomes safety events valueteam member ratings of team performance satisfaction and engagement

Because of the small sample we could not implement any statistically significant modelto predict team performance by observing knowledge flows although we got a cleardescription of a typical scenario in healthcare that could explain differences in performanceTeams who were recognized for their impactful work were engaged in more focusedcommunications within their team and with hospital units and had fewer members whoacted as brokering stars Replicating this study with a larger sample may help establishmore support for the theoretical relationships revealed from our study

Note

1 The Cincinnati Childrenrsquos Hospital Medical Center OH USA

References

Ancona D Bresman H and Caldwell D (2009) ldquoThe X-factor six steps to leading high-performingX-teamsrdquo Organizational Dynamics Vol 38 No 3 pp 217-224 doi 101016jorgdyn200904003

Awad SS Fagan SP Bellows C Albo D Green-Rashad B De la Garza M and Berger DH (2005)ldquoBridging the communication gap in the operating room with medicalteam trainingrdquo American Journal of Surgery Vol 190 No 5 pp 770-774 doi 101016jamjsurg200507018

Battistoni E and Fronzetti Colladon A (2014) ldquoPersonality correlates of key roles in informal advicenetworksrdquo Learning and Individual Differences Vol 34 pp 63-69 doi 101016jlindif201405007

Bodenheimer T and Sinsky C (2014) ldquoFrom triple to Quadruple aim care of the patient requires careof the providerrdquo Annals of Family Medicine Vol 12 No 6 pp 573-576 doi 101370afm1713

Burt RS (2004) ldquoStructural holes and good ideasrdquo American Journal of Sociology Vol 110 No 2pp 349-399

Burt RS (2005) Brokerage and Closure An Introduction to Social Capital Oxford University PressNew York NY doi 101007s13398-014-0173-72

Burt RS Meltzer DO Seid M Borgert A Chung JW Colletti RB Dellal G Kahn SAKaplan HC Peterson LE and Margolis P (2012) ldquoWhatrsquos in a name generator Choosing theright name generators for social network surveys in healthcare quality and safety researchrdquoBMJ Quality amp Safety Vol 21 No 12 pp 992-1000 doi 101136bmjqs-2011-000521

Burt RSSRS (1992) ldquoStructural holesrdquo Structural Holes The Social Structure of CompetitionHarvard University Press Cambridge MA p 324

Centola D (2015) ldquoThe social origins of networks and diffusionrdquo American Journal of SociologyVol 120 No 5 pp 1295-1338 doi 101086681275

Chin MH Cook S Drum ML Jin L Guillen M Humikowski CA Koppert J Harrison JFLippold S and Schaefer CT (2004) ldquoImproving diabetes care in Midwest community healthcenters with the health disparities collaborativerdquo Diabetes Care Vol 27 No 1 pp 2-8 doi102337diacare2712

Cohen SG and Bailey DE (1997) ldquoWhat makes teams work group effectiveness research fromthe shop floor to the executive suiterdquo Journal of Management Vol 23 No 3 pp 239-290doi 101177014920639702300303

Cross R Prusak L and Parker A (2002) ldquoWhere work happens the care and feeding of informalnetworks in organizationsrdquo Institute for Knowledge-based Organizations IBM Cambridge MA

Cummings JN (2004) ldquoWork groups structural diversity and knowledge sharing in a globalorganizationrdquo Management Science Vol 50 No 3 pp 352-364 doi 101287mnsc10300134

Delva D Jamieson M and Lemieux M (2008) ldquoTeam effectiveness in academic primary health careteamsrdquo Journal of Interprofessional Care Vol 22 No 6 pp 598-611 doi 10108013561820802201819

2249


teams

Dennis AR Fuller MF and Valacich JS (2008) ldquoMedia tasks and communication processes atheory of media synchronicityrdquo MIS Quarterly Vol 32 No 3 pp 575-600 available at httpsdoiorg10230725148857

Dutton RP Cooper C Jones A Leone S Kramer ME and Scalea TM (2003) ldquoDaily multidisciplinaryrounds shorten length of stay for trauma patientsrdquo The Journal of Trauma Injury Infection andCritical Care Vol 55 No 5 pp 913-919 doi 10109701TA00000933953409756

Everett M and Borgatti SP (2005) ldquoExtending centralityrdquo in Carrington PJ Scott J andWasserman S (Eds) Models and Methods in Social Network Analysis Cambridge UniversityPress Cambridge pp 57-76 doi 101017CBO9780511811395004

Fang C Lee J and Schilling MA (2010) ldquoBalancing exploration and exploitation through structuraldesign the isolation of subgroups and organizational learningrdquo Organization Science Vol 21No 3 pp 625-642 doi 101287orsc10900468

Gupta AK Smith KG and Shalley CE (2006) ldquoThe interplay between exploration and exploitationrdquoAcademy of Management Journal Vol 49 No 4 pp 693-706 doi 105465AMJ200622083026

Hackman JR (2004) ldquoLeading teamsrdquo Team Performance Management An International JournalVol 10 Nos 34 pp 84-88 doi 10110813527590410545081

Hansen MT (1999) ldquoThe search-transfer problem the role of weak ties in sharing knowledgeacross organization subunitsrdquo Administrative Science Quarterly Vol 44 No 1 pp 82-111doi 1023072667032

Hargadon AB (2005) ldquoBridging old worlds and building new ones toward a microsociology ofcreativityrdquo in Thompson LL and Choi H-S (Eds) Creativity and Innovation in OrganizationalTeams Lawrence Erbaum Associates Mahwah NJ pp 199-216

Harrod M Weston LE Robinson C Tremblay A Greenstone CL and Forman J (2016) ldquo lsquoIt goesbeyond good camaraderiersquo a qualitative study of the process of becoming an interprofessionalhealthcare lsquoteamletrsquo rdquo Journal of Interprofessional Care Vol 30 No 3 pp 295-300 doi 1031091356182020151130028

James JT (2013) ldquoA new evidence-based estimate of patient harms associated with hospital carerdquoJournal of Patient Safety Vol 9 No 3 pp 122-128 doi 101097PTS0b013e3182948a69

Kohn LT Corrigan JM and Donaldson MS (Eds) (2000) To Err is Human Building a Safer HealthSystem National Academies Press Washington DC p 287

Kossinets G and Watts DJJ (2009) ldquoOrigins of homophily in an evolving social networkrdquoAmerican Journal of Sociology Vol 115 No 2 pp 405-450 doi 101086599247

Landrigan CP Parry GJ Bones CB Hackbarth AD Goldmann DA and Sharek PJ (2010)ldquoTemporal trends in rates of patient harm resulting from medical carerdquo New England Journal ofMedicine Vol 363 No 22 pp 2124-2134 doi 101056NEJMsa1004404

Lemieux-Charles L and McGuire WL (2006) ldquoWhat do we know about health care teameffectiveness A review of the literaturerdquo Medical care research and review MCRR Vol 63No 3 pp 263-300 doi 1011771077558706287003

Long JC Cunningham FC and Braithwaite J (2013) ldquoBridges brokers and boundary spanners incollaborative networks a systematic reviewrdquo BMC Health Services Research Vol 13 No 158pp 1-13 doi 1011861472-6963-13-158

McPherson M (2004) ldquoA Blau space primer prolegomenon to an ecology of affiliationrdquo Industrial andCorporate Change Vol 13 No 1 pp 263-280 doi 101093icc131263

Makary MA and Daniel M (2016) ldquoMedical errormdashthe third leading cause of death in the USrdquoBmj Vol 2139 No 353 pp 1-5 doi 101136bmji2139

Orsquoreilly CA and Tushman ML (2004) ldquoThe ambidextrous organisationrdquo Harvard Business ReviewVol 82 No 4 pp 74-81

Palmieri PA DeLucia PR Oh TE Peterson LT and Green A (2008) ldquoThe anatomy andphysiology of error in adverse healthcare eventsrdquo Advance in Health Care Management Vol 7No 36 pp 33-68 doi 101016S1474-8231(08)07003-1

2250

MD5610

Perretti F and Negro G (2006) ldquoFilling empty seats how status and organizational hierarchies affectexploration versus exploitation in team designrdquo Academy of Management Journal Vol 49 No 4pp 759-777 doi 105465AMJ200622083032

Porter ME (2010) ldquoWhat is value in health carerdquo New England Journal of Medicine Vol 363 No 26pp 2477-2481 doi 101056NEJMp1011024

Rosenthal E (1997) ldquoSocial networks and team performancerdquo Team Performance Management Vol 3No 4 pp 288-294 doi 10110813527599710195420

Sexton JB Schwartz SP Chadwick WA Rehder KJ Bae J Bokovoy J Doram K Sotile WAdair KC and Profit J (2017) ldquoThe associations between work-life balance behavioursteamwork climate and safety climate cross-sectional survey introducing the work-life climatescale psychometric properties benchmarking data and future directionsrdquo BMJ Quality andSafety Vol 26 No 8 pp 632-640 doi 101136bmjqs-2016-006032

Shanafelt TD Hasan O Dyrbye LN Sinsky C Satele D Sloan J and West CP (2015) ldquoChangesin burnout and satisfaction with work-life balance in physicians and the general US workingpopulation between 2011 and 2014rdquo Mayo Clinic Proceedings Vol 90 No 12 pp 1600-1613doi 101016jmayocp201508023

Shanafelt TD Balch CM Bechamps G Russell T Dyrbye L Satele D Collicott P Novotny PJSloan J and Freischlag J (2010) ldquoBurnout and medical errors among American surgeonsrdquoAnnals of Surgery Vol 251 No 6 pp 995-1000 doi 101097SLA0b013e3181bfdab3

Shortell SM (2004) ldquoIncreasing value a research agenda for addressing the managerial andorganizational challenges facing health care delivery in the United Statesrdquo Medical CareResearch and Review Vol 61 No 3 pp 12S-30S doi 1011771077558704266768

Siggelkow N and Rivkin JW (2006) ldquoWhen exploration backfires unintended consequences ofmultilevel organizational searchrdquo Academy of Management Journal Vol 49 No 4 pp 779-795doi 105465AMJ200622083053

Solet DJ Norvell JM Rutan GH and Frankel RM (2005) ldquoLost in translation challenges andopportunities in physician-to-physician communication during patient handoffsrdquo Academicmedicine journal of the Association of American Medical Colleges Vol 80 No 12 pp 1094-1099doi 10109700001888-200512000-00005

Temkin-Greener H Gross D Kunitz SJ and Mukamel D (2004) ldquoMeasuring interdisciplinary teamperformance in a long-term care settingrdquo Medical Care Vol 42 No 5 pp 472-481 doi 10109701mlr000012430628397e2

Tuckman BW and Jensen MAC (1977) ldquoStages of small-group development revisitedrdquo Group ampOrganization Studies Vol 2 No 4 pp 419-427 doi 101177105960117700200404

Wagner EH Glasgow RE Davis C Bonomi AE Provost L McCulloch D Carver P and Sixta C(2001) ldquoQuality improvement in chronic illness care a collaborative approachrdquoJoint Commission Journal on Quality and Patient Safety Vol 27 No 2 pp 63-80

Walshe K and Rundall TG (2001) ldquoEvidence-based management from theory to practice in healthcarerdquo The Milbank Quarterly Vol 79 No 3 pp 429-457 doi 1011111468-000900214

Wasserman S and Faust K (1994) Social Network Analysis Methods and Applications CambridgeUniversity Press New York NY doi 101525ae1997241219

Corresponding authorFrancesca Grippa can be contacted at fgrippanortheasternedu


2251


teams

Letrsquos play the patients musicA new generation of performance measurement

systems in healthcareSabina Nuti Guido Noto Federico Vola and Milena Vainieri

Institute of Management Scuola Superiore SantrsquoAnna Pisa Italy

AbstractPurpose ndash Current performance measurement systems (PMSs) are mainly designed to measure performanceat the organizational level They tend not to assess the value created by the collaboration of multipleorganizations and by the involvement of users in the value creation process such as in healthcareThe purpose of this paper is to investigate the development of PMSs that can assess the population-basedvalue creation process across multiple healthcare organizations while adopting a patient-based perspectiveDesignmethodologyapproach ndash The paper analyzes the development of a new healthcare PMSaccording to a constructive approach through the development of a longitudinal case study The focus is onthe re-framing process of the PMS put in place by a large group of Italian regional health systems that haveadopted a collaborative assessment frameworkFindings ndash Framing information according to the population served and the patientsrsquo perspective supportsPMSs in assessing the value creation process by evaluating the contribution given by the multipleorganizations involved Therefore it helps prevent each service provider from working in isolation andavoids dysfunctional behaviors Re-framing PMSs contributes to re-focusing stakeholdersrsquo perspectivetoward value creation legitimizes organizational units specifically aimed at managing transversalcommunication cooperation and coordination supports the alignment of professionalsrsquo and organizationsrsquogoals and behaviors and fosters shared accountability among providersOriginalityvalue ndash The paper contributes to the scientific debate on PMSs by investigating a case that focuseson value creation by adopting a patient-centered perspective Although this case comes from the healthcaresector the underlying user-centered approach may be generalized to assess other environments processes orcontexts in which value creation stems from the collaboration of multiple providers (integrated co-production)Keywords Performance measurement systems Health care management Inter-organizational performancePatient-based perspectivePaper type Research paper

IntroductionPerformance measurement systems (PMSs) can be defined as a set of conceptual tools aimedat defining controlling and managing both the achievement of end-results (output oroutcomes) as well as the means used to achieve these results at various levels (eg societalorganizational and individual) (Broadbent and Laughlin 2009) These tools represent a keyfeature in every evidence-based management (EBM) process (Booth 2006) EBM promotesthe collection and use of performance measures and information in order to provide allstakeholders with evidence regarding the needs resources used and results obtained(Walshe and Rundall 2001 Lomas and Brown 2009) Without the support of PMSs decisionmakers and other stakeholders would not have evidence of whether the results achieved areconsistent with strategies and whether they are moving in the right direction (Marr 2006)

The first PMSs arose with the emergence of mass manufacturing models during theindustrial age (Bourne 2001 Bititci et al 2012) Since then these tools have evolved tomatch the changing needs of organizations and society both in the private and publicsectors (Radnor and Mcguire 2004)

According to Wilcox and Bourne (2002) and Bititci et al (2012) there are three mainphases of PMS evolution The first one (1890ndash1980) was developed from cost andmanagement accounting systems (Wilcox and Bourne 2002 Arnaboldi et al 2015)as part of which the ldquobudgetary controlrdquo form of performance measurement emergedThe PMSs developed in this period were designed to deal with the vertical hierarchy


Received 29 September 2017Revised 7 May 20187 May 2018Accepted 22 May 2018


2252

MD5610


principle that characterized organizations at that time and a distribution of powerconsistent with the organizational structure (Bititci et al 2012)

The second phase of performance measurement started in the 1980s and was aimed atovercoming the exclusive adoption of a financial perspective including multiple dimensionsof performance (Hayes and Abernathy 1980 Wilcox and Bourne 2002 Bititci et al 2012)During this phase the first ldquointegrated performance measurementrdquo systems were developedin order to deal with the switch from bureaucracy to adhocracy occurring in private andpublic organizations at that time

The third and most recent phase ( from the mid-1990s) was driven by the need to linkkey performance indicators to strategy (Kaplan and Norton 1992 1996 Wilcox andBourne 2002) In this period measurement started to be conceived as a tool to facilitatestrategic management practices in organizations eg mapping the process of value creationwithin and later on beyond organizational boundaries (Bititci et al 2012)

In the last few years the management literature has shown significant interest inanalyzing the opportunities and challenges of performance measurement applications ininter-organizational settings (Bititci et al 2012 Anderson and Dekker 2015 Dekker 2016)This increasing attention has coincided with a significant growth in collaborativerelationships between organizations in both the private (Anderson and Sedatole 2003Dekker 2016) and public sectors (Brignall and Modell 2000 Christensen and Laegreid 2007Bianchi 2010 Kurunmaumlki and Miller 2011 Halligan et al 2012)

Due to the institutional fragmentation characterizing the public sector the literature(see among others Ryan and Walsh 2004 Christensen and Laegreid 2007 Moore 2013Cuganesan et al 2014) has identified a need to focus performance measurement on anassessment of the value creation process and consequently to go beyond the organizationalboundaries and adopt a network perspective This trend in the design of PMSs is alsohappening in healthcare and the most recent evidence shows that this sector is evenanticipating many of the global dynamics and challenges

Healthcare systems are characterized by an intrinsic complexity derived from bothgovernance fragmentation as well as uncertainty pluralism and a multidisciplinaryenvironment (Plsek and Greenhalgh 2001 Lemieux-Charles et al 2003 Ramagem et al 2011)

Dealing with this complexity requires collaborative approaches among stakeholders inorder to better respond not just to patients and service users but also to the needs of thewhole population from a system perspective (Nuti Bini Ruggieri Piaggesi and Ricci 2016Gray et al 2017)

This paper focuses on performance measurement challenges and future perspectives inhealthcare The aim is to analyze how the healthcare system has followed the path of theglobal trend and how it can contribute to the research agenda of performance measurementThe paper provides the results of a constructive analysis of the evolution of PMSs based ona longitudinal case of the re-framing of the PMS by a large group of regional healthcaresystems in Italy that have adopted a network framework

The next section contextualizes the performance measurement and managementchallenges in the healthcare sector outlining its distinguishing characteristics The thirdsection presents the methodology and then the Italian case study on which this paper isbased and the fourth section explores its re-framing process The discussion andconclusions are then developed in the final sections

The evolution of PMSs in healthcareUntil the introduction of the New Public Management (NPM) paradigm the public sectors ofwestern countries adopted Weberrsquos model of ideal bureaucracy (Hood 1991 OrsquoFlynn 2007)whose system of control focused on input monitoring and process compliance (Head andAlford 2015)

2253

Performancemeasurement

systems

Management accounting forms of control were gradually introduced in the publichealthcare sector following the NPM reform of the 1980s which promoted the use of privatesector practices throughout the west (Hood 1991 Brignall and Modell 2000) The aim wasto overcome the shortcomings of the traditional paradigm of public administration basedon bureaucracy that did not focus on efficiency or results (Hood 1991 OrsquoFlynn 2007)Several healthcare public systems thus introduced the first generation of ldquobudgetarycontrolrdquo measurement systems mainly focused on financial measures volumes of servicesprovided and organizational responsibility assessments (Chua and Preston 1994 Ballantineet al 1998 Arnaboldi et al 2015 Naranjo-Gil et al 2016) This phase also known asldquomanagerialismrdquo or ldquomanaging for resultsrdquo led to the breakdown of organizations intovarious business units controlled by setting goals and monitoring performance resultsstressing departmentsrsquo productivity (Bouckaert and Halligan 2008 Head and Alford 2015)Although this generation of PMSs helped to overcome the bureaucratic model itstrengthened a ldquosilordquo structure where each provider and each organizational unit operatingin the healthcare system was monitored according to both the volume of activities(eg number of treatments) and financial measures such as revenues and costsThis approach frequently created internal competition within institutions especially interms of the allocation of financial resources (Chua and Preston 1994 Christensen andLaegreid 2007 Head and Alford 2015)

The strong focus on financial performance and the attribution of responsibilities toorganizational units of first generation PMSs limited the ability of healthcare stakeholdersto assess performance according to the public value paradigm which in the last fewdecades has become the reference paradigm of public administrations (OrsquoFlynn 2007Cuganesan et al 2014) Public value is a multidimensional construct that primarily resultsfrom government performance (Moore 1995 Bryson et al 2014) In healthcare publicvalue has been defined as the relationship between outcomes and resources (Porter 2010)from a population-based perspective (Gray and El Turabi 2012) The identification of valueas the key objective of healthcare systems (Porter 2010 Gray and El Turabi 2012Gray et al 2017 Lee et al 2017) requires PMSs to shift their focus toward the assessment ofhealth organizationsrsquo ability to take decisions and actions that effectively create and delivervalue to the reference population (Naranjo-Gil et al 2016) Population value in health caredoes not correspond to the volume of services delivered or the outcome achieved for thetreated patients but is the ability of the healthcare system to provide care to the people thatcould benefit most from it (Gray et al 2017)

In fact it is not uncommon for health services to be also provided to people that do not needthem and thus wasting resources (see Figure 1mdashgray area) Moreover the healthcare systemmay not be able to identify and provide care to those most in need (see Figure 1mdashwhite area)From the perspective of effectiveness healthcare systems create value for the population when

People who havereceived care

services

People who couldbenefit more from

care

PopulationValue

Source Adapted from Gray et al (2017)

Figure 1Population value

2254

MD5610

the people treated are those that benefit the most from the treatment (see Figure 1mdashblack area)(Gray et al 2017)

Performance measurement is thus required to overcome the traditional focuson the financial dimension and support a population value-based approach toperformance assessment

PMS in health care has thus followed the recommendations of many authors(Van Peursem et al 1995 Leggat et al 1998 Aidemark 2001 Arah et al 2006 Nuti et al2013) by implementing what Bititci et al (2012) have called ldquoIntegrated PerformanceMeasurement Systemsrdquo

This generation of PMSs in healthcare is characterized by

bull Multi-dimensionality PMSs provide measures that go beyond volumes of activitiesand financial aspects and are based on indicators related to structure processquality of care and equity from a population-based perspective and also the systemrsquosfinancial sustainability (Donabedian 1988 Ballantine et al 1998 Leggat et al 1998Arah et al 2006 Nuti et al 2013)

bull Evidence-based data collection and information provision providing support forstakeholders in decision making (Sackett et al 1996)

bull Shared design all stakeholders and particularly health professionals should beinvolved in providing insights and suggestions (eg new indicators revision ofexisting indicators) in a continuous fine-tuning process (Leggat et al 1998Nuti Vola Bonini and Vainieri 2016)

bull Systematic benchmarking of results benchmarking among providers and amonggeographic areas should be ensured in order to shift from monitoring to evaluation(Nuti et al 2013)

bull Transparent disclosure to stimulate data peer-review and together with systematicbenchmarking to leverage professional reputation (Hibbard 2003 Bevan andWilson 2013 Nuti Vola Bonini and Vainieri 2016 Bevan et al 2018)

bull Timeliness to allow policy makers to make decisions promptly and to increase trust inindicators (Davies and Lampel 1998 Bevan and Hood 2006 Wadmann et al 2013)

However even these PMSs present some limitations in addressing the new challengesof performance measurement because they are mainly designed according to anindividual healthcare providerrsquos perspective whereas most services are delivered topatients thanks to inter-organizational (ie across providers) relationships Especially inepidemiological conditions (eg chronic diseases cancer mental illnesses) the process ofvalue creation can only be measured effectively by assuming the value-delivery chainperspective which in healthcare corresponds to the patientsrsquo clinical pathwaysAs such the adoption of a care pathway perspective is pivotal in assessingperformance and consequently guiding policy makers and other stakeholdersrsquo actions(Nuti Bini Ruggieri Piaggesi and Ricci 2016)

Dealing with care pathways entails creating horizontal inter-organizational networks toallow coordination between health professionals across organizational boundaries Thesenetworks which may or may not be officially recognized are usually organized to take careof the patient along the different phases of the pathway The relationships among networkcomponents are characterized by interdependence complexity and continuous change andthe absence of a clear hierarchy makes their assessment problematic (van der Meer-Kooistraand Scapens 2008)

The management literature on performance assessment has tended to focus oninter-organizational performance assessment at the single-institution level (Cuganesan

2255


systems

et al 2014 Dekker 2016) Kurunmaumlki and Miller (2011) outlined the need to broaden thestudy of inter-organizational relations and performance management to include not onlyorganizational forms but the practices and processes through which they are madeoperable eg pathways

The limitations of current PMSsmdashwhich are related to collecting and displayingexclusively performance data from an organizational perspective (eg regional health systemlocal health authorities hospitals)mdashare linked to the risk of shifting professionalsrsquo attention tosub-optimal performance rather than delivering value to patients thus leading to performancedistortions and strategic inconsistency (Meyer and Gupta 1994 Van Thiel and Leeuw 2002Melnyk et al 2013) A lack of alignment between strategy and performance evaluationsystems may result in ldquoperformance trapsrdquo or ldquoperformance paradoxesrdquo (Meyer andGupta 1994 Van Thiel and Leeuw 2002 Lemieux-Charles et al 2003 Bevan and Hood 2006Wadmann et al 2013) Performance traps are related to narrow views and uses ofmeasurement which may lead for example to sub-optimization ( focusing on localperformance results rather than overall system goals) myopia ( focusing on short-term targetsat the expense of longer-term objectives) and tunnel vision (the narrowing of managerialattention) (Van Thiel and Leeuw 2002 Bevan and Hood 2006 Wadmann et al 2013Nuti Vainieri and Vola 2017) This is even more evident in highly fragmented governancestructures (Noto and Bianchi 2015)

There is thus a need for a PMS that measures the value created for the population(as with second generation PMSs) and also takes into account the patient perspectiveThis implies that PMSs in health should consider horizontal relationships betweenhealthcare organizations and professionals and mitigate professional and organizationalbarriers to networking (Berry 1994 van der Meer-Kooistra and Scapens 2008 Kurunmaumlkiand Miller 2011 Cuganesan et al 2014 Dekker 2016)

A key element in dealing with these challenges is the way performance data arereported so as to foster the sharing of results among stakeholders (Bititci et al 2016)The use of appropriate communication channels such as an effective visual system iscrucial in order to create commitment to achieving the desired performance andappropriate behaviors throughout all organizational levels (Kaplan and Norton 1992Otley 1999 Bititci et al 2016)

Performance visualization concerns the representation and framing of datainformation and knowledge in a graphical format which may lead to new insights andan understanding of the performance of the organizationsystem analyzedthus fostering stakeholder commitment to the strategic goals of the organization(Tversky and Kahneman 1981 Lengler and Eppler 2007) In fact since people are drivenby bounded rationality evidence-based decision-making is intrinsically mediated by theway evidence itself is communicated According to Bititci et al (2016) effective visualsystems for strategic and performance management support strategy developmentand implementation performance reviews internal and external communicationcollaboration and integration among different units and levels cultural changesand innovation

In order to benefit from PMSs performance information thus needs to be framed andcommunicated consistently with the aims and strategies (Teece 1990 Pettigrew 1992Bititci et al 2016) of health systems (Nuti et al 2013)

A shift from a single-organization performance assessment to an inter-organizationalassessment requires the integration of measures and representations that map the servicedelivery process that the network has to put in place which in the case of healthcare meansthe patient pathway PMSs are thus required to represent performance informationaccording to the goal of the system that is being measured (eg fostering collaborativepractices networking and shared accountability)

2256

MD5610

MethodThis paper describes the results of a longitudinal constructive study carried out in Italy onthe evolution of the Italian Regional Performance Evaluation System (IRPES) in healthcare

The IRPES was initially developed in 2004 thanks to a collaboration between theMes-LabmdashInstitute of Management of SantrsquoAnna School of Advanced Studies and theregional health system in Tuscany (Italy) Since 2008 the IRPES has been shared by manyother regional health systems in Italy so that they can benchmark their results against eachothersrsquo (Nuti et al 2013 Nuti Vola Bonini and Vainieri 2016) The IRPES is currently (2018)adopted by 11 Italian regions and two autonomous provinces (Apulia Basilicata CalabriaEmilia Romagna Friuli Venezia Giulia Liguria Lombardy Marche Tuscany UmbriaVeneto the Autonomous province of Bolzano the Autonomous province of Trento) coveringaround 190 health organizations providing health services for about 20 million inhabitantsThis PMS is currently used by these regional systems when producing regulations definingthe objectives and priorities of their health systems Some of these regulations have beendirectly based on the evidence produced by the IRPES[1]

What distinguishes the IRPES from other PMSs is the voluntary-based adoption byregional health systems and the role of the Mes-Lab in facilitating the continuousdevelopment of new analyses and tools to support stakeholders in interpreting data(Nuti and Vainieri 2016 Nuti Vainieri and Vola 2017)

TheMes-Lab has played a primary key role in both the development and the re-framing of theIRPES The constructive approach adopted aims to solve issues through the direct involvement ofresearchers in several phases of the innovation process such as testing solutions (Kasanen et al1993 Labro and Tero-Seppo 2003) The constructive approach is widely used in technicalsciences mathematics operations analysis and clinical medicine as well as in managementresearch (Kasanen et al 1993 Noslashrreklit et al 2016) The use of the constructive approach has shedlight on the principal issues involved in measuring and interpreting results Since the IRPES wasfirst set up the research group has interacted with policy makers managers and professionals ofthe health care sector The solutions implemented were thus designed to overcome its shortfallsThis paper discusses the contribution to the literature from this experience

The Italian Regional Performance Evaluation SystemThe IRPES system is made up of more than 300 indicators which measure themultidimensional performance of each healthcare organization The following aremonitored health status of the population capacity to pursue regional strategies clinicalperformance efficiency and financial performance patient satisfaction and staffsatisfaction (Nuti et al 2013) The indicators are calculated yearly using administrativedatabases The aim of the IRPES is to assess and monitor health system performanceat a regional and local level indicators are computed with regional and local granularity(both local health authorities and teaching hospitals)

The regional health systems adhering to the IRPES share a collaborative andconstructive approach with each other and with the Mes-Lab research group they discussthe definition of the indicators and on how they should be calculated Each regional healthsystem is responsible for processing its own data

About half of the 300 indicators are evaluated by comparing their results withinternational or nationallocal standards All regional health systems use the samestandards referring to the scientific literature normative standards or where these arelacking to the distribution of each indicator among health authorities Performance istherefore assessed according to five different performance tiers ranging from the worst(0mdashred) to the best (5mdashdark green)

Results are publicly disclosed through an open-access website and through an annualreport[2]

2257


systems

Each indicator is depicted using a wide range of graphical solutions Figure 2 useshistograms to report the results of one of the indicators used in the IRPES (ie waiting timesfor malignant breast cancer intervention)

The IRPES also exploits georeferencing data in order to display cartographicrepresentations (see Figure 3) Such graphical solutions depicting the performanceassociated with a specific geographical area are aimed at assessing value creation forgeographically delimited population groups

40

30

20

Day

s

10

0

40

30

20

Day

s

10

0

200

150

100

Day

s

10

0

Bol

zano

Lom

bard

ia

Tren

to

Ven

eto

Em

ilia-

Rom

agna

Pug

lia

FV

G

Um

bria

Tosc

ana

Ligu

ria

Mar

che

Bol

zano

Lom

bard

ia

Tren

to

Ven

eto

Em

ilia-

Rom

agna

Pug

lia

FV

G

Um

bria

Tosc

ana

Ligu

ria

Mar

che

2015

2016

AS

L F

oggi

aO

sped

ale

Val

duce

Azi

enda

PA

Bol

zano

AS

ST

di L

odi

AU

LSS

2 F

eltr

eU

SL

Um

bria

2A

SS

T D

el G

arda

OO

RR

Fog

gia

AS

L Le

cce

orig

gia

Pel

asci

ni -

Gra

v

AU

LSS

10

Ven

eto

Or

AU

LSS

7 P

ieva

di S

olig

oIR

CC

S P

ol S

Mat

teo

AS

ST

dei

Set

te L

aghi

EE

Cas

a S

ollie

voA

ULS

S 1

4 C

hiog

gia

AU

SL

1 Im

perie

seA

SS

T R

hode

nse

AS

ST

Di C

rem

aA

SS

T S

anti

Pao

lo e

Car

loIR

CC

S C

entr

o R

if O

ncol

A

ULS

S 9

Tre

viso

AU

SL

Sud

Est

Friu

li O

ccid

enta

leA

ULS

S 1

2 V

enez

iana

AS

ST

Di V

imer

cate

AO

Ter

niA

O R

eggi

o E

mili

aS

Raf

fael

e -

Mi

Mac

erat

aIs

tPol

iclS

Don

ato

Sen

igal

liaF

erm

oA

SU

I Udi

neA

US

L M

oden

aA

US

L 4

Chi

avar

ese

AU

SL

Cen

tro

AO

U M

oden

aA

O P

erug

iaA

ULS

S 6

Vic

enza

Osp

Eva

ngel

ico

AS

L B

rindi

siA

SL

Bar

letta

-And

ria-T

rani

Source 2016 datamdashavailable at httpperformancesssupitnetval

C1041 Waiting times for malignant breast cancer intervention mdash 2016



Figure 2Waiting times formalignant breastcancer intervention

2258

MD5610

In order to provide an overview of each organizationrsquos performance the wholeset of indicators is currently composed of a subset of ldquomacro-indicatorsrdquo which isrepresented through a target chart (a ldquodartboardrdquo) in which the highest scores(dark-green band) are positioned in the center and the lowest ones (red band) are in theouter circle

Figure 4 shows an example of the Friuli Venezia Giulia resultsAccording to the taxonomy reported in first section IRPES can be considered

as an integrated performance management system (Bititci et al 2012 Nuti et al 2013Nuti Vola Bonini and Vainieri 2016) It can be deemed to comply with the set of proceduralrequirements mentioned above

bull Multi-dimensionality this goes beyond the assessment of financial sustainability andconsiders measures related to clinical processes appropriateness quality of carepatient satisfaction and staff satisfaction

bull Evidence-based data collection and information provision the IRPES is based onboth administrative data and data collected ad hoc whose standardization andnormalization follows rigorous and standard scientific criteria

bull Systematic benchmarking the PMS described here compares the performanceacross regional health systems and providers on a yearly basis The evaluation foreach indicator is based on gold standards or on the distribution of results across theorganizations participating in the system

bull Transparent disclosure the IRPES is publicly reported annually both through aprinted report and via the web (httpperformancesssupitnetval)

bull Timeliness data and indicators are collected and calculated every year and publiclydisclosed within six months from the end of the reference year

Because of this design and the effective visual representation the system has aidedregional and local organizations in improving their performance and reducing


Figure 3Cartographic

representations ofwaiting times formalignant breast

cancer intervention

2259


systems

unwarranted variations (Nuti Vola Bonini and Vainieri 2016) The IRPES has stimulatedprofessionals and other stakeholders to focus on population value creation through theinclusion of a large set of outcome measures also by considering the residentsrsquogeographical area

However the IRPES is currently anchored to an ldquoorganization-focusedrdquo perspectiveie it monitors and reports each unit and organization performance separately Althoughevidence provided by this measurement system is key to assessing organizationperformance focusing on the single tiles may be misleading given that patientsrsquo carepaths that generally cross different care settings In reality emerging healthcare needsrequire coordinated responses and shared responsibility by a wide range of providersThus evaluation systems need to be reframed accordingly in order to detect thecontribution of all the links of the healthcare value chain and to highlight the sharedresponsibility of the different organizations contributing to the care pathway

Populationrsquos health mdash 2010ndash2012

A4 Suicide mortality

B2Healthylifestylespromotion

F19Costfordiagnostictests

F18Averagecostforhospitalcare

F17Healthexpenditure F15

WorkplaceHealthandSafety

F10bPharmaceuticalgovernance

F12aDrugprescriptionefficiency

D18AMAdischarges

D9EmergencyDepertmentLWBS

C21Pharmaceuticalcompliance

C18Electivesurgeryappropriateness

C16EmergencyDepartment

C15Mentalhealth

C13aDiagnosticappropriatenessC11a

Chroniccaremanagement

C8aHospital-primarycareintegration

C7Maternalandchildcare

C5Qualityofthecareprocess

C4Surgicalappropriateness

C14Appropriatenessofcare

C2aCLOS(surgicalDRGs)

C2aMLOS(medicalDRGs)

C1Demandmanagement

B28Homecare

B7Vaccinecoverage

B5Cancerscreenings

B4Opioidconsumption

C10Oncologicalpathway

C9Appropriateprescribingofmedication

A2 Cancer mortality A10 Lifestyles A1 Infant mortality A3 Circulatory disease mortality


Figure 4An example of theFriuli Venezia GiuliaRegion IRPESdartboard

2260

MD5610

To overcome these limitations the IRPES now takes into account the population value chainperspective The next section describes the re-framing process that has been implemented inorder to integrate the organizational perspective with the patient-based perspective

Re-framing the IRPESAfter a decade of IRPES use the research team together with the regional stakeholdersrecognized the need to analyze performance information also at a pathway level

In order to offer an effective graphical representation by shifting the focus from singleorganizationsrsquo perspective to care pathways results the original graph (ie the dartboard) wasintegrated with a new tool that represents the care pathwaysrsquo performance by relying on themetaphor of the ldquostaverdquo ie the set of horizontal lines and spaces used in sheet music Both themetaphors share a common characteristic they hint at a ldquopositiverdquo allusion by referring torecreational and artistic activities This is intended to stimulate a favorable approachby the user especially by leveraging on the framing effect (Tversky and Kahneman 1981)The metaphor of the stave conveys is intended to transmit the message that the health caresystem should play the patientsrsquo music following step by step hisher pathway

As shown in Table I a selection of the original indicators used in the IRPES wererepositioned according to the different phases that the patients cross along the pathways(Nuti De Rosis Bonciani and Murante 2017) So far five pathways have been selectedaccording to their relevance the maternal and pediatric pathway the oncological pathway thechronic diseases pathway the mental health pathway and the emergency care pathwayTheir design involved the selection of the most appropriate indicators in order to effectivelyrepresent the different phases each care path is composed of

As an example the case of the oncologic pathway is reported and describedThe stave like the dartboard uses five color bands ( from red to dark-green)

These bands are now displayed horizontally and are framed to represent the differentphases of care pathways This view allows users to focus on the strengths and weaknessesthat characterize the healthcare service delivery in the different pathway phases

In order to further investigate performance according to a patient-based perspective thisstructure has been integrated with patient-related experience measures (PREMs) and in thenear future it will also consider patient-related outcomes measures (PROMs)mdashcurrently inthe experimental phase These measures are calculated by conducting standardized andcontinuous surveys with patients to get their feedback on outcomes and care experiencesThese surveys assess quality of life and patient outcome (PROMs) during pre-treatmentstreatments and follow-up phases and patient experiences (PREMs) by collecting data oninformation and support received during access to care (eg screening) treatments(eg surgery) and follow-up

Staves are designed to display the pathwaysrsquo performance both at regional and locallevels Regional pathways report regional performance without detailing the providersLocal pathways instead show performances achieved by each provider in a geographicalarea in order to highlight the individual contribution to the overall care pathway and tofocus the viewerrsquos attention on (joint) value creation for each local area population

As shown in Figures 5 and 6 each dot reports the evaluation associated withthe performance achieved by each provider (colors represent different organizations) in thegeographical area with regards to the pathwayrsquos indicators

The dots on the stave are thus associated with the name of different healthorganizations In Tuscany (Figure 6) the performance of both the local health authorityrsquos(AUSL Centro) and an autonomous hospital (AOU Careggi) are reported in thePadua area three providers cooperate to provide oncological care and are therefore jointlyreported by the stave the local health authority (AULSS 16 Padova) and two autonomoushospitals (AO Padova and IOV)

2261


systems

By adopting a pathway perspective the stave meets two goals First it steers the userrsquosattention toward the patient perspective by embracing the value creation paradigmSecond by showing the performance of the different organizations that servethe population of a geographical area in each pathway phase the stave highlights thecontribution that each organization provides stressing joint responsibility in the overallresults of the care pathway Thus it is easier for the stakeholders of the healthcaresystem to understand the criticalities in delivering value to their reference populationThrough this visual representation managers may be able to assess the performance of

Oncologic pathway

ScreeningB511 Screening extension breastB512 Screening adhesion breastB514 Voluntary screening adhesion breastB515 women visited within 20 days from positive screeningB516 visit adhesion after positive screeningB521 Screening extension cervixB522 Screening adhesion cervixB524 Voluntary screening adhesion cervixB531 Screening extension rectal colonB532 Screening adhesion rectal colonB535 Voluntary screening adhesion rectal colon

DiagnosisC105 Prescriptive appropriateness of tumor biomarkers

TreatmentC1041 Waiting times for malignant breast cancer interventionC1042 Waiting times for malignant prostate cancer interventionC1043 Waiting times for malignant colon cancer interventionC1044 Waiting times for malignant rectum cancer interventionC1045 Waiting times for malignant lung cancer interventionC1046 Waiting times for malignant uterus cancer interventionC1711 Percentage of admissions over the volume threshold for breast cancerC1712 Index of dispersion of cases in wards under the volume threshold for breast cancerC1751 Percentage of admissions over the volume threshold for prostate cancerC1752 Index of dispersion of cases in wards under the volume threshold for prostate cancerC1021 of breast-conserving surgeries (nippleskin sparing) for breast cancerC1022 of women who undergo sentinel lymph node excisionC10221 of women who undergo radical axillary lymph node excisionC1024 of women treated with radiotherapy within 4 month from breast surgeryC1025 Administration within 8 weeks of chemotherapy in subject with breast cancerC1031 of patients undergoing re-intervention within 30 days of hospitalization for colon (three-year)C1032 of patients undergoing re-intervention within 30 days of hospitalization for rectum (three-year)C1033 Administration within 8 weeks of chemotherapy in subject with colon cancerC1061 of men undergoing radiotherapy who begin treatment within 6 months from interventionF1021c Average expenditure for oncology medicines (local health authority)F1021d Average expenditure for oncology medicines (hospital)

End of lifeC281 of deceased oncologic patients within the palliative care networkC282 of patients with maximum waiting time between reporting and hospitalization in hospice

⩽3 daysC282b of oncologic patients with maximum waiting time between reporting and hospitalization in

hospice ⩽3 daysC283 of hospice admissions with a period of hospitalization greater than 30 days

Table IList of the indicatorsthat constitute theoncological pathwaygrouped according tothe different phasesbased onadministrative data

2262

MD5610

Scr

eeni

ngex

tens

ion

brea

st

Wai

ting

time

hosp

ice

Scr

eeni

ngD

iagn

osis

Trea

tmen

tE

nd o

f Life

Two

or m

ore

orga

niza

tions

hav

ing

the

sam

e ev

alua

tion

AO

Pad

ova

AU

LSS

16

Pad

ova

Ist

Onc

Ven

eto

(IO

V)

Scr

eeni

ngex

tens

ion

cerv

ix

Scr

eeni

ngad

hesi

onbr

east

Scr

eeni

ngad

hesi

once

rvix

Scr

eeni

ngex

tens

ion

colo

n

Scr

eeni

ngad

hesi

onco

lon

App

ropr

bi

omar

ker

Wai

ting

time

trea

tbr

east

Wai

ting

time

trea

tpr

osta

te

Wai

ting

time

trea

tco

lon

Wai

ting

time

trea

tre

ctum

Wai

ting

time

trea

tut

erus

Adm

issi

ons

over

thre

shol

dbr

east

Inde

x of

disp

ersi

onbr

east

Adm

issi

ons

over

thre

shol

dpr

osta

te

Inde

x of

disp

ersi

onbr

east

o

f(n

ippl

esk

insp

arin

g)su

rger

ies

Pal

liativ

eca

rene

twor

k

012345

Evaluation Sour

ce 2

016

data

mdashav

aila

ble

at h

ttp

perf

orm

ance

sssu

pit

netv

al

Figure 5An example ofthe stave in the

geographical area ofPadova (Veneto)

2263


systems

012345

Evaluation

Scr

eeni

ngD

iagn

osis

Trea

tmen

tE

nd o

f Life

Two

or m

ore

orga

niza

tions

hav

ing

the

sam

e ev

alua

tion

AO

U C

areg

giA

US

L C

entr

o

Scr

eeni

ngex

tens

ion

brea

st

Scr

eeni

ngex

tens

ion

cerv

ix

Scr

eeni

ng

adhe

sion

brea

st

Scr

eeni

ngad

hesi

once

rvix

Scr

eeni

ngex

tens

ion

colo

n

Scr

eeni

ngad

hesi

onco

lon

App

ropr

bi

omar

ker

Wai

ting

time

trea

tbr

east

Wai

ting

time

trea

tpr

osta

te

Wai

ting

time

trea

tco

lon

Wai

ting

time

trea

tre

ctum

Wai

ting

time

trea

tlu

ng

Wai

ting

time

hosp

ice

Adm

issi

ons

over

thre

shol

dbr

east

Inde

x of

disp

ersi

onbr

east

Adm

issi

ons

over

thre

shol

dpr

osta

te

Inde

x of

disp

ersi

onbr

east

o

f(n

ippl

esk

insp

arin

g)su

rger

ies

Pal

liativ

eca

rene

twor

k

Wai

ting

time

trea

tut

erus

Sour

ce 2

016

data

mdashav

aila

ble

at h

ttp

perf

orm

ance

sssu

pit

netv

al

Figure 6An example ofthe stave in thegeographical area ofcentral area (Tuscany)

2264

MD5610

the service supply in the various phases that make up a care pathway and consequentlyto attribute co-responsibilities to the multiplicity of providers involved in the servicedelivery of each phase

As previously mentioned the stave is currently adopted by 13 health systems (11 regionsand 2 autonomous provinces) These pathways can be viewed both at the regional andat the intra-regional ie geographical area level The performance achieved by the81 geographical areas which reflect the perimeters of the local health authorities ofthe network-adhering regions is publicly disclosed so that local populations can assessthe value created (wwwperformancesssupitnetval)

DiscussionThe previous section described the development of a major performance evaluationsystem in Italy starting from its design in 2004 till the most recent developments in 2017There have been two main phases

(1) The IRPES was first created in 2004 in Tuscany in order to integrate financialinformation concerning the regional health care system with evidence on qualityequity efficiency appropriateness effectiveness and responsiveness The aim wasto make such information available to stakeholders in the healthcare system(regional managers and administrators professionals patients citizens etc)Since 2008 an increasing number of regional health systems in Italy have beenadopting the same IRPES resulting in an inter-regional performance comparison

This comparison was enhanced by integrating the original financial dimensionswith the others and by enlarging the range of monitored units Consequently healthcare institutions have been monitored in terms a wider range of perspectives andbenchmarked against a growing number of comparable providers

Comparing this phase with the previously mentioned theoretical frameworks onPMS this transition reflects first the introduction of a ldquobudgetary controlrdquo approach(measuring financial performance of the systemrsquos units) and subsequently its shifttoward ldquointegrated performance measurementrdquo (measuring the multidimensionalperformance of the systemrsquos units) (Chua and Preston 1994 Ballantine et al 1998Bititci et al 2012 Naranjo-Gil et al 2016) The focus of the performance evaluationprocess has been the same throughout the ten years of the project health careorganizations in their different granularity (regions health authorities hospitalshealth districts etc) The limitations encountered adopting this approach were thusrelated to the difficulty of assessing the value created by the joint actions of theproviders involved in the health service delivery

(2) In 2016 the IRPES was reframed in order to collect and to report data that analyzeand illustrate the performance achieved by one or more providers The key toanalyzing the activity of a network of health care providers involved in theservice delivery is to adopt a patient-based perspective (Gray and El Turabi 2012Nuti Vola Bonini and Vainieri 2016) The IRPESrsquos analytical focus has integratedthe evaluation of individual institutions with the evaluation of patient care pathsThe introduction of a new data visualization toolmdashthe above-mentionedstavemdashillustrates the theoretical foundations of this integrative perspective Thusthe new PMS enables the adoption of the patient care paths perspective ie clinicalactivities performed by multiple providers in order to take care of complex healthproblems that require clinical assistance and coordination over time

The PMS evolution should be interpreted according to the modifications of the ldquocontextrdquo thePMS is developed in (Bititci et al 2012) Phase 2 above reflects the dynamic process of

2265


systems

alignment of the IRPES to the evolving contextual institutional organizational andstrategic situation

Since this paper deals with PMSs in the health care sector the context analysis needs tocarefully assess the recent revolutionary shiftmdashpartially due to ICT innovationmdashconcerningthe patientsrsquo role in steering their health care choices and related outcomes (Richards et al 2013)The transition from Phases 1 to 2 was aimed at fine-tuning the performance evaluation processwith the opportunities offered by the patientsrsquo new role

Integrating the previous perspective with a new approach aimed at assessing healthcareorganizationsrsquo performance in co-producing value for patients implied designing a newarchitecture of the evaluation process While the analytical perspective remained the samethe focus shifted as a result of exploiting a multidimensional approach The interest in theoverall performance of divisional units was integrated by monitoring the performance inindividual geographical areas during specific macro-activities (care paths) that involve aplurality of organizations

In this case the theoretical taxonomy proposed by Bititci et al (2012) might be somewhatmisleading if uncritically applied to the interpretation of this process Bititci interpreted thegeneral transition of PMSs from ldquointegrated performance measurementrdquo to ldquointegratedperformance managementrdquo as a shift from ldquosingle organizationsrdquo to monitoring ldquocollaborativeorganizationsrdquo the latter intended as ldquovirtual organizations that are additional to theorganizations that are participating in the collaborative enterpriserdquo (Bititci et al 2012)The re-framing process of the described PMS should not be interpreted as an integration ofprevious performance monitoring approach by including performance implications ofautonomous but relevant organizations (such as those supporting the supply chain) Instead itrepresented the shift from an organization-focused PMS to a strategic activities-focused PMSIn other words the PMS is now assessing the ability of the health care system to manage itscore activities through the integrations of its organizations Individual institutions whichrepresented the focus of IRPES phase 1 now become an ldquoinstrumental focusrdquo Maybecounterintuitively the label coined by Bititci and colleagues to identify the most recentgeneration of PMSsmdashldquointegrated performance managementrdquomdashbetter complies with PMSs inhealth care than in other sectors their focus actually shifts from individual organizations tothe integration of individual organizations within the (health care) system

Flanking the previous organization-centered perspective with the patient-focusedapproach entailed designing an evaluation system aimed at assessing how healthcaresystems create value for their respective populations This implied assessing

(1) different providersrsquo contributions in joint value creation and

(2) value creation throughout the various phases of the care paths referring to differentcare settings and different providers

The adoption of the new perspective has therefore been pre-conditional to designing aperformance evaluation system capable of assessing two fundamental elements of valuecreation in healthcare co-production and integration

Evidence on the effectiveness of this new approach is not yet available However thereframed PMS has four possible benefits

(1) strategic re-focusing shifting the focus from organizationsrsquo performance tointegrated activitiesrsquo performance may help stakeholders become more aware ofthe ldquonewrdquo strategic goals of health care systems

(2) legitimization the new approach may contribute to legitimizing organizationalunits specifically aimed at managing transversal communication cooperationand coordination such as the above-mentioned inter-authority departments(Lemieux-Charles et al 2003)

2266

MD5610

(3) alignment since it focuses on care paths the new approach is more in line withclinical activity and therefore more easily understood and accepted by professionalsthereby fostering their engagement and

(4) shared accountability integrating the results of different providers in a singleperformance management framework fosters the shared accountability of thenetwork of organizations participating in service delivery

ConclusionsThis paper investigated the results of a constructive research experience related to thetransition of a PMS in order to identify potential improvement of PMSs in health care Due tothe active involvement of the research team in the development of the case described theapproach used in this paper did not adopt an evolutionary approach but opted for aconstructive approach being inspired by the literature on healthcare managementand PMSs the collaboration between the research team and the stakeholders allowed tore-design the IRPES starting from the patient perspective

The IRPES experience helped to reverse the deterministic and reactive interpretation ofthe relationship linking the contextual situation with the PMS aimed at evaluating itThe new role of patients in healthcare today is not merely in terms of new informationalneeds ( for instance the introduction of PROMs and PREMs) but relates to a new perspectivethat assesses two fundamental determinants of value creation in healthcaremdashieco-production and integration

In conclusion three final issues should be mentioned the toolrsquos replicability thelimitations of the research and its potential developments

In terms of the toolrsquos replicability the IRPES case suggests the need for PMSs tointegrate the classic organizational perspective with a user-centered perspective whenthe aim is to assess environments processes or contexts in which value creation stemsfrom the collaboration of multiple providers (integrated co-production)

Contingent limitationsmdashsuch as data unavailability or unreliabilitymdashmay of coursehinder the generalizability of such an instrument but do not invalidate its underlyinginnovative approach In fact the used approach may prove fundamental in evaluating areaswhere the userrsquos role is becoming essential in co-determining value creation For example

bull Other healthcare systems regardless of differences in epidemiological needsstrategic responses and institutional architecture

bull Other service-oriented areas such as education both in the public and in theprivate sector

bull Some manufacturing sectors where the customersrsquo role is relevant in valuecreation The literature tracing the evolution of PMSs usually highlights how thePMSs in the manufacturing sector and private sector have helped develop PMSs inthe service sector and public sector respectively The case described here mightrepresent a double pay back with an innovation in a service-oriented and publicsector (the Italian health care sector) paving the way for future improvements in theevolution of PMSs

With regard to potential developments of our PMS it may be useful to recall that the healthcare sector in the west experiencedmdashprobably before other sectorsmdashthe need to integratethe activities of the various organizations that jointly contribute to value creation(ie ldquointegrated co-productionrdquo) acknowledge and potentially manage the impact thatactors belonging to different but related systems (such as social care) have on the health caresystem itself

2267


systems

The re-framing of PMS accounts for the first need (inter-organizational assessment) butdoes not yet respond to the second (inter-systemic assessment) While previouscontributions called for PMSs aimed at evaluating the performance of ldquocollaborativeorganizationsrdquo the experience described here may suggest the need to design PMSs able toevaluate ldquocollaborative systemsrdquo in order to assess the reciprocal interactions connectingthe health care system the social system the environmental system and so on The newhealth care context seems to call for widening the perspective of PMSs toward anldquoopen evaluationrdquo approach by integrating the performance of systems other than those inthe health care sector

The paper relies on a longitudinal experience to thoroughly investigate itsdynamics by identifying the problematic issues it tackled and the solution it devisedComparisons with other cases were not made thus further studies could investigate there-framing process described in this paper by analyzing multiple experiences or casesfrom different contexts

Notes

1 See for instance the government acts of Basilicata Veneto and Tuscany available at wwwregionebasilicataitgiuntasitegiuntadepartmentjspdep=100061amparea=585290ampotype=1059ampid=2996190 httpsburregionevenetoitBurvServicespubblicaDettaglioDgraspxid=356632 wwwregionetoscanaitbancadatiattiContenutoxmlid=124931ampnomeFile=Delibera_n675_del_05-08-2013

2 wwwperformancesssupitnetval

References

Aidemark L (2001) ldquoThe meaning of balanced scorecard in the health care organizationrdquoFinancial Accountability amp Management Vol 17 No 1 pp 23-40

Anderson SW and Dekker HC (2015) ldquoThe role of management controls in transforming firmboundaries and sustaining hybrid organizational formsrdquo Foundations and Trends inAccounting Vol 8 No 2 pp 75-141

Anderson SW and Sedatole K (2003) ldquoManagement accounting for the extended enterpriseperformance management for strategic alliances and networked partnersrdquo in Bhimani A (Ed)Management Accounting in the Digital Economy Oxford University Press Oxford

Arah OA Westert GP Hurst J and Klazinga NS (2006) ldquoA conceptual framework for the OECDhealth care quality indicators projectrdquo International Journal for Quality in Health Care Vol 18No 1 pp 5-13 doi 101093intqhcmzl024

Arnaboldi M Lapsley I and Steccolini I (2015) ldquoPerformance management in the public sector the ultimate challengerdquo Financial Accountability and Management Vol 31 No 1 pp 1-22

Ballantine J Brignall S and Modell S (1998) ldquoPerformance measurement and management in publichealth services a comparison of UK and Swedish practicerdquo Management Accounting ResearchVol 9 No 1 pp 71-94

Berry AJ (1994) ldquoSpanning traditional boundaries organization and control of embeddedoperationsrdquo Leadership amp Organization Development Journal Vol 15 No 7 pp 4-10doi 10110801437739410066478

Bevan G and Hood C (2006) ldquoWhatrsquos measured is what matters targets and gaming in the Englishpublic health care systemrdquo Public Administration Vol 84 No 3 pp 517-538 doi 101111j1467-9299200600600x

Bevan G and Wilson D (2013) ldquoDoes lsquonaming and shamingrsquo work for schools and hospitalsLessons from natural experiments following devolution in England and Walesrdquo Public Money ampManagement Vol 33 No 4 pp 245-252 doi 101080095409622013799801

2268

MD5610

Bevan G Evans A and Nuti S (2018) ldquoReputations count why benchmarking performance isimproving health care across the worldrdquo Health Economics Policy and Law CambridgeUniversity Press pp 1-21 doi 101017S1744133117000561

Bianchi C (2010) ldquoImproving performance and fostering accountability in the public sector throughsystem dynamics modelling from an lsquoexternalrsquo to an lsquointernalrsquo perspectiverdquo Systems Researchand Behavioral Science Vol 27 pp 361-384 doi 101002sres

Bititci U Cocca P and Ates A (2016) ldquoImpact of visual performance management system on theperformance management practices of organizationsrdquo International Journal of ProductionResearch Vol 54 No 6 pp 1571-1593

Bititci U Garengo P Doumlrfler V and Nudurupati S (2012) ldquoPerformance measurement challengesfor tomorrowrdquo International Journal of Management Reviews Vol 14 No 3 pp 305-327doi 101111j1468-2370201100318x

Booth A (2006) ldquoCounting what counts performance measurement and evidence-basedpracticerdquo Performance Measurement and Metrics Vol 7 No 2 pp 63-74 doi 10110814678040610679452

Bouckaert G and Halligan J (2008) Managing Performance International Comparisons RoutledgeAbingdon Oxon

Bourne M (2001) The Handbook of Performance Measurement Gee Publishing AbingdonOxon London

Brignall S and Modell S (2000) ldquoAn institutional perspective on performance measurement andmanagement in the lsquonew public sectorrsquo rdquoManagement Accounting Research Vol 11 pp 281-306doi 101006mare20000136

Broadbent J and Laughlin R (2009) ldquoPerformance management systems a conceptual modelrdquoManagement Accounting Research Vol 20 No 4 pp 283-295 doi 101016jmar200907004

Bryson JM Crosby BC and Bloomberg L (2014) ldquoPublic value governance moving beyondtraditional public administration and the new public managementrdquo Public AdministrationReview Vol 74 No 4 pp 445-456 doi 101111puar12238Public

Christensen T and Laegreid P (2007) ldquoThe whole-of-government approach to public sector reformrdquoPublic Administration Review Vol 67 No 6 pp 1059-1066 doi 101111j1540-6210200700797x

Chua WF and Preston A (1994) ldquoWorrying about accounting in health carerdquoAccounting Auditing ampAccountability Journal Vol 7 No 3 pp 4-17

Cuganesan S Jacobs K and Lacey D (2014) ldquoBeyond new public management does performancemeasurement drive public value in networksrdquo in Guthrie J Marcon G Russo S andFarneti F (Eds) Public Value Management Measurement and Reporting (Studies in Public andNon-Profit Governance) Vol 3 pp 21-42

Davies HTO and Lampel J (1998) ldquoTrust in performance indicatorsrdquo Quality in Health Care Vol 7No 3 pp 159-162

Dekker HC (2016) ldquoOn the boundaries between intrafirm and interfirm managementaccounting researchrdquo Management Accounting Research Vol 31 pp 86-99 doi 101016jmar201601001

Donabedian A (1988) ldquoThe quality of care how can it be assessedrdquo The Journal of the AmericanMedical Association Vol 260 No 12 pp 1743-1748

Kasanen E Lukka K and Siitonen A (1993) ldquoThe constructive approach in management accountingresearchrdquo Journal of Management Accounting Research Vol 5 pp 243-264

Gray M and El Turabi A (2012) ldquoOptimising the value of interventions for populationsrdquoBritish Medical Journal Vol 345 pp 1-2 doi 101136bmje6192

Gray M Pitini E Kelley T and Bacon N (2017) ldquoManaging population healthcarerdquo Journal of theRoyal Society of Medicine Vol 110 No 11 pp 434-439

2269


systems

Halligan J Sarrico CS and Rhodes ML (2012) ldquoOn the road to performance governance in the publicdomainrdquo International Journal of Productivity and Performance Management Vol 61 No 3pp 224-234 doi 10110817410401211205623

Hayes RH and Abernathy WJ (1980) ldquoManaging our way to economic declinerdquo Harvard BusinessReview Vol 58 pp 67-77

Head BW and Alford J (2015) ldquoWicked problems implications for public policy and managementrdquoAdministration amp Society Vol 47 No 6 pp 711-739 doi 1011770095399713481601

Hibbard JH Stockard J and Tusler M (2003) ldquoDoes publicizing hospital performancestimulate quality improvement effortsrdquo Health Affairs Vol 22 No 2 pp 84-94 doi 101377hlthaff22284

Hood C (1991) ldquoA public management for all seasonsrdquo Public Administration Vol 69 No 1 pp 3-20

Kaplan RS and Norton DP (1992) ldquoThe balanced scorecard ndash measures that drive performancerdquoHarvard Business Review Vol 70 Nos 1 pp 71-79

Kaplan RS and Norton DP (1996) ldquoUsing the balanced scorecard as a strategic managementsystemrdquo Harvard Business Review Vol 85 Nos 7-8 pp 37-60

Kurunmaumlki L and Miller P (2011) ldquoRegulatory hybrids partnerships budgeting and modernisinggovernmentrdquo Management Accounting Research Vol 22 No 4 pp 220-241 doi 101016jmar201008004

Labro E and Tero-Seppo T (2003) ldquoOn bringing more action into management accounting researchprocess considerations based on two constructive case studiesrdquo European Accounting ReviewVol 12 No 3 pp 409-442

Lee VS Kawamoto K Hess R Park C Young J Hunter C Johnson S Gulbransen S Pelt CEHorton DJ and Graves KK (2017) ldquoImplementation of a value-driven outcomes program toidentify high variability in clinical costs and outcomes and association with reduced cost andimproved qualityrdquo Journal of the American Medical Association Vol 316 No 10 pp 1061-1072doi 101001jama201612226

Leggat SG Narine L Lemieux-Charles L Barnsley J Baker GR Sicotte C Champagne F andBilodeau H (1998) ldquoA review of organisational performance assessment in healthcarerdquoHealth Services Management Research Vol 11 pp 3-23 doi 101177095148489801100102

Lemieux-Charles L McGuire W Champagne F Barnsley J Cole D and Sicotte C (2003) ldquoThe useof multilevel performance indicators in managing performance in health care organizationsrdquoManagement Decision Vol 41 No 8 pp 760-770 doi 10110800251740310496279

Lengler R and Eppler M (2007) ldquoTowards a periodic table of visualization methods for managementrdquoLASTED International Conference on Graphics and Visualization in Engineering ClearwaterFlorida

Lomas J and Brown A (2009) ldquoResearch and advice giving a functional view of evidence-informedpolicy advice in a Canadian Ministry of Healthrdquo The Milbank Quarterly Vol 87 No 4 pp 903-926

Marr B (2006) Strategic Performance Management Leveraging and Measuring your Intangible ValueDrivers Butterworth-Heinemann Oxford

Melnyk SA Bititci U Platts K Tobias J and Andersen B (2013) ldquoIs performance measurementand management fit for the futurerdquo Management Accounting Research Vol 25 No 2pp 173-186 doi 101016jmar201307007

Meyer MW and Gupta V (1994) ldquoThe performance paradoxrdquo Research in Organizational BehaviorVol 16 pp 309-369

Moore MH (1995) Creating Public Value Strategic Management in Government Harvard UniversityPress Harvard MA

Moore MH (2013) Recognizing Public Value Harvard University Press Harvard MA

Naranjo-Gil D Saacutenchez-Expoacutesito MJ and Goacutemez-Ruiz L (2016) ldquoTraditional vs contemporarymanagement control practices for developing public health policiesrdquo International Journal ofEnvironmental Research and Public Health Vol 13 pp 713-726 doi 103390ijerph13070713

2270

MD5610

Noslashrreklit H Raffnsoslashe-Moslashller M and Mitchell F (2016) ldquoA pragmatic constructivist approach toaccounting practice and researchrdquo Qualitative Research in Accounting amp Management Vol 13No 3 pp 266-277

Noto G and Bianchi C (2015) ldquoDealing with multi-level governance and wicked problems inurban transportation systems the case of Palermo municipalityrdquo Systems Vol 3 No 3 pp 62-80doi 103390systems3030062

Nuti S and Vainieri M (2016) ldquoStrategies and tools to manage variation in regional governance systemsrdquoin Johnson A and Stuckel T (Eds) Medical Practice Variations Springer New York NYpp 433-457

Nuti S Seghieri C and Vainieri M (2013) ldquoAssessing the effectiveness of a performance evaluationsystem in the public health care sector Some novel evidence from the Tuscany regionexperiencerdquo Journal of Management and Governance Vol 17 No 1 pp 59-69 doi 101007s10997-012-9218-5

Nuti S Vainieri M and Vola F (2017) ldquoPriorities and targets supporting target-setting inhealthcarerdquo Public Money amp Management Vol 37 No 4 pp 277-284 doi 1010800954096220171295728

Nuti S De Rosis S Bonciani M and Murante AM (2017) ldquoRethinking healthcare performanceevaluation systems towards the people-centredness approach their pathways their experiencetheir evaluationrdquo Healthcare Papers Vol 17 No 2 pp 56-64

Nuti S Bini B Ruggieri TG Piaggesi A and Ricci L (2016) ldquoBridging the gap between theory andpractice in integrated care the case of the diabetic foot pathway in Tuscanyrdquo InternationalJournal of Integrated Care Vol 16 No 2 pp 1-14

Nuti S Vola F Bonini A and Vainieri M (2016) ldquoMaking governance work in the health care sectorevidence from a lsquonatural experimentrsquo in Italyrdquo Health Economics Policy and Law Vol 11 No 1pp 17-38 doi 101017S1744133115000067

OrsquoFlynn J (2007) ldquoFrom new public management to public value paradigmatic changeand managerial implicationsrdquo The Australian Journal of Public Administration Vol 66 No 3pp 353-366 doi 101111j1467-8500200700545x

Otley D (1999) ldquoPerformance management a framework for management control systems researchrdquoManagement Accounting Research Vol 10 No 4 pp 363-382 doi 101006mare19990115

Pettigrew A (1992) ldquoThe character and significance of strategy process researchrdquo StrategicManagement Journal Vol 13 pp 5-16


Porter ME (2010) ldquoWhat is value in health carerdquo The New England Journal of Medicine Vol 363No 26 pp 2477-2481

Radnor Z and Mcguire M (2004) ldquoPerformance management in the public sector fact or fictionrdquoInternational Journal of Productivity and Performance Management Vol 53 No 3 pp 245-260doi 10110817410400410523783

Ramagem C Urrutia S Griffith T Cruz M Fabrega R Holder R and Montenegro H (2011)ldquoCombating health care fragmentation through integrated health services delivery networksrdquoInternational Journal of Integrated Care Vol 11 August pp 1-2 doi 10-1-101537ijic2011-100

Richards T Montori VM Godlee F Lapsley P and Paul D (2013) ldquoLet the patient revolutionbeginrdquo British Medical Journal Vol 346 May p f2614 doi 101136bmjf2614

Ryan C and Walsh P (2004) ldquoCollaboration of public sector agencies reporting and accountabilitychallengesrdquo The International Journal of Public Sector Management Vol 17 Nos 6-7pp 621-631

Sackett DL Rosenberg WM Gray JM Haynes RB and Richardson WS (1996) ldquoEvidence basedmedicine what it is and what it isnrsquotrdquo British Medical Journal Vol 312 No 71 pp 71-72

2271


systems

Teece DJ (1990) ldquoContributions and impediments of economic analysis to the study of strategicmanagementrdquo Perspective on Strategic Management pp 39-80

Tversky A and Kahneman D (1981) ldquoThe framing of decisions and the psychology of choicerdquoScience Vol 221 No 4481 pp 453-458

van der Meer-Kooistra J and Scapens RW (2008) ldquoThe governance of lateral relations betweenand within organizationsrdquo Management Accounting Research Vol 19 No 4 pp 365-384doi 101016jmar200808001

Van Peursem KA Prat MJ and Lawrence SR (1995) ldquoHealth management performance a review ofmeasures and indicatorsrdquo Accounting Auditing amp Accountability Journal Vol 8 No 5 pp 34-70

Van Thiel S and Leeuw FL (2002) ldquoThe performance paradox in the public sectorrdquo PublicPerformance amp Management Review Vol 25 No 3 pp 267-281 doi 1023073381236

Wadmann S Johansen S Lind A Birk HO and Hoeyer K (2013) ldquoAnalytical perspectives onperformance-based management an outline of theoretical assumptions in the existing literaturerdquoHealth Economics Policy and Law Vol 8 No 4 pp 511-527 doi 101017S174413311300011X

Walshe K and Rundall TG (2001) ldquoEvidence-based management from theory to practice in healthcarerdquo The Milbank Quarterly Vol 79 No 3 pp 429-457

Wilcox M and Bourne M (2002) ldquoPerformance measurement and management research and actionrdquoPerformance Management Association Conference Boston MA

Corresponding authorSabina Nuti can be contacted at snutisantannapisait


2272

MD5610

Hospital unit understaffingand missed treatments

primary evidenceAshley Y Metcalf

College of Business Ohio University Athens Ohio USAYong Wang

West Chester University Philadelphia Pennsylvania USA andMarco Habermann

College of Business Ohio University Athens Ohio USA

AbstractPurpose ndash Hospitals throughout the USA are facing increasing patient demand and employee shortages Thiscapacity issue has led to understaffing in some hospital areas The purpose of this paper is to examine theunderstaffing in hospital-unit respiratory care and the impact to error rates specifically missed treatments ratesThe moderating effects of teamwork and standardized integrated information systems are also consideredDesignmethodologyapproach ndash Survey methodology is used for data collection of respiratory caremanagers within hospital units Regression is used to test the hypotheses in this studyFindings ndash The regression results show that higher rates of understaffing are associated with more missedtreatments In addition both teamwork and integrated information systems are associated with lower missedtreatments Finally the moderating effect of teamwork is also highly significant within the model whileintegrated information systems are not a significant moderatorPractical implications ndash Managers working within understaffed hospital units can try to reduce missedtreatment rates by both integrated information systems and teamwork among employees Additional benefitscan be gained from teamwork due to the indirect effects (moderating effects) as well This indicates teamworktraining can be useful for quality initiativesOriginalityvalue ndash Understaffing is associated with higher missed treatments in hospital unitsStandardized integrated information systems within a hospital are associated with less missed treatmentsFurthermore employee teamwork within a hospital unit is associated with a direct effect on missed treatmentrates as well as an indirect effect by weakening the negative impact of understaffingKeywords Information systems Teamwork Healthcare Hospital units Medical staffing UnderstaffingPaper type Research paper

IntroductionDemand for many healthcare frontline workers (nurses therapists etc) is expected toincrease at above-average rates between the years 2016-2026 due to the aging population inthe USA The demand for registered nurses is expected to increase 15 percent (BLS 2018a)Demand for respiratory therapists is expected to increase 23 percent (BLS 2018b) whereasthe demand for nursing assistants is expected to increase 11 percent (BLS 2018c)In addition to increasing demand existing staffing shortages and employee turnover inhospitals has become a critical area of concern for healthcare administration (Aiken et al2011 Jacobson 2015) In fact even if nurses are available in the labor market manyhospitals are still refusing to hire because of budget constraints ( Jacobson 2015)This means that nationwide nursing shortages combined with hospital budget constraintsare leading to a long-term capacity imbalance Managers within US hospitals have to dealwith chronic understaffing and subsequent impacts to patient care ( Jacobson 2015)

In healthcare practice understaffing of frontline workers influences core managementdecisions because its consequence is associated with higher error rates and poorer quality ofcare (Lang et al 2004 Twigg et al 2015) Managing the issue of understaffing also falls withinevidence-based healthcare management For example Walshe and Rundall (2001) proposed



0025-1747DOI 101108MD-09-2017-0908

Received 29 September 2017Revised 6 March 2018Accepted 18 April 2018


2273

Hospital unitunderstaffing


that an evidence-based healthcare system should be implemented to assess and prevent theoveruse underuse and misuse of healthcare resources Understaffing is certainly related to theunderuse or misuse of resources High demands on frontline employees and lack of staffing tomeet those demands can lead to increased error rates ( Jacobson 2015 Twigg et al 2015) andhigher rates of missed treatments Missed treatments are treatments that have been scheduledas part of a patientrsquos care plan but are missed by the frontline employee Previous research inevidence-based management considered missed treatments a type of medical error and anoutcome of poor management decisions (Arndt and Bigelow 2009)

In addition to staffing concerns teamwork among frontline employees is particularlyimportant in a hospital environment and is related to the application of clinical expertise inevidence-based healthcare management (Sackett et al 1996) Hospitals are very labor-intensiveservice environments that have to meet the demands of diverse patient needs Teamworkamong hospital caregivers enhances communication and coordination as well as increases thequality of care to patients (Institute of Medicine (IOM) 2000 2001 Pronovost and Vohr 2010)

Elements of communication and coordination can also be seen within the hospitalrsquosinformation systems infrastructure Hospital information systems have also been linked tonurse training and staffing decisions as well as quality of care (Li and Benton 2006) andmedical errors (Radley et al 2013) Integrated information systems have been shown toincrease quality of care decrease medical errors and decrease hospital costs (Angst et al2011 Li and Benton 2006 Radley et al 2013) Further in line with evidence-basedmanagement philosophy integrated information systems are essential in providing dataand analytics used for decision making (Guo et al 2017)

The current study seeks to examine the direct relationships between teamwork andmissed treatments information systems and missed treatments as well as understaffingand missed treatments In addition this study investigates if teamwork among frontlineemployees or the existence of integrated information systems can be effective at weakeningthe effects of understaffing on error rates The empirical relationships among these fourfactors are critical for crafting healthcare management strategies Thus the findings candirectly assist researchers and managers who are focused on evidence-based managementwhich is defined as the ldquosystematic application of the best available evidence to theevaluation of managerial strategies for improving the performance of health servicesorganizationsrdquo (Kovner and Rundall 2006 p 6) So the key research question is

RQ1 Does employee teamwork or integrated information systems moderate theunderstaffing-missed treatment relationship

Conceptual frameworkHospital staffing (and subsequent understaffing) has been of increasing interest in thehealthcare literature Understaffing of nurses is a major impediment to providing high-qualitycare (Aiken et al 2001 2002 Needleman et al 2002) In fact Twigg et al (2015) found that evenafter controlling for patient characteristics understaffing in nurses was associated with higherodds of infection pressure injury pneumonia deep vein thrombosis sepsis and gastrointestinalbleed In addition the reader should refer to two literature reviews Lang et al (2004) Kane et al(2007) regarding the impact of nurse staffing on patient outcomes The overall conclusions fromthese literature reviews were consistent adequate levels of staffing (or lower nurse-to-patientratios) are associated with higher quality of care measures

Furthermore hospital understaffing is associated with lower levels of job satisfactionand higher levels of staff burnout (Aiken et al 2002) This burnout and lack of jobsatisfaction will only make a staffing situation worse in a particular hospital Understaffingshould be actively managed in order to maintain current staff satisfaction reduce turnoverand improve hospital quality of care (Needleman et al 2002 Twigg et al 2015)

2274

MD5610

Understaffing is a particular problem in a labor-intensive environment like healthcare wheredemand for services is increasing Chronic understaffing of nurses is associated with higherhospital-acquired infection rates higher rates of pneumonia and higher rates of sepsis (Needlemanet al 2002 Twigg et al 2015) Rogowski et al (2013) confirmed this trend with NICU nurseswhere higher rates of understaffing significantly raised the infection rate for critical infants

In the current study missed treatments are considered a type of medical error Missedtreatments are treatments that are scheduled per the patientrsquos care plan but are missed bythe frontline employee When medical staff have to deal with too many tasks they usuallyfind themselves in high role stress at work (Peiro et al 2001) The role stress often includerole ambiguity (ie medical staffrsquos confusion about the expectations and requirements formanaging extra workload) role conflict (ie medical staffrsquos inability to meet supervisorrsquos andpatientrsquos simultaneous demands) and role overload (ie medical staffrsquos inability in servicingexisting patients together with new patients with limited time) (Schaubroeck et al 1989Chiu et al 2015) Such role stress makes it easier for medical errors to happenIf understaffing is present certainly there would be more opportunity for an employee tomiss a patientrsquos treatment because of their high workload Therefore it is expected thathigher levels of understaffing will be associated with higher rates of missed treatments

H1 Higher levels of understaffing are associated with higher missed treatment rates

Teamwork in a healthcare environment is particularly important due to the high laborintensity required for patient care The lack of teamwork and communication increases errorrates in hospitals (IOM 2000 2001 Pronovost and Vohr 2010) The Institute of Medicine(IOM 2001) indicated that teamwork can be a valuable mechanism to combat medical errorsFurthermore recent work in rapid-response teams has indicated these specialized teamshave resulted in lower levels of cardiac arrests and lower mortality rates (Berwick et al2006 Buist et al 2002 Chan et al 2010 DeVita et al 2004)

From a managerial perspective effective teamwork helps to achieve superior organizationalresults due to the synergistic effect (Hertel 2011 Sandoff and Nilsson 2016) To achieve betterresult in person-centered care the synergy from multiple personnel and units closely workingtogether is valuable for healthcare professionals (Rosengren 2016) Training hospitalemployees for better teamwork skills has been associated with better patient-safety culturebetter communications about errors and staff working together across hospital units ( Joneset al 2013) In addition nurses with greater teamwork have higher levels of job satisfactionlower burnout and higher perceived quality of care for their patients (Rafferty et al 2001)

In a recent statement by the American Heart Association many preventable hospitalerrors are due to breakdowns in communication collaboration and teamwork (Wahr et al2013) In hospital units where teamwork is present the element of communication andcollaboration may aid to ensure no patient treatments are missed Therefore in this studygreater levels of teamwork (and its associated communicationcollaboration efforts) areexpected to be associated with lower missed treatment rates

H2 Higher levels of teamwork within a hospital unit are associated with lower missedtreatment rates

The use of information systems and computerized physician order entry has beenassociated with lower rates of medical errors (Radley et al 2013) Information systems areadopted in healthcare settings to improve the delivery of services and documentation ofrecords (Angst et al 2011 Devaraj et al 2013 Meyer and Collier 2001) In addition Li andBenton (2006) show that information systems can lower the cost and increase the quality ofhealthcare in the nursing sector

The medical community has seen a greater emphasis on information systems when theAmerican Recovery and Reimbursement Act (Federal Register 2010) began enforcing

2275


penalties in 2015 to hospitals that have not implementing electronic health recordsAs of 2008 only 15 percent of hospitals had a comprehensive electronic records system andonly 76 percent had a basic electronic records system ( Jha et al 2009) Since that timechanges in legislation due to the affordable care act has tied full reimbursement (of Medicareand Medicaid) to a hospitalrsquos adoption of electronic records Though rates of adoption havesignificantly increased since then many hospitals nationwide still do not have an integratedinformation system An integrated information system is standardized and integratedacross departments to facilitate information flow across a hospital

The adoption of electronic records and physician order entry can reduce waiting timesreduce reporting times increase medication accuracy and reduce transcription errors(Kaushal et al 2003 Mekhjian et al 2002 Radley et al 2013) Integrated information systemscan result in process simplification and therefore higher levels of patient safety (Bates et al2001) In addition integrated information systems can increase patient flow through thehospital and therefore reduce length of stay measures (Devaraj et al 2013) An integratedinformation system can increase the absorptive capacity of the hospital unit providing theunit a stronger ability in identifying transforming synthesizing analyzing and reportinginformation and knowledge about patients (Zahra and George 2002 Todorova and Durisin2007) At the individual staff level an integrated information system can help to create a goodldquocognitive fitrdquo when appropriate information is needed for various tasks (Vessey 1991 Dillaand Steinbart 2005) A medical professional becomes better at problem solving if sufficientinformation is readily available and timely presented during task completion Thus the bettera hospital unit becomes in information absorption and integration the better the outcomepatients receive In this study it is expected that hospital units that have standardizedintegrated information systems will have lower rates of missed treatments

H3 Greater availability of standardized integrated information systems is associatedwith lower missed treatment rates

Coordination and information exchange are critical to achieving better patient outcomes(Boyer and Pronovost 2010 Gittell et al 2000 Pronovost and Vohr 2010) Betterinformation exchange enhances healthcare delivery and reduces medical errors (White et al2004) With this in mind it is expected that coordination and collaboration among caregiverswill create a working environment that will lessen the effects of understaffing on medicalerrors Coordination collaboration and information exchange can occur in the form ofrelationships (via employee teamwork) or in the form of technology (via integratedinformation systems) In hospital units with high levels of understaffing the existence ofteamwork and integrated information systems can lessen the impact of the staffingproblems on missed treatment rates Therefore this study predicts negative moderating(ie dampening) effects by teamwork and integrated information systems Figure 1 providesa conceptual model for this study

H4a Higher levels of teamwork negatively moderate the relationship betweenunderstaffing and missed treatments

H4b Greater availability of standardized integrated information systems negativelymoderates the relationship between understaffing and missed treatments

MethodInstrument development and research settingThis study was carried out within the field of respiratory care using a nationwide set of USnon-governmental hospitals (ie VA hospitals were not included as they have differentmanagerial structures and incentives) Respiratory care is a specialized industry in the USA

2276

MD5610

where respiratory therapists treat patients with breathing difficulties and lung diseasesRespiratory therapists are frontline caregivers that commonly provide treatments forconditions such as Asthma COPD and Pneumonia (BLS 2018b)

The level of analysis in the study is the hospital unit Each hospital in the study hasmultiple hospital units These hospital units have different levels of staffing teamwork andmissed treatments as the respiratory care needs can vary depending on the hospital unitTherefore this study considers four potential units within a hospital EmergencyDepartment (ED) Intensive Care Unit (ICU) Neonatal Intensive Care Unit (NICU) and AdultInpatient Floors (AI) In each of these units respiratory therapists are required to provide avariety of care regarding respiratory services In fact respiratory therapists are the primaryfrontline employees that deliver respiratory care treatments (relative to nurses who deliver abroad range of treatments) Therefore the survey was designed to be completed by therespiratory care managersupervisor for that particular hospital unit

Survey questions for understaffing and teamwork were developed by working closely withour industry managers Both understaffing and teamwork are well understood variables thatrespiratory care managers are consistently aware Understaffing is the degree to which ahospital unit is understaffed in respiratory therapists Teamwork is the degree to which thefrontline employees (nurses therapists etc) work together to solve problems for patient care

Information systems describe the availability standardization and use of informationsystems within the hospital The information systems scale was drawn from priorhealthcare literature (Goldstein and Naor 2005 Meyer and Collier 2001)

In order to measure error rates as an outcome measure a variable was needed that wasconsistently monitored at the hospital unit-level between hospitals settings Most variablesare aggregated up to the hospital-level for government reporting Other objective data areavailable at the patient-level (but not necessarily defined by hospital unit) In additionpatient-level data requires significant IRB approvals from each participating hospitalbecause of privacy rights So our industry partners and the American Association forRespiratory Care (AARC) were contacted to determine if there were any measuresconsistently tracked by respiratory care managers at the unit-level within a hospital

One variable emerged that is known by respiratory care managers across the country(and measured in a consistent way) missed treatment rates In fact the AARC maintains aproprietary benchmarking database that tracks missed treatments rates for its participatinghospitals The AARC variable for missed treatments is defined as the percentage ofldquotreatments ordered but not delivered within a given time periodrdquo (AARC 2017) While thefull missed treatment database was not available from the AARC the associationprovided us with blind (no hospital identifiers) annual numbers for missed treatments

Understaffing

Teamwork

MissedTreatments

ControlsFor ProfitTeaching StatusHospital SizeUrbanRuralHospital Unit

H1H4a

H2

InformationSystems

H4b

H3

Figure 1Conceptual model

2277


Quartile calculations from the AARC database for missed treatments were used to developthe scale cutoffs which were then used as the survey response options for the missedtreatments variable in this study By using this scale for missed treatments respiratory caremanagers were much more comfortable providing a response than if we asked for an open-ended number on missed treatment rates Also keep in mind that the missed treatmentsdiscussed in this study are treatments missed in the respiratory care plan for the patientMissed treatments in other areas of the patientrsquos care plan (outside of respiratory therapy)were not considered in this study Details of all survey items are presented in the Appendix

Prior to data collection the University of South Carolinarsquos Institutional Review Board(IRB) approved the survey and its distribution as ldquoIRB-exemptrdquo from written consentHospital control variables such as hospital size (measured as number of beds) profit vsnon-profit teaching vs non-teaching and urban vs rural were obtained from the AmericanHospital Association database for our participating hospitals

Data collectionThe data collection for this study involved several stages a pre-test revision and the maindata collection The proposed research questions in this study are dependent on thepractical relevance of the survey questions and the full understanding of the survey itemsby responding practitioners Several rounds of instrument pre-testing with hospital partnersin South Carolina (respiratory care managerstherapistspulmonologists) were used toensure the content validity of the survey constructs and question wording Content validitydefined as the ldquoadequacyrdquo in which the content in question has been sampled (Nunnally1978) is commonly assessed through the evaluation of the survey items by content expertsAs such four academics (professors in operations management) and six practitioners(two respiratory care managers two pulmonology physicians and four respiratorytherapists) reviewed each of the items included in the survey If survey items were confusingor unclear the item was revised and then reviewed again by these experts

For our main data collection the survey was distributed online using the Qualtricssoftware in Spring 2013 Respiratory care managerssupervisors were asked to respond tothe survey for the specific unit in the hospital (ED ICU NICU or AI) that they managedIn total usable responses were received from 105 respiratory care managerssupervisors(ie hospital units) from 45 different hospitals A summary of hospital units used in thisstudy is presented in Table I

Data analysisPrior to performing analysis tests for reliability and validity were performed on theinformation systems scale Since the scale was drawn from existing literature (Goldstein andNaor 2005 Meyer and Collier 2001) confirmatory factor analysis was performed Allgoodness of fit values (CFIfrac14 097 SRMRfrac14 004 CDfrac14 096) were well within acceptablecutoff limits (Hu and Bentler 1999) All factor loadings were all above 06 indicating properconvergent validity Cronbachrsquos α was 089 indicating the scale has high levels of internal

Hospital unit summary

Total number of hospitals 45Number of units ICU 38Number of units NICU 12Number of units ED 24Number of units adult inpatient 31Number of states represented 21

Table ISummary ofparticipating hospitalunits

2278

MD5610

consistency (Nunnally 1978) The indicators for the information systems scale wereaveraged to determine a single score for information systems that was used in theregression analysis Furthermore descriptive statistics of all variables in our model arepresented in Table II

Multiple regression was used in the STATA 13 software to test the hypotheses Model 1tests only the direct effects Model 2 adds the moderating effects to the model In addition tothe main variables each model also contains control variables for profit-status (For Profit)teaching status (Teaching) Size Urban vs Rural and hospital unit (ICU NICU ED) Thedummy variables for hospital unit (ICU NICU ED) are interpreted relative to the ldquogeneraladult inpatientrdquo units Post-regression tests for heteroscedasticity and multicollinearity wereconducted and did not show any problems with the regression models

ResultsTable III presents the regression results Model 1 examines only the direct effectsH1 H2 andH3 are all supported ( po001) Higher levels of understaffing are associated with significantlyhigher levels of missed treatments Greater levels of teamwork within the hospital unit are

Variable Mean SD Min Max

Focal variablesUnderstaffing 351 117 1 5Information systems 392 086 1 5Teamwork 41 072 2 5Missed treatments 199 119 1 5

Control variablesFor profit 005 021 0 1Teaching status 065 048 0 1Size 411 299 25 1637Urban 086 035 0 1ICU 035 048 0 1NICU 011 032 0 1ED 022 042 0 1

Table IIDescriptive statistics

of variables

DV missed treatments Model 1 Model 2

Understaffing 0270 (0005) 0238 (0010)Information systems minus0318 (0009) minus0283 (0026)Teamwork minus0507 (0002) minus0574 (0000)Understaffingtimes IS ndash minus0026 (0783)UnderstaffingtimesTW ndash minus0284 (0019)For profit minus1059 (0041) minus1338 (0011)Teaching minus0237 (0348) minus0117 (0645)Size 0000 (0538) 0000 (0644)Urban minus0061 (0850) minus0004 (0989)ICU minus0193 (0443) minus0192 (0439)NICU minus0932 (0017) minus1106 (0005)ED minus0688 (0019) minus0648 (0025)n 105 105R2 034 038Notes Values in parentheses are p-values po005 po001

Table IIIRegression results

2279


associated with lower levels of missed treatments Greater availability of standardizedintegrated information systems is associated with lower levels of missed treatments

Model 2 examines the moderating effects of teamwork and information systems H4a isalso supported ( po005) Greater levels of teamwork within a hospital unit dampens therelationship between understaffing and missed treatments However H4b is not supportedThe level of use of information systems did not impact the understaffing to missedtreatments relationship

Control variables for ldquoFor Profitrdquo ldquoNICUrdquo and ldquoEDrdquo are significant in both modelsSo for-profit hospitals have lower levels of missed treatments relative to non-profit facilitiesFurthermore neonatal ICUs and EDs have lower missed treatment rates relative to generaladult inpatient units Finally teaching status size of hospital urban environments andICUs were not significant predictors of missed treatment rates

DiscussionHealthcare researchers and practitioners with an evidence-based management philosophyconstantly seek causal links for rational decision-making (Arndt and Bigelow 2009)The findings of this study provide empirical evidence for improving the performance ofhealthcare organizations which is a core task of evidence-based management in healthcare(Kovner and Rundall 2006 Guo et al 2017) The results show that higher levels ofunderstaffing are associated with higher missed treatment rates It is no surprise that in anenvironment where frontline employees have a high workload (due to inadequate staffing) amistake is more likely to occur This is consistent with prior literature stating the lack ofadequate staffing increases error rates in hospitals (Aiken et al 2001 2002 Jacobson 2015Lang et al 2004 Twigg et al 2015)

Previous research in evidence-based healthcare management rarely takes intoconsideration organizational problems (eg understaffing) together with solutions(eg teamwork and information systems) in one research model Our research fills thevoid In this study we examine not only the effect of understaffing but also the effects ofteamwork and integrated information systems side by side The results of the direct effectsalso show that higher levels of teamwork and availability of integrated information systemsboth significantly decrease the missed treatment rates This could be due to the increasedlevels of communication collaboration and subsequent information sharing for patient careThe results highlight the importance of information sharing via teamwork and informationsystems in achieving lower missed treatment rates in healthcare Teamwork by frontlineemployees appears to be a top determinant to solving the missed treatment problemThe result is in line with previous findings that team collaboration is the key to achievingsuperior outcomes in healthcare management (Sackett et al 1996) In addition a hospitalrsquosintegrated information systems infrastructure also helps mitigating missed treatments dueto less time spent in communication and better coordination among employees Our resultsupports the notion that shared data and analytics are essential in healthcare decisionmaking (Guo et al 2017) and justifies hospitalsrsquo continuous investment in maintaining andupdating their information systems

Furthermore a noteworthy finding in this study is the significant moderating effect ofteamwork Higher levels of teamwork can be used to weaken the negative effect ofunderstaffing on missed treatments So not only does teamwork decrease missedtreatments directly but it also weakens the adverse impact of understaffing on missedtreatments This provides a useful solution to the understaffing issue encountered byhospital unit managers trying to maintain high quality of care While the personalinteractions via teamwork are shown to weaken the negative effect of understaffingthe availability of integrated information systems has statistically insignificant effect onthe understaffing-missed treatments relationship in this study In theory and in practice

2280

MD5610

the use of information systems is directly associated with better outcomes in healthcareas indicated by the direct relationship between information systems and missedtreatments However the insignificant moderating effect provides initial evidence thatwhen understaffed hospital units still suffers from missed treatments even thoughadvanced information systems are available It is possible that in the circumstances ofhigh levels of understaffing the integrated information systems cannot be effectivelyutilized for communication and coordination by a much smaller number of frontlineemployees who remain at work The insignificant moderating effect of integratedinformation systems provides evident caution for healthcare managers who reply oninformation systems and add important contribution to previous information systemsresearch in healthcare

The strategic use of integrated information systems can certainly improve the qualityof healthcare by speeding up patient flow as well as the delivery of services(Mekhjian et al 2002 Devaraj et al 2013 Radley et al 2013) However whenunderstaffed the process in which patient information and knowledge move along orcirculate may be slowed down or even disabled The various touchpoints in informationdiscovery entry transfer and reporting need to be actively managed by the differentemployees in patient services When one touchpoint in the chain is missing information itcan affect all that follows For example if the electronic diagnosis record is not created inthe beginning of patient service due to the shortage of medical staff subsequentdiagnosis treatment and reporting can be more difficult resulting in extended waitingtimes and increased transcription error rates The finding extends cognitive fit theory(Vessey 1991) into the organizational level suggesting that external task (eg obtainingpatient information and knowledge) and internal structure (eg availability of staff ) mustfit each other in order to achieve superior information systems performance in anorganization (eg hospital unit)

Control variables of profit status ED and NICU are significant in both models From amanagerial perspective the significant control variables help describe the variance of missedtreatments due to a hospital unitrsquos risk taking levels For-profit hospitals have lowermissed treatment rates relative to non-profit hospitals One potential explanation for the resultcould be hospital unit managersrsquo risk aversion due to profit orientation Medical errorscan be expensive therefore for-profit hospital units need to be more active in loweringmissed treatment rates Furthermore the results show that EDs and NICUs have lower missedtreatment rates compared to the ldquogeneral adult inpatientrdquo units One potential explanation isdue to the critical nature of medical risk in these departments Anymissed treatment in EDs orNICUs can potentially cause irreversible medical accident Thus these types of hospital unitsneed to pay high attention in fulfilling treatment plans among their critical patients to mitigatehigh treatment risk In view of evidence-based management in healthcare the results based onthe control variables can also offer managerial insights into managing missed treatments forrisk reduction

Practical implicationsBased on primary empirical evidence our findings shed light on how to reduce missedtreatments in healthcare Healthcare managers should be aware of the critical negativeconsequences of understaffing In view of evidence-based healthcare managementguidelines (Walshe and Rundall 2001) understaffing can be understood as an issuerelated to underuse or misuse of healthcare resources The former includes the shortage ofemployees in healthcare and the latter includes overwork or misplacement of medicalstaff Both underuse and misuse of medical staff appear to be more than a temporaryhuman resource problem in healthcare and can unavoidably cause medical errors such asmissed treatments

2281


To reduce medical errors related to missed treatments we suggest managers to resort toenhanced teamwork practice and better use of integrated information systems On one handteamwork practice can enhance the coordination among frontline medical staff To do somanagers should offer training to employees in order for them to better understand thenature and scope of teamwork Managers should also adopt teamwork performanceevaluation systems and teamwork incentive programs in human resource management Onthe other hand it is necessary to improve and update a hospitalrsquos information systemsinfrastructure on a regular basis

More importantly when understaffing takes place in a hospital unit managers shouldunderstand that teamwork can be helpful in avoiding foreseeable medical errors In suchcircumstances managers should deploy management practices such as cross-functionalteam leaders (Sarin and McDermott 2003) and heavyweight managers for internalcoordination (Koufteros et al 2010) for better teamwork outcomes In the meantimemanagers should not assume that having integrated information systems would necessarilymitigate the negative outcomes when essential medical personnel are absent As explainedpatient information cannot move along efficiently when certain employees in patientservices are not in place Fully counting on information systems to solve problems duringunderstaffed time periods will potentially cause medical catastrophe

To summarize the major practical implications of this study for healthcare managersare the following first unsurprisingly understaffing can be a source of errors and qualityproblems within hospital units Second both integrated information systems andteamwork among frontline employees are associated with less missed treatmentsThese effects are likely due to increased information sharing and collaborationFurthermore teamwork among frontline employees within a unit can also weaken theperilous effects caused by understaffing but the availability of information systemscannot The implications are that employee dynamics (such as teamwork) play animportant role not only in direct impact to outcomes but also through indirect channels bydampening problems cause by understaffing

To be clear by highlighting the positive roles of teamwork and integrated informationsystems we are not intending to promote deliberate understaffing of hospital units Howeverif a manager is stuck with a chronically understaffed environment efforts for teamwork andcollaboration can help reduce missed treatments and potentially maintain a higher quality ofpatient care Managers who practice evidence-based management should considerorganization-wide efforts to deliberately increase the level of teamwork among frontlineemployees within hospital units These efforts could include initiatives such as daily teambuilding or work-design to facilitate teamwork and collaboration among employees The sametype of efforts with integrated information systems can provide direct benefits to missedtreatments but cannot capture the indirect benefits of dampening the understaffing problemsEmployee multi-tasking with the different information system functions may potentially buildan underlying link between understaffing and missed treatments

Limitations and future researchWhile this study provides interesting implications for teamwork in an understaffedenvironment it is not without limitations This study asks respiratory care managers aboutunderstaffing but does not use actual staffing ratios Future studies can use objective dataon staffing numbers across hospitalshospital units and associated error rates to confirm thefindings in this paper In addition future work should consider individual factors (ie staffpersonality traits burnout individual workloads etc) that influence individual missedtreatment rates This study only considered one type of medical error missed treatments inrespiratory care Future work should consider other types of errors and even extend thesetting beyond respiratory care to include other specialties or nursing care

2282

MD5610

In addition future work can also consider organizational-level initiatives cultures andor technologies that drive medical errors This study has a sample where over 50 percent ofparticipating hospital units were from teaching hospitals While this study did not find anysignificant effects based on teaching status future work can examine how teaching statusor other organizational-level traits impact medical errors and quality of care It is possiblethat teaching status would impact hospital culture or level of training of frontlinecaregivers which then could impact the quality of care

Understaffing teamwork and information system capacity are all organizationalcontingencies To a broader extent it is expected that these contingency variables affectpatient missed treatment rates situationally depending on the fit with organizationalcharacteristics or hospital unit resource attributes Thus future research may furtherexamine the ldquofitrdquo between the independent variables used in the current study andorganizational characteristics to see how medical errors can be mitigated Future work couldalso consider data collection of error rates beforeafter information system implementationand beforeafter staff teamwork training to determine causal relationships

References

AARC (2017) ldquoMissed treatments about these metricsrdquo AARC benchmarking system available athttpcaarcorgresourcesbenchmarkingindexcfm (accessed September 23 2017)

Aiken LH Clarke SP Sloane DM Sochalski J and Silber JH (2002) ldquoHospital nurse staffing andpatient mortality nurse burnout and job dissatisfactionrdquo JAMA Vol 288 No 16 pp 1987-1993

Aiken LH Cimiotti J Sloane DM Smith HL Flynn L and Neff D (2011) ldquoThe effects of nursestaffing and nurse education on patient deaths in hospitals with different nurse workenvironmentsrdquo Medical Care Vol 49 No 12 pp 1047-1053

Aiken LH Clarke SP Sloane DM Sochalski JA Busse R Clarke H Giovannetti P Hunt JRafferty AM and Shamian J (2001) ldquoNursesrsquo reports on hospital care in five countriesrdquo HealthAffairs Vol 20 No 3 pp 43-53

Angst CM Devaraj S Queenan CC and Greenwood B (2011) ldquoPerformance effects related to thesequence of integration of healthcare technologiesrdquo Production and Operations ManagementVol 20 No 3 pp 319-333


Bates DW Cohen M Leape LL Overhage JM Shabot MM and Sheridan T (2001) ldquoReducingthe frequency of errors in medicine using information technologyrdquo Journal of the AmericanMedical Informatics Association Vol 8 No 4 pp 299-308

Berwick DM Calkins DR McCannon CJ and Hackbarth AD (2006) ldquoThe 100000 lives campaignsetting a goal and a deadline for improving health care qualityrdquo JAMA Vol 295 No 3pp 324-327

BLS (2018a) Occupational Outlook Handbook Registered Nurses Bureau of Labor Statistics USDepartment of Labor available at wwwblsgovoohhealthcareregistered-nurseshtm (accessedMay 4 2018)

BLS (2018b) Occupational Outlook Handbook Respiratory Therapists Bureau of Labor Statistics USDepartment of Labor available at wwwblsgovoohhealthcarerespiratory-therapistshtm(accessed May 4 2018)

BLS (2018c) Occupational Outlook Handbook Nursing Assistants and Orderlies Bureau of LaborStatistics US Department of Labor available at wwwblsgovoohhealthcarenursing-assistantshtm (accessed May 4 2018)

Boyer KK and Pronovost P (2010) ldquoWhat medicine can teach operations what operations can teachmedicinerdquo Journal of Operations Management Vol 28 No 5 pp 367-371

2283


Buist MD Moore GE Bernard SA Waxman BP Anderson JN and Nguyen TV (2002) ldquoEffectsof a medical emergency team on reduction of incidence of and mortality from unexpectedcardiac arrests in hospital preliminary studyrdquo BMJ Vol 324 No 7334 pp 387-390

Chan PS Jain R Nallmothu BK Berg RA and Sasson C (2010) ldquoRapid response teamsa systematic review and meta-analysisrdquo Arch Internal Medicine Vol 170 No 1 pp 18-26

Chiu S Yeh S-P and Huang TC (2015) ldquoRole stressors and employee deviance the moderatingeffect of social supportrdquo Personnel Review Vol 44 No 2 pp 308-324

Devaraj S Ow TT and Kohli R (2013) ldquoExamining the impact of information technology andpatient flow on healthcare performance a theory of swift and even flow (TSEF) perspectiverdquoJournal of Operations Management Vol 31 No 4 pp 181-192

DeVita MA Braithwaite RS Mahidhara R Stuart S Foraida M and Simmons RL (2004) ldquoUse ofmedical emergency team responses to reduce hospital cardiopulmonary arrestsrdquo BMJ Qualityand Safety Vol 13 No 4 pp 251-254

Dilla WN and Steinbart PJ (2005) ldquoUsing information display characteristics to provide decisionguidance in a choice task under conditions of strict uncertaintyrdquo Journal of Information SystemsVol 19 No 2 pp 29-55

Federal Register (2010) ldquoDepartment of health and human servicesrdquo Rules and Regulations Vol 75No 144 42 CFR Parts 412 413 422 and 495 available at wwwgpogovfdsyspkgFR-2010-07-28pdf2010-17207pdf

Gittell JH Fairfield KM Bierbaum B Head W Jackson R Kelly M Laskin R Lipson S Siliski JThornhill T and Zuckerman J (2000) ldquoImpact of relational coordination on quality of carepostoperative pain and functioning and length of stay a nine-hospital study of surgical patientsrdquoMedical Care Vol 38 No 8 pp 807-819

Goldstein SM and Naor M (2005) ldquoLinking publicness to operations management practices a studyof quality management practices in hospitalsrdquo Journal of Operations Management Vol 23 No 2pp 209-228

Guo R Berkshire SD Fulton LV and Hermanson PM (2017) ldquoUse of evidence-based management inhealthcare administration decision-makingrdquo Leadership in Health Services Vol 30 No 3 pp 330-342

Hertel G (2011) ldquoSynergetic effects in working teamsrdquo Journal of Managerial Psychology Vol 26 No 3pp 176-184

Hu L and Bentler PM (1999) ldquoCutoff criteria for fit indexes in covariance structure analysisconventional criteria versus new alternativesrdquo Structural Equation Model Vol 6 No 1 pp 1-55

Institute of Medicine (IOM) (2000) To err is Human Building a Safer Health System National AcademyPress Washington DC

Institute of Medicine (IOM) (2001) Crossing the Quality Chasm A New Health System for the 21stCentury National Academy Press Washington DC

Jacobson R (2015) ldquoWidespread understaffing of nurses increases risk to patientsrdquo Scientific Americana division of Nature America Inc July 14 available at wwwscientificamericancomarticlewidespread-understaffing-of-nurses-increases-risk-to-patients (accessed September 23 2017)

Jha AK DesRoches CM Campbell EG Donelan K Rao SR Ferris TG Shields A Rosenbaum Sand Blumenthal D (2009) ldquoUse of electronic health records in US hospitalsrdquo The New EnglandJournal of Medicine Vol 360 No 16 pp 1-11

Jones F Podila P and Powers C (2013) ldquoCreating a culture of safety in the emergency department thevalue of teamwork trainingrdquo Journal of Nursing Administration Vol 43 No 4 pp 194-200

Kane RL Shamliyan TA Mueller C Duval S and Wilt TJ (2007) ldquoThe association of registerednurse staffing levels and patient outcomes systematic review and meta-analysisrdquoMedical CareVol 45 No 12 pp 1195-1204

Kaushal R Shojania KG and Bates DW (2003) ldquoEffects of computerized physician order entry andclinical decision support systems on medication safety a systematic reviewrdquo Archives ofInternal Medicine Vol 163 No 12 pp 1409-1416

2284

MD5610

Koufteros XA Rawski GE and Rupak R (2010) ldquoOrganizational integration for productdevelopment the effects on glitches on‐time execution of engineering change orders andmarket successrdquo Decision Sciences Vol 41 No 1 pp 49-80


Lang TA Roman PS Hodge M Kravitz RL and Olson V (2004) ldquoNurse-patient ratios asystematic review on the effects of nurse staffing on patient nurse employee and hospitaloutcomesrdquo JONA Vol 34 Nos 78 pp 326-337

Li L and Benton WC (2006) ldquoHospital technology and nurse staffing management decisionsrdquoJournal of Operations Management Vol 24 No 5 pp 676-691

Mekhjian HS Kuman RR Kuehn L Bentley TD Teater P Thomas A Payne B and Ahmad A(2002) ldquoImmediate benefits realized following implementation of physician order entry at anacademic medical centerrdquo Journal of the American Medical Informatics Association Vol 9 No 5pp 529-539

Meyer SM and Collier DA (2001) ldquoAn empirical test of the causal relationships in the Baldrige healthcare pilot criteriardquo Journal of Operations Management Vol 19 No 4 pp 403-425

Needleman J Buerhaus P Mattke S Stewart M and Zelevinsky K (2002) ldquoNurse-staffing levelsand the quality of care in hospitalsrdquo New England Journal of Medicine Vol 346 No 22pp 1715-1722

Nunnally JC (1978) Psychometric Theory 2nd ed McGraw-Hill New York NY

Peiro JM Gonzalez-Roma V Tordera N and Manas MA (2001) ldquoDoes role stress predict burnoutover time among health care professionalsrdquo Psychol amp Health Vol 16 No 5 pp 511-525

Pronovost P and Vohr E (2010) Safe Patients Smart Hospitals How One Doctorrsquos Checklist Can HelpUs Change Health Care from the Inside Out Hudson Street Press New York NY

Radley DC Wasserman MR Olsho LE Shoemaker SJ Spranca MD and Bradshaw B (2013)ldquoReduction in medication errors in hospitals due to adoption of computerized provider orderentry systemsrdquo Journal of the American Medical Informatics Association Vol 20 No 3pp 470-476

Rafferty AM Ball J and Aiken LH (2001) ldquoAre teamwork and professional autonomy compatibleand do they result in improved hospital carerdquo BMJ Quality and Safety Vol 10 No S2pp ii32-ii37

Rogowski JA Staiger D Patrick T Horbar J Kenny M and Lake ET (2013) ldquoNurse staffing andNICU infection ratesrdquo JAMA Pediatrics Vol 167 No 5 pp 444-450

Rosengren K (2016) ldquoPerson-centered care a qualitative study on first line managersrsquo experiences onits implementationrdquo Health Services Management Research Vol 29 No 3 pp 42-49

Sackett DL Rosenberg WM Gray JM Haynes RB and Richardson WS (1996) ldquoEvidence basedmedicine what it is and what it isnrsquotrdquo British Medical Journal Vol 312 pp 71-72

Sandoff M and Nilsson K (2016) ldquoHow staff experience teamwork challenges in a new organizationalstructurerdquo Team Performance Management Vol 22 Nos 78 pp 415-427

Sarin S and McDermott C (2003) ldquoThe effect of team leader characteristics on learning knowledgeapplication and performance of cross‐functional new product development teamsrdquoDecision Sciences Vol 34 No 4 pp 707-739

Schaubroeck J Cotton JL and Jennings KR (1989) ldquoAntecedents and consequences ofrole stress a covariance structure analysisrdquo Journal of Organizational Behavior Vol 2 No 1pp 35-58

Todorova G and Durisin B (2007) ldquoAbsorptive capacity valuing a reconceptualizationrdquoAcademy of Management Review Vol 32 No 3 pp 774-786

Twigg DE Gelder L and Myers H (2015) ldquoThe impact of understaffed shifts on nurse-sensitiveoutcomesrdquo Journal of Advanced Nursing Vol 71 No 7 pp 1564-1572

2285


Vessey I (1991) ldquoCognitive fit a theory-based analysis of the graphs versus tables literaturerdquoDecision Sciences Vol 22 No 2 pp 219-240

Wahr JA Prager RL Abernathy J3 Martinez EA Salas E Seifert PC Groom RC Spiess BDSearles BE Sundt TM and Sanchez JA (2013) ldquoPatient safety in the cardiac operating roomhuman factors and teamwork a scientific statement from the American heart associationrdquoCirculation doi 101161CIR0b013e3182a38efa

Walshe K and Rundall TG (2001) ldquoEvidence‐based management from theory to practice in healthcarerdquo Milbank Quarterly Vol 79 No 3 pp 429-457

White AA Wright SW Blanco R Lemonds B Sisco J Bledsoe S Irwin C Isenhour J andPichert JW (2004) ldquoCause-and-effect analysis of risk management files to assess patient care inthe emergency departmentrdquo Academic Emergency Medicine Vol 11 No 10 pp 1035-1041

Zahra SA and George G (2002) ldquoAbsorptive capacity a review reconceptualization and extensionrdquoAcademy of Management Review Vol 27 No 2 pp 185-203

Appendix Survey items

Hospital unitResponses (1-ICU 2-NICU 3-ED 4-General Adult Inpatient)

(1) Which best describes your hospital unit

UnderstaffingResponses (1-Strongly Disagree 5-Strongly Agree)

(1) This unit is often understaffed in respiratory therapists

TeamworkResponses (1-Strongly Disagree 5-Strongly Agree)

(1) The members of this unit work together as a team for patient care

Information systemsResponses (1-Strongly Disagree 5-Strongly Agree)

(1) Our electronic information systems are standardized across departments

(2) Our electronic information systems are integrated across departments

(3) Our electronic information systems support frontline employees

(4) Both hardware and software are reliable

(5) Electronic information systems are used to link care givers actions with patient outcomes

Missed treatmentsResponses [0-023 024-065 066-185 186-5 above 5]

(1) What is your average missed treatments (in this unit)

Corresponding authorAshley Y Metcalf can be contacted at metcalfaohioedu


2286

MD5610

Emerald is excited to announce a recent partnership with Peerwith a platform that provides authors with a variety of services

The Emerald Peerwith site can be found herehttpsauthorservicesemeraldpublishingcom

Peerwith connects academics seeking support for their work with a relevant expert to get their research submission-ready Peerwith experts can help with the following language editing copy editing scientific editing translation services statistical support funding application support visuals video publication support literature search peer review and indexing services Authors post their assignments on the Peerwith site experts provide a quote and the fee and conditions are then agreed upon directly between the author and the expert

While we are not of course guaranteeing publication upon use of Peerwith we hope that being able to direct academics to this resource either before submission or during the peer review process will help authors further improve the quality of their papers and increase their chances of positive reviews and acceptance

Academics with relevant expertise can sign up as an expert on the Peerwith system herehttpswwwpeerwithcomservicesoffer

12750_PeerwithJournalAd_Quarto_v01indd 1 01032017 110718

insight

Backfiles CollectionsPreserving over 100 years of management research online

A lifetime investment for your institution Emerald Backfi les will signifi cantly enhance your libraryrsquos o ering by providing access to over 125000 articles from more than 260 journals dating back to 1898

Visit emeraldinsightcom

Get Backfi les Collections for your libraryRecommend Backfi les to your librarian today Find out more emeraldpublishingcombackfi lescollections















Man

age

me

nt D

ec

ision

Vo

lum

e 5

6 N

um

be

r 10 2

018

em

erald

pu

blish

ing

co

m




Management Decision

Management Decision


Number 10


ISBN 978-1-78973-015-9


















Outline placeholder















































































































0025-1747

2061

Editorialadvisory

board

















































































2062

MD5610

Guest editorial







0025-1747DOI 101108MD-10-2018-004

2063

Guest editorial










2064

MD5610







2065

Guest editorial






Napoli Italy







References







2066

MD5610














2067

Guest editorial





2068

MD5610












0025-1747DOI 101108MD-10-2017-1022




2069






2070

MD5610






2071










2072

MD5610





Iden

tific

atio

nSc

reen

ing

Elig

ibili

tyIn

clud

ed




Reason out of scope







N=30


N=69




2073





2002



2003



2009




2010




2011



ManagementDecision

2012




2013




2013




2014

10 Systematicreview




2012

11 Empiricalarticle




2005

12 Empiricalarticle



2007

13 Empiricalarticle




2008

14 Empiricalarticle




2011

15 Empiricalarticle




2011

16 Empiricalarticle



PLoS One 2012

(continued )


2074

MD5610




17 Empiricalarticle



2012

18 Empiricalarticle




2012

19 Empiricalarticle



2012

20 Empiricalarticle




2012

21 Empiricalarticle




2013

22 Empiricalarticle




2013

23 Empiricalarticle




2014

24 Empiricalarticle




2014

25 Empiricalarticle




2014

26 Empiricalarticle




2014

27 Empiricalarticle




2015

28 Empiricalarticle


Nurse Leader 2015

29 Empiricalarticle




2017

30 Empiricalarticle




2017

Table I

2075





Evidence)













Managers

Policy-makers

Researchers


Data Analyses




organization







2076

MD5610











Policy-makers










Researchers





Managers






Healthprofessionals













(continued )


2077








Expertspreferences



therapists(PT))



Expertspreferences



Managers





Managers






Managers






Healthprofessionals






Managers








2078

MD5610











IT leadersManagers






Managers






Managers






Managers


Nelson KENS Pilon B












2079

















Managers

Table II

2080

MD5610



References








2081






















2082

MD5610














2083







2084

MD5610











0025-1747DOI 101108MD-09-2017-0920




2085

Evidence-based

management










2086

MD5610







2087

Evidence-based

management




Source Definition











2088

MD5610





2089

Evidence-based

management









2090

MD5610










2091

Evidence-based

management









However when the conditions are unclear or if the decision makers are unsure whether theconditions are clear then relying on the contingency specification of EBMgt becomesproblematic The problem is not merely an issue of bounded rationality but derives from themathematics of diversity and the epistemological problem of the underspecification of theoriesby evidence To the extent that we know what drives performance ldquowe should select the bestcollection on the basis of that information [hellip] [however] if we are not sure of what wersquoredoing we should err toward greater diversityrdquo (011) particularly ldquoon complex tasksthat involve multiple dimensions or variablesrdquo (Page 2017) The challenge of identifyingwhich parameters should be incorporated in an EBMgt strategy suggests a different solutionDrawing from the cybernetic tradition this paper extends a third interpretation of ldquobestnessrdquo

2092

MD5610






2093

Evidence-based

management






Know aboutproblems





















2094

MD5610







2095

Evidence-based

management





References










2096

MD5610






















2097

Evidence-based

management

























2098

MD5610
























2099

Evidence-based

management




New York NY



2100

MD5610











0025-1747DOI 101108MD-10-2017-0997


13 May 201814 July 2018



2101











2102

MD5610








2103








2104

MD5610





(n=79)

(n=21)

(n=98)

(n=170)

(n=196)(n=2131)

Comorbidity

215

57 462

266

Frailty





2105








2106

MD5610










2107











2108

MD5610









2109







2110

MD5610





ELECTIVEEMERGENT

MEDICAL AREA


OPERATINGTHEATRE

NOAT HOME


YESHOSPITAL

ADMISSION

NOAT HOME

GPsMEDICAL

PRACTICE


YES

NO

EMERGENCYDEPARTMENT


SURGICAL AREA


READYTO BE

DISCHARGED

NEEDFURTHER

ASSISTANCE

Figure 2Flowchart



2111










2112

MD5610











2113









2114

MD5610




EMERGENT

EMERGENCYDEPARTMENT

HOSPITALADMISSION

NOAT HOME

YESMEDICAL AREA

BEDMANAGER

SURGICAL AREA



NO

YES

OPERATINGTHEATRE

READYTO BE

DISCHARGED

NEEDFURTHER

ASSISTANCE

NOAT HOME



GPsMEDICAL

PRACTICE


CENTRALDISCHARGE

GATEWAY (CDG)




FRAILPATIENT


ELECTIVE



2115






Objective Indicator




P




P


P



P


P


P


P


O



2116

MD5610








2117








2118

MD5610

References





















2119



















2120

MD5610



















2121



















2122

MD5610



















2123












2124

MD5610









0025-1747DOI 101108MD-10-2017-1010


19 March 201829 April 2018




2125



Principles

Criticism

sSolutio

nsandim

plications

tomanagers

Challeng

ein

transferring

EBM

into

clinical

practice

1The

bestavailable

practiceshould

beused

2Evidenceshould

bebasedon

anevaluatio

nof

allevidence

3Baseclinical

decisionson

patientsrsquovalues

andpreferences

Randomized

ControlT

ests

(RCT

s)are

considered

ashigh

estqu

ality

evidence

Shortcom

ings

ofRCT

s1RCT

smay

lack

applicability

because

aItisbasedon

ldquoaveragerdquo

rand

omized

patient

bThe

clinical

guidelines

that

are

basedon

RCT

smay

ignore

variablesthat

may

affect

treatm

entoutcom

es2RCT

sdo

notpresentresults

that

consider

effectsof

multi-

orco-

morbidity

3Indu

stry

pressuresmay

indu

cebias

inresultsintrodu

ceconflicts

ofinterest

Solutio

ns1Broadeningof

RCT

popu

latio

nsm

akeRCT

smorerealistic



principle3of

EBM

aAttem

ptto

mitigate

problem

ofno

sing

legu

idelineapplying

specifically

toan

individu

alpatient

bCa

nusethekn

owledg

eof


profile

iTodo

thisn

eedto

usecompu

ting

technology

andinform

atics

iiCa

ndevelopldquom


Implications

ofthesesolutio

nsto

managers

1Managersshould


patient


into

the

healthcare

theirorganizatio

nprovides

2Managersshould

understand

theshortcom

ings

ofgu

idelines

andavoidmeasuring

performance

very

narrow

lywith

respectto

guideline

adherence

3Managersshouldun

derstand

biases

ofresearch

andconsider

theinflu

ence

ofincentives

onall

partsof

theirorganizatio

n4Im

plem

entatio

nof


requ

ires

useof

healthcare

inform

atics

Problems

1Not

enough

timeforph

ysicians

tokeep

upwith

theincreasedrate

ofavailabilityof

evidence

(medical

know

ledg

e)2Researchevidence

istranslated

into

practice

usinggu

idelinesG

uidelin

eshave

issues

aHighdegree


idelines

bGuidelin

esdo


interm

sof

financialor


orotherchanges(th

eseareim

portantto

managers)

cPa

tientsdo

notco-operate

infollowing

guidelinesorhave


andaskfortreatm

entthat

deviatefrom

guidelines

dGuidelin

esthem

selves

arelow

quality

eDogmaticrelianceon

guidelines

isnot

optim

alSolutio

nsIncrease

adherenceanduseof

guidelines

by1Ch

anging

thepatterns


educationremindersp

atient

education

decision

supp

ortincentives

2Im

provingleadership

acceptance

(som

ething

managerscanim

pact)


2126

MD5610





2127








2128

MD5610





2129







2130

MD5610








2131







2132

MD5610







2133







a

e c

dI II

IIIb








EBM

Legend




JointDecision


Primary

Patient Behavior

Physician Behavior


2134

MD5610





2135






2136

MD5610






2137






2138

MD5610









2139






2140

MD5610





2141







2142

MD5610






2143












2144

MD5610


References
















2145
























2146

MD5610














2147







Napoli Italy






2148

MD5610










2149


comparativeanalysis







2150

MD5610





2151


comparativeanalysis








2152

MD5610






Use of visualcues




















(continued )


triage process

2153


comparativeanalysis















2154

MD5610










2155


comparativeanalysis








2156

MD5610









⩽2 16 55⩾3 and ⩽4 68 37W4 16 9



Alpha ED Beta ED





Yes No


Alpha and Beta

2157


comparativeanalysis








725

20

AverageYHS

AverageYED

AverageYTED

ALPHA

BETA

15

10

5

0

6

5

4

No

of N

urse

s

3

2

10 5 10 15 20


25 30 35 40 45


2158

MD5610













fsQCA analysis

2159


comparativeanalysis












2160

MD5610







Variable Measure







2161


comparativeanalysis

Alpha

Beta

Configuration

RAW

COVERAGE

UNIQUE

COVERAGE

CONSIST

ENCY

CONFIGURATION

RAW

COVERAGE

UNIQUE

COVERAGE

CONSIST

ENCY

Simple

POY

HSsimYEDsim

YTEDP

EDP

TM

0442105

00547369

0697674

simPO

YHSYEDY

TEDsim

PEDsim

PTMC

T00392402

00392402

1PO

simYEDsim

YTEDP

EDP

TMC

T0393684

000631577

0653846

POY

HSYEDY

TEDsim

PEDP

TMC

T0172524

0137345

0717517

POY

HSsimYTEDP

EDP

TMC

T0406316

00189474

0725564

simPO

simYHSsimYEDsim

YTEDsim

PEDsim

PTMC

T00652632

00652632

054386

Solutio

ncoverage

0532632

Solutio

ncoverage

0176586

Solutio

nconsistency

0575

Solutio

nconsistency

0765574

Interm

ediate

POY

HSsimYEDsim

YTEDP

TMsim

CT0472258

0104516

0831818

simPO

YHSsimYTEDP

EDP

TMsim

CT0137203

00764015

0918954

POsim

YEDsim

YTEDP

EDP

TMC

T0296774

00309677

0804196

POY

HSYEDY

TEDP

TMsim

CT0114192

00651566

0993228

POY

HSsimYTEDP

EDP

TMC

T0265806

00774436

simPO

simYHSsimYEDsim

YTEDP

EDP

TMC

T0111387

00548768

0938837

simPO

simYHSsimYEDsim

YTEDsim

PEDP

TMC

T0211613

0211613

0811881

POY

HSsimYEDsim

YTEDsim

PEDP

TMC

T00497569

00124166

0984593

POY

HSYEDY

TEDsim

PEDP

TMC

T0154839

00258064

0736196

Solutio

ncoverage

0767742

Solutio

ncoverage

0306993

Solutio

nconsistency

0750315

Solutio

nconsistency

0938447

Com

plex

POsim

YHSsimYEDsim

YTEDsim

PEDC

T0217628

00620239

1simPO

YHSYEDY

TEDsim

PEDsim

PTMC

T00278783

00278782

1simYHSsimYEDsim

YTEDsim

PEDP

TMC

T0309032

00261153

0879257

POY

HSYEDY

TEDsim

PEDP

TMsim

CT0172524

0172524

0946586

POsim

YHSsimYEDsim

YTEDP

TMC

T0311208

00772579

0953333

simPO

YHSYEDY

TEDP

EDP

TMsim

CT0249273

0249273

0958017

simPO

YHSYEDsim

YTEDsim

PEDP

TMsim

CT0198041

00707291

0764706

POY

HSYEDY

TEDsim

PEDP

TMC

T009358

00304679

1PO

YHSYEDsim

YTEDP

EDP

TMC

T0085963

000761694

0918605

Solutio

ncoverage

0635473

Solutio

ncoverage

0449675

Solutio

nconsistency

0870343

Solutio

nconsistency

0956075


2162

MD5610










2163


comparativeanalysis






2164

MD5610







2165


comparativeanalysis






2166

MD5610






References








2167


comparativeanalysis






















2168

MD5610





















2169


comparativeanalysis





















2170

MD5610








2171


comparativeanalysis

















2172

MD5610






2173












2174

MD5610







2175




(1) events

(2) statuses and

(3) triggers



T3 N0

Any T N1-2






Surgical resection






2176

MD5610






BEGIN0

Imaging Detected100


Visit detected100

Surg int detected40

RT detected60

Treated100



End of Treatment

CHT detected10

EOT100

Patient operated40

Not treated yet100








January 27 2000


25 days

14 days

0 days

20 days

20 days

Waiting for therapy



Treated

Not treated yet



2177















2178

MD5610

BE

GIN

No

17

74

in p

ath

BN

o 9

51in

pat

h C

No

141

in p

ath

AN

o 2

30

attiv

a pa

th A

(T

3 N

0)N

o 2

30

is n

ad R

T d

ose

25 G

ray

No

0

nad

RT

dos

e is

25

Gra

yN

o 0

no c

hem

io p

ath

A

No

0ch

emo

is fl

uoro

OR

cap

ecit

pat

h A

N

o 8

3ch

emo

is fl

uoro

OR

cap

ecit

pat

h B

N

o 6

23ch

emo

is fl

uoro

OR

cap

ecit

pat

h C

N

o 1

00

wai

ting

for

surg

ery

B1

2N

o 6

23w

aitin

g fo

r su

rger

y C

No

100

wai

ting

for

surg

ery

A1

2N

o 8

3w

aitin

g fo

r su

rger

y A

3N

o 0

is s

urge

ry p

erfo

rmed

pat

h A

3

No

0

surg

ery

perf

orm

ed p

ath

A3

No

0su

rger

y pe

rfor

med

pat

hA

12

No

55

surg

ery

perf

orm

ed p

ath

B1

2N

o 3

70su

rger

y pe

rfor

med

pat

hC

No

36

is s

urge

ry p

erfo

rmed

pat

h A

12

N

o 5

5

is s

urge

ry p

erfo

rmed

pat

h B

12

N

o 3

70is

sur

gery

per

form

ed p

ath

C

No

36

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

A

No

123

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th A

No

123

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

B

No

623

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th B

No

623

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

C

No

100

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th C

No

100

attiv

a pa

th B

(T

1-4

N1-

2)N

o 9

51at

tiva

path

C (

T4

N0-

3 T

1-4

N3)

No

141

Not

es T

he g

reen

circ

le o

n th

e to

p is

the

entry

poi

nt in

the

clin

ical

pat

hway

lab

eled

as ldquo

begi

nrdquo

and

cont

ains

17

74 p

atie

nts

From

ther

est

art t

he th

ree

bran

ches

A (l

eft)

B (m

iddl

e) a

nd C

(rig

ht)

whi

ch c

onta

in 2

30 9

51 a

nd 1

41 p

atie

nts

resp

ectiv

ely

The

box

es in

bet

wee

nth

e st

ates

repr

esen

t the

trig

gers

that

act

ivat

e th

e re

spec

tive

trans

ition



2179






2180

MD5610







2181






005

004

003

002

001

000

30

Den

sity

40 50 60 70 80

Path APath BPath C



2182

MD5610




No 17 186 18Yes 33 162 14





the 3 paths

10

08

06

04

02

00

0 100 200 300 400

Path APath BPath C




execution

2183









References



2184

MD5610



















2185






2186

MD5610












0025-1747DOI 101108MD-09-2017-0917


29 April 20184 June 2018



2187


hospital sector











2188

MD5610










2189


hospital sector







2190

MD5610




2191


hospital sector






making model




applying DEMATEL

VIKOR application

START




weights




END



2192

MD5610




Yn0kfrac141

akij

1=n

(1)


A frac14

1 a12 a1na21 1 a2n an1 an2 1

26664

37775 (2)

B frac14

1 b12 b1nb21 1 b2n bn1 bn2 1

26664

37775 (3)




LWci frac14

Qnjfrac141 bij

1=nPn

ifrac141

Qnjfrac141 bij

1=n i j frac14 1 2 n (4)

2193


hospital sector

FWci frac14

Qnjfrac141 aij

1=nPn

ifrac141

Qnjfrac141 aij

1=n i j frac14 1 2 n (5)


CI frac14 lmaxnn1

(6)

CR frac14 CIRI

(7)






2194

MD5610


zij frac141l

Xlkfrac141



X frac14 Zs (10)


Xnjfrac141

zij max1p ipn

Xnifrac141

zij

( ) (11)





R frac14Xnjfrac141

tij (13)

D frac14Xnifrac141

tij (14)


y frac14Pn

ifrac141

Pnjfrac141 tij

n2 (15)

2195


hospital sector



A frac14

P1

P2

P3

Pm

SC1 SC2 SCn

f 11 f 12 f 1nf 21 f 22 f 2nf 31 f 32 f 3n

f m1 f m2 f mn

2666666666664

3777777777775 (16)




( ) i frac14 1 2 m (17)



( ) i frac14 1 2 m (18)


Si frac14Xnjfrac141


(19)


0

1A (20)


2196

MD5610


Qi frac14 vSiSn


RRn (21)






Q P 2eth THORN

Q P 1eth THORN

XDQ (24)







Q P meth THORN

Q P 1eth THORN

oDQ (26)


2197


hospital sector










2198

MD5610









2199


hospital sector


Patient (C1)

Age (S1)

Background(S2)



Social andcultural

aspects (S5)

Patientpersonality

(S6)

State ofmedical

equipment (S7)


equipment (S8)



(S10)

Work overload(S11)


Shift pattern(S13)

Laboratmosphere

(S14)

Fatigue (S15)

Drowsiness(S16)


(S17)


(S18)



protocols (S20)


protocols (S21)


(S22)



(S24)




Team work (C6)

Work methods(C5)

Work force (C4)

Environment(C3)

Technology (C2)

H1

H2

Hm

Goa

l E

valu

ate

the

risk

of a

dver

se e

vent

s in

the

hosp

ital s

ecto

r


2200

MD5610






Technology(C2)



Environment(C3)



Work force(C4)



Workmethods (C5)



Team work(C6)




2201


hospital sector






Age

Age

Age

Age

Age

Is

Is

Is

Is

Is

Much less

Much less

Much less

Much less

Much less

Less

Less

Less

Less

Less

Equally

Equally

Equally

Equally

Equally

Moderately

Moderately

Moderately

Moderately

Moderately

Strongly

Strongly

Strongly

Strongly

Strongly

Important than

Important than

Important than

Important than

Important than

Background

Disease Complexity



Patient personality


2202

MD5610







State ofmedical

equipment



State ofmedical

equipment


equipment




equipment





Has

Has

Has

Has

Has

Has

No influence

No influence

No influence

No influence

No influence

No influence

Low influence

Low influence

Low influence

Low influence

Low influence


Mediuminfluence

Mediuminfluence

Mediuminfluence

Mediuminfluence

Mediuminfluence

High influence

High influence

High influence

High influence

High influence


Very highinfluence

Very highinfluence

Very highinfluence

Very highinfluence

Very highinfluence

on

on

on

on

on

on


force cluster)

SI S2 S3 S4 S5 S6

SI 1 214 214 5 3 5S2 047 1 1 228 5 341S3 047 1 1 1 387 451S4 020 044 1 1 451 368S5 033 02 026 022 1 13256 020 029 022 027 076 1



2203


hospital sector



Cluster GW LW



Cluster CR



2204

MD5610



WGc frac14

P1

P2

P3

Py

SC1 SC2 SCz

r11 r12 r1zr21 r22 r2zr31 r32 r3z

ry1 ry2 ryz

2666666666664

3777777777775

GW 1

GW 2

GW 3

GWm

26666666664

37777777775 (27)

WFc frac14

P1

P2

P3

Py

SC1 SC2 SCz

r11 r12 r1zr21 r22 r2zr31 r32 r3z

ry1 ry2 ryz

2666666666664

3777777777775

FW 1

FW 2

FW 3

FWm

26666666664

37777777775 (28)





conditionSocial and


personality



cluster)

2205


hospital sector




conditionSocial and


personality



Cluster GW LW



2206

MD5610






2001801601401201008060402000

Workmethods


Criterion

Glo

bal w

eigh

t

198179

171 167 165

120


hospital sector



157

Background184


414


270


317


249Social and

cultural aspects99


(a) (b)

Figure 6LW values

2207


hospital sector




Laboratmosphere

205

Spaceconditions

219

Shift pattern163

Work overload165


200


248


202


132

Fatigue150

Drowsiness177


139


(a) (b)

Figure 7LW values


263



214


268

Lack ofmonitoring

359

Lack ofLeadership

329


312


(a) (b)

Figure 8LW values

2208

MD5610





C6

C5

C4

C3

C2

C1

0 005 01 015 02 025 03 035 04

Global weight (GW)

01710149

01980116

01650176

0167012

01790071

0120368

Crit

erio

n




2209


hospital sector




014

012

01

008

006

004

002


Sub-criterion

Glo

bal w

eigh

t (G

W)



2210

MD5610


15

05

ndash05315 32 325 33 335 34 345 35

ndash15

ndash2

ndash1

0

15

1

0

ndash1

ndash15

ndash05

05

1

S21

S23

S24

S22

D +R D +R

DndashR

DndashR

204 2045 205 2055 206 2065 207 2075 208 2085 209

S26

S25

S27

(a) (b)



map for criteria

1

08

06

04

02

0

ndash0255 6 65 7 75 8

ndash04

ndash06

DndashR

DndashR

D +RD +R

25

2

15

1

05

0

ndash05

ndash1

ndash15

12 125 13 135 14 145 15 155 16 165

S13

S14

S12

S11

S10 S16

S17

S20

S15

S18 S19

(a) (b)



map for criteria

06

04

02

ndash02

ndash04

ndash06

ndash08

04 45 5 55 6 65 7 75 8 85 9

D+R

DndashR

06

03

04

05

02

0

ndash02

ndash01

ndash04

ndash03

01

DndashR

S6

S1

S4

S3

S2

S5

214 216 218 22 222 224 226 228 23 232 234

D+R

S8

S7

S9

(a) (b)



map for criteria

2211


hospital sector







2212

MD5610









T for criteria

2213


hospital sector







2214

MD5610




































(continued )



2215


hospital sector




























2216

MD5610



P1 0148 1 0033 1 0000 1P2 0369 2 0057 2 0548 2P3 0581 3 0074 3 1000 3






compromise solution

S26S27

S1S2

S3

S4

S5

S6

S7

S8

S9

S10

S11

S12

S13S14S15

S16

S17

S18

S19

S20

S21

S22

S23

S24

S25

006

005

004

003

002

001

0

P1 P2 P3



2217


hospital sector





2218

MD5610


References
















2219


hospital sector






















2220

MD5610


















2221


hospital sector




















2222

MD5610





















2223


hospital sector

















Further reading




2224

MD5610













0025-1747DOI 101108MD-06-2017-0550




2225







2226

MD5610








2227








2228

MD5610










2229









2230

MD5610









2231




Constant 23725 o001






2232

MD5610







2233







2234

MD5610






sensitivity tests

2235






2236

MD5610


References



















2237


















2238

MD5610












0025-1747DOI 101108MD-10-2017-1001




2239


teams









2240

MD5610







2241


teams





2242

MD5610




Metric Definition





Closenesscentrality




Networkconstraint





2243


teams





Team Performance

Connectivity

Degree Centrality





Brokerage Capacity

Network Constraints



4

3

2

1

0Operational

Num

ber

of T

op B

roke

rs

Cost

Team Name

EETF OutcomesPFE

5


2244

MD5610





Outcome

Hospital

University

Industry

Personal

GovernmentIHI

PFE

EEFT

Cost

Figure 3Out-group

communications

2245


teams

12

10

6

4

2

0Outcome


PFE EETF Cost

8 36

28

3638 38

34 36

4 43632

3

Team Performance



Outgroup

Connectivity

200

PFE

PFE

PFE

PFE030

025

020

015

010

005

000

035030

025020

015

005010

0000605

0403

0201

00

EETF

EETF

EETF

EETF

Outcome

Outcome

Outcome

Outcome

Cost

Cost

Cost

Cost

Avg Constraint

Avg Constraint

Avg Betweenness

Avg Betweenness

Cro

ss T

ies

Cro

ss T

ies

Avg

Clo

sene

ssA

vg C

lose

ness

Indegree

Indegree

Outdegree

Outdegree

150

100

50

0

000005

010015

020025

030035

040000 005 010

015 020 025030 035

040

07

06

05

04

03

02

01

00000

005

010

015

0000

100

80

60

40

20

0010

008

004006

00206 000

0504

0302

0100

0002 000400060008001000120014

0016


2246

MD5610





2247


teams







2248

MD5610



Note


References















2249


teams




















2250

MD5610

















2251


teams











2252

MD5610












2253


systems





services


care

PopulationValue



2254

MD5610














2255


systems








2256

MD5610









2257


systems



40

30

20

Day

s

10

0

40

30

20

Day

s

10

0

200

150

100

Day

s

10

0

Bol

zano

Lom

bard

ia

Tren

to

Ven

eto

Em

ilia-

Rom

agna

Pug

lia

FV

G

Um

bria

Tosc

ana

Ligu

ria

Mar

che

Bol

zano

Lom

bard

ia

Tren

to

Ven

eto

Em

ilia-

Rom

agna

Pug

lia

FV

G

Um

bria

Tosc

ana

Ligu

ria

Mar

che

2015

2016

AS

L F

oggi

aO

sped

ale

Val

duce

Azi

enda

PA

Bol

zano

AS

ST

di L

odi

AU

LSS

2 F

eltr

eU

SL

Um

bria

2A

SS

T D

el G

arda

OO

RR

Fog

gia

AS

L Le

cce

orig

gia

Pel

asci

ni -

Gra

v

AU

LSS

10

Ven

eto

Or

AU

LSS

7 P

ieva

di S

olig

oIR

CC

S P

ol S

Mat

teo

AS

ST

dei

Set

te L

aghi

EE

Cas

a S

ollie

voA

ULS

S 1

4 C

hiog

gia

AU

SL

1 Im

perie

seA

SS

T R

hode

nse

AS

ST

Di C

rem

aA

SS

T S

anti

Pao

lo e

Car

loIR

CC

S C

entr

o R

if O

ncol

A

ULS

S 9

Tre

viso

AU

SL

Sud

Est

Friu

li O

ccid

enta

leA

ULS

S 1

2 V

enez

iana

AS

ST

Di V

imer

cate

AO

Ter

niA

O R

eggi

o E

mili

aS

Raf

fael

e -

Mi

Mac

erat

aIs

tPol

iclS

Don

ato

Sen

igal

liaF

erm

oA

SU

I Udi

neA

US

L M

oden

aA

US

L 4

Chi

avar

ese

AU

SL

Cen

tro

AO

U M

oden

aA

O P

erug

iaA

ULS

S 6

Vic

enza

Osp

Eva

ngel

ico

AS

L B

rindi

siA

SL

Bar

letta

-And

ria-T

rani






2258

MD5610













cancer intervention

2259


systems












D18AMAdischarges





C15Mentalhealth










C1Demandmanagement

B28Homecare

B7Vaccinecoverage

B5Cancerscreenings

B4Opioidconsumption






2260

MD5610











2261


systems


Oncologic pathway








2262

MD5610

Scr

eeni

ngex

tens

ion

brea

st

Wai

ting

time

hosp

ice

Scr

eeni

ngD

iagn

osis

Trea

tmen

tE

nd o

f Life

Two

or m

ore

orga

niza

tions

hav

ing

the

sam

e ev

alua

tion

AO

Pad

ova

AU

LSS

16

Pad

ova

Ist

Onc

Ven

eto

(IO

V)

Scr

eeni

ngex

tens

ion

cerv

ix

Scr

eeni

ngad

hesi

onbr

east

Scr

eeni

ngad

hesi

once

rvix

Scr

eeni

ngex

tens

ion

colo

n

Scr

eeni

ngad

hesi

onco

lon

App

ropr

bi

omar

ker

Wai

ting

time

trea

tbr

east

Wai

ting

time

trea

tpr

osta

te

Wai

ting

time

trea

tco

lon

Wai

ting

time

trea

tre

ctum

Wai

ting

time

trea

tut

erus

Adm

issi

ons

over

thre

shol

dbr

east

Inde

x of

disp

ersi

onbr

east

Adm

issi

ons

over

thre

shol

dpr

osta

te

Inde

x of

disp

ersi

onbr

east

o

f(n

ippl

esk

insp

arin

g)su

rger

ies

Pal

liativ

eca

rene

twor

k

012345

Evaluation Sour

ce 2

016

data

mdashav

aila

ble

at h

ttp

perf

orm

ance

sssu

pit

netv

al



2263


systems

012345

Evaluation

Scr

eeni

ngD

iagn

osis

Trea

tmen

tE

nd o

f Life

Two

or m

ore

orga

niza

tions

hav

ing

the

sam

e ev

alua

tion

AO

U C

areg

giA

US

L C

entr

o

Scr

eeni

ngex

tens

ion

brea

st

Scr

eeni

ngex

tens

ion

cerv

ix

Scr

eeni

ng

adhe

sion

brea

st

Scr

eeni

ngad

hesi

once

rvix

Scr

eeni

ngex

tens

ion

colo

n

Scr

eeni

ngad

hesi

onco

lon

App

ropr

bi

omar

ker

Wai

ting

time

trea

tbr

east

Wai

ting

time

trea

tpr

osta

te

Wai

ting

time

trea

tco

lon

Wai

ting

time

trea

tre

ctum

Wai

ting

time

trea

tlu

ng

Wai

ting

time

hosp

ice

Adm

issi

ons

over

thre

shol

dbr

east

Inde

x of

disp

ersi

onbr

east

Adm

issi

ons

over

thre

shol

dpr

osta

te

Inde

x of

disp

ersi

onbr

east

o

f(n

ippl

esk

insp

arin

g)su

rger

ies

Pal

liativ

eca

rene

twor

k

Wai

ting

time

trea

tut

erus

Sour

ce 2

016

data

mdashav

aila

ble

at h

ttp

perf

orm

ance

sssu

pit

netv

al


2264

MD5610









2265


systems












2266

MD5610












2267


systems



Notes



References










2268

MD5610




















2269


systems





















2270

MD5610



















2271


systems











2272

MD5610











0025-1747DOI 101108MD-09-2017-0908



2273










2274

MD5610










2275









2276

MD5610







Understaffing

Teamwork

MissedTreatments


H1H4a

H2

InformationSystems

H4b

H3


2277










2278

MD5610








of variables




2279








2280

MD5610





2281







2282

MD5610



References













2283





















2284

MD5610





















2285
























2286

MD5610







insight



















Man

age

me

nt D

ec

ision

Vo

lum

e 5

6 N

um

be

r 10 2

018

em

erald

pu

blish

ing

co

m




Management Decision

Management Decision


Number 10


ISBN 978-1-78973-015-9


















Outline placeholder

































































































0025-1747

2061

Editorialadvisory

board

















































































2062

MD5610

Guest editorial







0025-1747DOI 101108MD-10-2018-004

2063

Guest editorial










2064

MD5610







2065

Guest editorial






Napoli Italy







References







2066

MD5610














2067

Guest editorial





2068

MD5610












0025-1747DOI 101108MD-10-2017-1022




2069






2070

MD5610






2071










2072

MD5610





Iden

tific

atio

nSc

reen

ing

Elig

ibili

tyIn

clud

ed




Reason out of scope







N=30


N=69




2073





2002



2003



2009




2010




2011



ManagementDecision

2012




2013




2013




2014

10 Systematicreview




2012

11 Empiricalarticle




2005

12 Empiricalarticle



2007

13 Empiricalarticle




2008

14 Empiricalarticle




2011

15 Empiricalarticle




2011

16 Empiricalarticle



PLoS One 2012

(continued )


2074

MD5610




17 Empiricalarticle



2012

18 Empiricalarticle




2012

19 Empiricalarticle



2012

20 Empiricalarticle




2012

21 Empiricalarticle




2013

22 Empiricalarticle




2013

23 Empiricalarticle




2014

24 Empiricalarticle




2014

25 Empiricalarticle




2014

26 Empiricalarticle




2014

27 Empiricalarticle




2015

28 Empiricalarticle


Nurse Leader 2015

29 Empiricalarticle




2017

30 Empiricalarticle




2017

Table I

2075





Evidence)













Managers

Policy-makers

Researchers


Data Analyses




organization







2076

MD5610











Policy-makers










Researchers





Managers






Healthprofessionals













(continued )


2077








Expertspreferences



therapists(PT))



Expertspreferences



Managers





Managers






Managers






Healthprofessionals






Managers








2078

MD5610











IT leadersManagers






Managers






Managers






Managers


Nelson KENS Pilon B












2079

















Managers

Table II

2080

MD5610



References








2081






















2082

MD5610














2083







2084

MD5610











0025-1747DOI 101108MD-09-2017-0920




2085

Evidence-based

management










2086

MD5610







2087

Evidence-based

management




Source Definition











2088

MD5610





2089

Evidence-based

management









2090

MD5610










2091

Evidence-based

management









However when the conditions are unclear or if the decision makers are unsure whether theconditions are clear then relying on the contingency specification of EBMgt becomesproblematic The problem is not merely an issue of bounded rationality but derives from themathematics of diversity and the epistemological problem of the underspecification of theoriesby evidence To the extent that we know what drives performance ldquowe should select the bestcollection on the basis of that information [hellip] [however] if we are not sure of what wersquoredoing we should err toward greater diversityrdquo (Page 2011) particularly ldquoon complex tasksthat involve multiple dimensions or variablesrdquo (Page 2017) The challenge of identifyingwhich parameters should be incorporated in an EBMgt strategy suggests a different solutionDrawing from the cybernetic tradition this paper extends a third interpretation of ldquobestnessrdquo

2092

MD5610






2093

Evidence-based

management






Know aboutproblems





















2094

MD5610







2095

Evidence-based

management





References










2096

MD5610






















2097

Evidence-based

management

























2098

MD5610
























2099

Evidence-based

management




New York NY



2100

MD5610











0025-1747DOI 101108MD-10-2017-0997


13 May 201814 July 2018



2101











2102

MD5610








2103








2104

MD5610





(n=79)

(n=21)

(n=98)

(n=170)

(n=196)(n=2131)

Comorbidity

215

57 462

266

Frailty





2105








2106

MD5610










2107











2108

MD5610









2109







2110

MD5610





ELECTIVEEMERGENT

MEDICAL AREA


OPERATINGTHEATRE

NOAT HOME


YESHOSPITAL

ADMISSION

NOAT HOME

GPsMEDICAL

PRACTICE


YES

NO

EMERGENCYDEPARTMENT


SURGICAL AREA


READYTO BE

DISCHARGED

NEEDFURTHER

ASSISTANCE

Figure 2Flowchart



2111










2112

MD5610











2113









2114

MD5610




EMERGENT

EMERGENCYDEPARTMENT

HOSPITALADMISSION

NOAT HOME

YESMEDICAL AREA

BEDMANAGER

SURGICAL AREA



NO

YES

OPERATINGTHEATRE

READYTO BE

DISCHARGED

NEEDFURTHER

ASSISTANCE

NOAT HOME



GPsMEDICAL

PRACTICE


CENTRALDISCHARGE

GATEWAY (CDG)




FRAILPATIENT


ELECTIVE



2115






Objective Indicator




P




P


P



P


P


P


P


O



2116

MD5610








2117








2118

MD5610

References





















2119



















2120

MD5610



















2121



















2122

MD5610



















2123












2124

MD5610









0025-1747DOI 101108MD-10-2017-1010


19 March 201829 April 2018




2125



Principles

Criticism

sSolutio

nsandim

plications

tomanagers

Challeng

ein

transferring

EBM

into

clinical

practice

1The

bestavailable

practiceshould

beused

2Evidenceshould

bebasedon

anevaluatio

nof

allevidence

3Baseclinical

decisionson

patientsrsquovalues

andpreferences

Randomized

ControlT

ests

(RCT

s)are

considered

ashigh

estqu

ality

evidence

Shortcom

ings

ofRCT

s1RCT

smay

lack

applicability

because

aItisbasedon

ldquoaveragerdquo

rand

omized

patient

bThe

clinical

guidelines

that

are

basedon

RCT

smay

ignore

variablesthat

may

affect

treatm

entoutcom

es2RCT

sdo

notpresentresults

that

consider

effectsof

multi-

orco-

morbidity

3Indu

stry

pressuresmay

indu

cebias

inresultsintrodu

ceconflicts

ofinterest

Solutio

ns1Broadeningof

RCT

popu

latio

nsm

akeRCT

smorerealistic



principle3of

EBM

aAttem

ptto

mitigate

problem

ofno

sing

legu

idelineapplying

specifically

toan

individu

alpatient

bCa

nusethekn

owledg

eof


profile

iTodo

thisn

eedto

usecompu

ting

technology

andinform

atics

iiCa

ndevelopldquom


Implications

ofthesesolutio

nsto

managers

1Managersshould


patient


into

the

healthcare

theirorganizatio

nprovides

2Managersshould

understand

theshortcom

ings

ofgu

idelines

andavoidmeasuring

performance

very

narrow

lywith

respectto

guideline

adherence

3Managersshouldun

derstand

biases

ofresearch

andconsider

theinflu

ence

ofincentives

onall

partsof

theirorganizatio

n4Im

plem

entatio

nof


requ

ires

useof

healthcare

inform

atics

Problems

1Not

enough

timeforph

ysicians

tokeep

upwith

theincreasedrate

ofavailabilityof

evidence

(medical

know

ledg

e)2Researchevidence

istranslated

into

practice

usinggu

idelinesG

uidelin

eshave

issues

aHighdegree


idelines

bGuidelin

esdo


interm

sof

financialor


orotherchanges(th

eseareim

portantto

managers)

cPa

tientsdo

notco-operate

infollowing

guidelinesorhave


andaskfortreatm

entthat

deviatefrom

guidelines

dGuidelin

esthem

selves

arelow

quality

eDogmaticrelianceon

guidelines

isnot

optim

alSolutio

nsIncrease

adherenceanduseof

guidelines

by1Ch

anging

thepatterns


educationremindersp

atient

education

decision

supp

ortincentives

2Im

provingleadership

acceptance

(som

ething

managerscanim

pact)


2126

MD5610





2127








2128

MD5610





2129







2130

MD5610








2131







2132

MD5610







2133







a

e c

dI II

IIIb








EBM

Legend




JointDecision


Primary

Patient Behavior

Physician Behavior


2134

MD5610





2135






2136

MD5610






2137






2138

MD5610









2139






2140

MD5610





2141







2142

MD5610






2143












2144

MD5610


References
















2145
























2146

MD5610














2147







Napoli Italy






2148

MD5610










2149


comparativeanalysis







2150

MD5610





2151


comparativeanalysis








2152

MD5610






Use of visualcues




















(continued )


triage process

2153


comparativeanalysis















2154

MD5610










2155


comparativeanalysis








2156

MD5610









⩽2 16 55⩾3 and ⩽4 68 37W4 16 9



Alpha ED Beta ED





Yes No


Alpha and Beta

2157


comparativeanalysis








725

20

AverageYHS

AverageYED

AverageYTED

ALPHA

BETA

15

10

5

0

6

5

4

No

of N

urse

s

3

2

10 5 10 15 20


25 30 35 40 45


2158

MD5610













fsQCA analysis

2159


comparativeanalysis












2160

MD5610







Variable Measure







2161


comparativeanalysis

Alpha

Beta

Configuration

RAW

COVERAGE

UNIQUE

COVERAGE

CONSIST

ENCY

CONFIGURATION

RAW

COVERAGE

UNIQUE

COVERAGE

CONSIST

ENCY

Simple

POY

HSsimYEDsim

YTEDP

EDP

TM

0442105

00547369

0697674

simPO

YHSYEDY

TEDsim

PEDsim

PTMC

T00392402

00392402

1PO

simYEDsim

YTEDP

EDP

TMC

T0393684

000631577

0653846

POY

HSYEDY

TEDsim

PEDP

TMC

T0172524

0137345

0717517

POY

HSsimYTEDP

EDP

TMC

T0406316

00189474

0725564

simPO

simYHSsimYEDsim

YTEDsim

PEDsim

PTMC

T00652632

00652632

054386

Solutio

ncoverage

0532632

Solutio

ncoverage

0176586

Solutio

nconsistency

0575

Solutio

nconsistency

0765574

Interm

ediate

POY

HSsimYEDsim

YTEDP

TMsim

CT0472258

0104516

0831818

simPO

YHSsimYTEDP

EDP

TMsim

CT0137203

00764015

0918954

POsim

YEDsim

YTEDP

EDP

TMC

T0296774

00309677

0804196

POY

HSYEDY

TEDP

TMsim

CT0114192

00651566

0993228

POY

HSsimYTEDP

EDP

TMC

T0265806

00774436

simPO

simYHSsimYEDsim

YTEDP

EDP

TMC

T0111387

00548768

0938837

simPO

simYHSsimYEDsim

YTEDsim

PEDP

TMC

T0211613

0211613

0811881

POY

HSsimYEDsim

YTEDsim

PEDP

TMC

T00497569

00124166

0984593

POY

HSYEDY

TEDsim

PEDP

TMC

T0154839

00258064

0736196

Solutio

ncoverage

0767742

Solutio

ncoverage

0306993

Solutio

nconsistency

0750315

Solutio

nconsistency

0938447

Com

plex

POsim

YHSsimYEDsim

YTEDsim

PEDC

T0217628

00620239

1simPO

YHSYEDY

TEDsim

PEDsim

PTMC

T00278783

00278782

1simYHSsimYEDsim

YTEDsim

PEDP

TMC

T0309032

00261153

0879257

POY

HSYEDY

TEDsim

PEDP

TMsim

CT0172524

0172524

0946586

POsim

YHSsimYEDsim

YTEDP

TMC

T0311208

00772579

0953333

simPO

YHSYEDY

TEDP

EDP

TMsim

CT0249273

0249273

0958017

simPO

YHSYEDsim

YTEDsim

PEDP

TMsim

CT0198041

00707291

0764706

POY

HSYEDY

TEDsim

PEDP

TMC

T009358

00304679

1PO

YHSYEDsim

YTEDP

EDP

TMC

T0085963

000761694

0918605

Solutio

ncoverage

0635473

Solutio

ncoverage

0449675

Solutio

nconsistency

0870343

Solutio

nconsistency

0956075


2162

MD5610










2163


comparativeanalysis






2164

MD5610







2165


comparativeanalysis






2166

MD5610






References








2167


comparativeanalysis






















2168

MD5610





















2169


comparativeanalysis





















2170

MD5610








2171


comparativeanalysis

















2172

MD5610






2173












2174

MD5610







2175




(1) events

(2) statuses and

(3) triggers



T3 N0

Any T N1-2






Surgical resection






2176

MD5610






BEGIN0

Imaging Detected100


Visit detected100

Surg int detected40

RT detected60

Treated100



End of Treatment

CHT detected10

EOT100

Patient operated40

Not treated yet100








January 27 2000


25 days

14 days

0 days

20 days

20 days

Waiting for therapy



Treated

Not treated yet



2177















2178

MD5610

BE

GIN

No

17

74

in p

ath

BN

o 9

51in

pat

h C

No

141

in p

ath

AN

o 2

30

attiv

a pa

th A

(T

3 N

0)N

o 2

30

is n

ad R

T d

ose

25 G

ray

No

0

nad

RT

dos

e is

25

Gra

yN

o 0

no c

hem

io p

ath

A

No

0ch

emo

is fl

uoro

OR

cap

ecit

pat

h A

N

o 8

3ch

emo

is fl

uoro

OR

cap

ecit

pat

h B

N

o 6

23ch

emo

is fl

uoro

OR

cap

ecit

pat

h C

N

o 1

00

wai

ting

for

surg

ery

B1

2N

o 6

23w

aitin

g fo

r su

rger

y C

No

100

wai

ting

for

surg

ery

A1

2N

o 8

3w

aitin

g fo

r su

rger

y A

3N

o 0

is s

urge

ry p

erfo

rmed

pat

h A

3

No

0

surg

ery

perf

orm

ed p

ath

A3

No

0su

rger

y pe

rfor

med

pat

hA

12

No

55

surg

ery

perf

orm

ed p

ath

B1

2N

o 3

70su

rger

y pe

rfor

med

pat

hC

No

36

is s

urge

ry p

erfo

rmed

pat

h A

12

N

o 5

5

is s

urge

ry p

erfo

rmed

pat

h B

12

N

o 3

70is

sur

gery

per

form

ed p

ath

C

No

36

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

A

No

123

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th A

No

123

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

B

No

623

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th B

No

623

is n

ad R

T d

ose

mor

e or

eq

to 4

5 G

ray

path

C

No

100

nad

RT

dos

e m

ore

or e

q to

45

Gra

y pa

th C

No

100

attiv

a pa

th B

(T

1-4

N1-

2)N

o 9

51at

tiva

path

C (

T4

N0-

3 T

1-4

N3)

No

141

Not

es T

he g

reen

circ

le o

n th

e to

p is

the

entry

poi

nt in

the

clin

ical

pat

hway

lab

eled

as ldquo

begi

nrdquo

and

cont

ains

17

74 p

atie

nts

From

ther

est

art t

he th

ree

bran

ches

A (l

eft)

B (m

iddl

e) a

nd C

(rig

ht)

whi

ch c

onta

in 2

30 9

51 a

nd 1

41 p

atie

nts

resp

ectiv

ely

The

box

es in

bet

wee

nth

e st

ates

repr

esen

t the

trig

gers

that

act

ivat

e th

e re

spec

tive

trans

ition



2179






2180

MD5610







2181






005

004

003

002

001

000

30

Den

sity

40 50 60 70 80

Path APath BPath C



2182

MD5610




No 17 186 18Yes 33 162 14





the 3 paths

10

08

06

04

02

00

0 100 200 300 400

Path APath BPath C




execution

2183









References



2184

MD5610



















2185






2186

MD5610












0025-1747DOI 101108MD-09-2017-0917


29 April 20184 June 2018



2187


hospital sector











2188

MD5610










2189


hospital sector







2190

MD5610




2191


hospital sector






making model




applying DEMATEL

VIKOR application

START




weights




END



2192

MD5610




Yn0kfrac141

akij

1=n

(1)


A frac14

1 a12 a1na21 1 a2n an1 an2 1

26664

37775 (2)

B frac14

1 b12 b1nb21 1 b2n bn1 bn2 1

26664

37775 (3)




LWci frac14

Qnjfrac141 bij

1=nPn

ifrac141

Qnjfrac141 bij

1=n i j frac14 1 2 n (4)

2193


hospital sector

FWci frac14

Qnjfrac141 aij

1=nPn

ifrac141

Qnjfrac141 aij

1=n i j frac14 1 2 n (5)


CI frac14 lmaxnn1

(6)

CR frac14 CIRI

(7)






2194

MD5610


zij frac141l

Xlkfrac141



X frac14 Zs (10)


Xnjfrac141

zij max1p ipn

Xnifrac141

zij

( ) (11)





R frac14Xnjfrac141

tij (13)

D frac14Xnifrac141

tij (14)


y frac14Pn

ifrac141

Pnjfrac141 tij

n2 (15)

2195


hospital sector



A frac14

P1

P2

P3

Pm

SC1 SC2 SCn

f 11 f 12 f 1nf 21 f 22 f 2nf 31 f 32 f 3n

f m1 f m2 f mn

2666666666664

3777777777775 (16)




( ) i frac14 1 2 m (17)



( ) i frac14 1 2 m (18)


Si frac14Xnjfrac141


(19)


0

1A (20)


2196

MD5610


Qi frac14 vSiSn


RRn (21)






Q P 2eth THORN

Q P 1eth THORN

XDQ (24)







Q P meth THORN

Q P 1eth THORN

oDQ (26)


2197


hospital sector










2198

MD5610









2199


hospital sector


Patient (C1)

Age (S1)

Background(S2)



Social andcultural

aspects (S5)

Patientpersonality

(S6)

State ofmedical

equipment (S7)


equipment (S8)



(S10)

Work overload(S11)


Shift pattern(S13)

Laboratmosphere

(S14)

Fatigue (S15)

Drowsiness(S16)


(S17)


(S18)



protocols (S20)


protocols (S21)


(S22)



(S24)




Team work (C6)

Work methods(C5)

Work force (C4)

Environment(C3)

Technology (C2)

H1

H2

Hm

Goa

l E

valu

ate

the

risk

of a

dver

se e

vent

s in

the

hosp

ital s

ecto

r


2200

MD5610






Technology(C2)



Environment(C3)



Work force(C4)



Workmethods (C5)



Team work(C6)




2201


hospital sector






Age

Age

Age

Age

Age

Is

Is

Is

Is

Is

Much less

Much less

Much less

Much less

Much less

Less

Less

Less

Less

Less

Equally

Equally

Equally

Equally

Equally

Moderately

Moderately

Moderately

Moderately

Moderately

Strongly

Strongly

Strongly

Strongly

Strongly

Important than

Important than

Important than

Important than

Important than

Background

Disease Complexity



Patient personality


2202

MD5610







State ofmedical

equipment



State ofmedical

equipment


equipment




equipment





Has

Has

Has

Has

Has

Has

No influence

No influence

No influence

No influence

No influence

No influence

Low influence

Low influence

Low influence

Low influence

Low influence


Mediuminfluence

Mediuminfluence

Mediuminfluence

Mediuminfluence

Mediuminfluence

High influence

High influence

High influence

High influence

High influence


Very highinfluence

Very highinfluence

Very highinfluence

Very highinfluence

Very highinfluence

on

on

on

on

on

on


force cluster)

SI S2 S3 S4 S5 S6

SI 1 214 214 5 3 5S2 047 1 1 228 5 341S3 047 1 1 1 387 451S4 020 044 1 1 451 368S5 033 02 026 022 1 13256 020 029 022 027 076 1



2203


hospital sector



Cluster GW LW



Cluster CR



2204

MD5610



WGc frac14

P1

P2

P3

Py

SC1 SC2 SCz

r11 r12 r1zr21 r22 r2zr31 r32 r3z

ry1 ry2 ryz

2666666666664

3777777777775

GW 1

GW 2

GW 3

GWm

26666666664

37777777775 (27)

WFc frac14

P1

P2

P3

Py

SC1 SC2 SCz

r11 r12 r1zr21 r22 r2zr31 r32 r3z

ry1 ry2 ryz

2666666666664

3777777777775

FW 1

FW 2

FW 3

FWm

26666666664

37777777775 (28)





conditionSocial and


personality



cluster)

2205


hospital sector




conditionSocial and


personality



Cluster GW LW



2206

MD5610






2001801601401201008060402000

Workmethods


Criterion

Glo

bal w

eigh

t

198179

171 167 165

120


hospital sector



157

Background184


414


270


317


249Social and

cultural aspects99


(a) (b)

Figure 6LW values

2207


hospital sector




Laboratmosphere

205

Spaceconditions

219

Shift pattern163

Work overload165


200


248


202


132

Fatigue150

Drowsiness177


139


(a) (b)

Figure 7LW values


263



214


268

Lack ofmonitoring

359

Lack ofLeadership

329


312


(a) (b)

Figure 8LW values

2208

MD5610





C6

C5

C4

C3

C2

C1

0 005 01 015 02 025 03 035 04

Global weight (GW)

01710149

01980116

01650176

0167012

01790071

0120368

Crit

erio

n




2209


hospital sector




014

012

01

008

006

004

002


Sub-criterion

Glo

bal w

eigh

t (G

W)



2210

MD5610


15

05

ndash05315 32 325 33 335 34 345 35

ndash15

ndash2

ndash1

0

15

1

0

ndash1

ndash15

ndash05

05

1

S21

S23

S24

S22

D +R D +R

DndashR

DndashR

204 2045 205 2055 206 2065 207 2075 208 2085 209

S26

S25

S27

(a) (b)



map for criteria

1

08

06

04

02

0

ndash0255 6 65 7 75 8

ndash04

ndash06

DndashR

DndashR

D +RD +R

25

2

15

1

05

0

ndash05

ndash1

ndash15

12 125 13 135 14 145 15 155 16 165

S13

S14

S12

S11

S10 S16

S17

S20

S15

S18 S19

(a) (b)



map for criteria

06

04

02

ndash02

ndash04

ndash06

ndash08

04 45 5 55 6 65 7 75 8 85 9

D+R

DndashR

06

03

04

05

02

0

ndash02

ndash01

ndash04

ndash03

01

DndashR

S6

S1

S4

S3

S2

S5

214 216 218 22 222 224 226 228 23 232 234

D+R

S8

S7

S9

(a) (b)



map for criteria

2211


hospital sector







2212

MD5610









T for criteria

2213


hospital sector







2214

MD5610




































(continued )



2215


hospital sector




























2216

MD5610



P1 0148 1 0033 1 0000 1P2 0369 2 0057 2 0548 2P3 0581 3 0074 3 1000 3






compromise solution

S26S27

S1S2

S3

S4

S5

S6

S7

S8

S9

S10

S11

S12

S13S14S15

S16

S17

S18

S19

S20

S21

S22

S23

S24

S25

006

005

004

003

002

001

0

P1 P2 P3



2217


hospital sector





2218

MD5610


References
















2219


hospital sector






















2220

MD5610


















2221


hospital sector




















2222

MD5610





















2223


hospital sector

















Further reading




2224

MD5610













0025-1747DOI 101108MD-06-2017-0550




2225







2226

MD5610








2227








2228

MD5610










2229









2230

MD5610









2231




Constant 23725 o001






2232

MD5610







2233







2234

MD5610






sensitivity tests

2235






2236

MD5610


References



















2237


















2238

MD5610












0025-1747DOI 101108MD-10-2017-1001




2239


teams









2240

MD5610







2241


teams





2242

MD5610




Metric Definition





Closenesscentrality




Networkconstraint





2243


teams





Team Performance

Connectivity

Degree Centrality





Brokerage Capacity

Network Constraints



4

3

2

1

0Operational

Num

ber

of T

op B

roke

rs

Cost

Team Name

EETF OutcomesPFE

5


2244

MD5610





Outcome

Hospital

University

Industry

Personal

GovernmentIHI

PFE

EEFT

Cost

Figure 3Out-group

communications

2245


teams

12

10

6

4

2

0Outcome


PFE EETF Cost

8 36

28

3638 38

34 36

4 43632

3

Team Performance



Outgroup

Connectivity

200

PFE

PFE

PFE

PFE030

025

020

015

010

005

000

035030

025020

015

005010

0000605

0403

0201

00

EETF

EETF

EETF

EETF

Outcome

Outcome

Outcome

Outcome

Cost

Cost

Cost

Cost

Avg Constraint

Avg Constraint

Avg Betweenness

Avg Betweenness

Cro

ss T

ies

Cro

ss T

ies

Avg

Clo

sene

ssA

vg C

lose

ness

Indegree

Indegree

Outdegree

Outdegree

150

100

50

0

000005

010015

020025

030035

040000 005 010

015 020 025030 035

040

07

06

05

04

03

02

01

00000

005

010

015

0000

100

80

60

40

20

0010

008

004006

00206 000

0504

0302

0100

0002 000400060008001000120014

0016


2246

MD5610





2247


teams







2248

MD5610



Note


References















2249


teams




















2250

MD5610

















2251


teams











2252

MD5610












2253


systems





services


care

PopulationValue



2254

MD5610














2255


systems








2256

MD5610









2257


systems



40

30

20

Day

s

10

0

40

30

20

Day

s

10

0

200

150

100

Day

s

10

0

Bol

zano

Lom

bard

ia

Tren

to

Ven

eto

Em

ilia-

Rom

agna

Pug

lia

FV

G

Um

bria

Tosc

ana

Ligu

ria

Mar

che

Bol

zano

Lom

bard

ia

Tren

to

Ven

eto

Em

ilia-

Rom

agna

Pug

lia

FV

G

Um

bria

Tosc

ana

Ligu

ria

Mar

che

2015

2016

AS

L F

oggi

aO

sped

ale

Val

duce

Azi

enda

PA

Bol

zano

AS

ST

di L

odi

AU

LSS

2 F

eltr

eU

SL

Um

bria

2A

SS

T D

el G

arda

OO

RR

Fog

gia

AS

L Le

cce

orig

gia

Pel

asci

ni -

Gra

v

AU

LSS

10

Ven

eto

Or

AU

LSS

7 P

ieva

di S

olig

oIR

CC

S P

ol S

Mat

teo

AS

ST

dei

Set

te L

aghi

EE

Cas

a S

ollie

voA

ULS

S 1

4 C

hiog

gia

AU

SL

1 Im

perie

seA

SS

T R

hode

nse

AS

ST

Di C

rem

aA

SS

T S

anti

Pao

lo e

Car

loIR

CC

S C

entr

o R

if O

ncol

A

ULS

S 9

Tre

viso

AU

SL

Sud

Est

Friu

li O

ccid

enta

leA

ULS

S 1

2 V

enez

iana

AS

ST

Di V

imer

cate

AO

Ter

niA

O R

eggi

o E

mili

aS

Raf

fael

e -

Mi

Mac

erat

aIs

tPol

iclS

Don

ato

Sen

igal

liaF

erm

oA

SU

I Udi

neA

US

L M

oden

aA

US

L 4

Chi

avar

ese

AU

SL

Cen

tro

AO

U M

oden

aA

O P

erug

iaA

ULS

S 6

Vic

enza

Osp

Eva

ngel

ico

AS

L B

rindi

siA

SL

Bar

letta

-And

ria-T

rani






2258

MD5610













cancer intervention

2259


systems












D18AMAdischarges





C15Mentalhealth










C1Demandmanagement

B28Homecare

B7Vaccinecoverage

B5Cancerscreenings

B4Opioidconsumption






2260

MD5610











2261


systems


Oncologic pathway








2262

MD5610

Scr

eeni

ngex

tens

ion

brea

st

Wai

ting

time

hosp

ice

Scr

eeni

ngD

iagn

osis

Trea

tmen

tE

nd o

f Life

Two

or m

ore

orga

niza

tions

hav

ing

the

sam

e ev

alua

tion

AO

Pad

ova

AU

LSS

16

Pad

ova

Ist

Onc

Ven

eto

(IO

V)

Scr

eeni

ngex

tens

ion

cerv

ix

Scr

eeni

ngad

hesi

onbr

east

Scr

eeni

ngad

hesi

once

rvix

Scr

eeni

ngex

tens

ion

colo

n

Scr

eeni

ngad

hesi

onco

lon

App

ropr

bi

omar

ker

Wai

ting

time

trea

tbr

east

Wai

ting

time

trea

tpr

osta

te

Wai

ting

time

trea

tco

lon

Wai

ting

time

trea

tre

ctum

Wai

ting

time

trea

tut

erus

Adm

issi

ons

over

thre

shol

dbr

east

Inde

x of

disp

ersi

onbr

east

Adm

issi

ons

over

thre

shol

dpr

osta

te

Inde

x of

disp

ersi

onbr

east

o

f(n

ippl

esk

insp

arin

g)su

rger

ies

Pal

liativ

eca

rene

twor

k

012345

Evaluation Sour

ce 2

016

data

mdashav

aila

ble

at h

ttp

perf

orm

ance

sssu

pit

netv

al



2263


systems

012345

Evaluation

Scr

eeni

ngD

iagn

osis

Trea

tmen

tE

nd o

f Life

Two

or m

ore

orga

niza

tions

hav

ing

the

sam

e ev

alua

tion

AO

U C

areg

giA

US

L C

entr

o

Scr

eeni

ngex

tens

ion

brea

st

Scr

eeni

ngex

tens

ion

cerv

ix

Scr

eeni

ng

adhe

sion

brea

st

Scr

eeni

ngad

hesi

once

rvix

Scr

eeni

ngex

tens

ion

colo

n

Scr

eeni

ngad

hesi

onco

lon

App

ropr

bi

omar

ker

Wai

ting

time

trea

tbr

east

Wai

ting

time

trea

tpr

osta

te

Wai

ting

time

trea

tco

lon

Wai

ting

time

trea

tre

ctum

Wai

ting

time

trea

tlu

ng

Wai

ting

time

hosp

ice

Adm

issi

ons

over

thre

shol

dbr

east

Inde

x of

disp

ersi

onbr

east

Adm

issi

ons

over

thre

shol

dpr

osta

te

Inde

x of

disp

ersi

onbr

east

o

f(n

ippl

esk

insp

arin

g)su

rger

ies

Pal

liativ

eca

rene

twor

k

Wai

ting

time

trea

tut

erus

Sour

ce 2

016

data

mdashav

aila

ble

at h

ttp

perf

orm

ance

sssu

pit

netv

al


2264

MD5610









2265


systems












2266

MD5610












2267


systems



Notes



References










2268

MD5610




















2269


systems





















2270

MD5610



















2271


systems











2272

MD5610











0025-1747DOI 101108MD-09-2017-0908



2273










2274

MD5610










2275









2276

MD5610







Understaffing

Teamwork

MissedTreatments


H1H4a

H2

InformationSystems

H4b

H3


2277










2278

MD5610








of variables




2279








2280

MD5610





2281







2282

MD5610



References













2283





















2284

MD5610





















2285
























2286

MD5610







insight



















Man

age

me

nt D

ec

ision

Vo

lum

e 5

6 N

um

be

r 10 2

018

em

erald

pu

blish

ing

co

m




Management Decision

Management Decision


Number 10


ISBN 978-1-78973-015-9


















Outline placeholder
