designing computerized decision support that works for ... · designing computerized decision...

hat(

Designing Computerized Decision Support That Worksfor Clinicians and Families

Alexander G. Fiks, MD, MSCE

Evidence-based decision-making is central to the practice ofpediatrics. Clinical trials and other biomedical research pro-vide a foundation for this process, and practice guidelines,drawing from their results, inform the optimal management ofan increasing number of childhood health problems. However,many clinicians fail to adhere to guidelines. Clinical decisionsupport delivered using health information technology, often inthe form of electronic health records, provides a tool to deliverevidence-based information to the point of care and has thepotential to overcome barriers to evidence-based practice. Anincreasing literature now informs how these systems should bedesigned and implemented to most effectively improve out-comes in pediatrics. Through the examples of computerizedphysician order entry, as well as the impact of alerts at
the point of care on immunization rates, the delivery
a

doi:10.1016/j.cppeds.2010.10.006

60

of evidence-based asthma care, and the follow-up of childrenwith attention deficit hyperactivity disorder, the following re-view addresses strategies for success in using these tools. Thefollowing review argues that, as decision support evolves, theclinician should no longer be the sole target of information andalerts. Through the Internet and other technologies, families areincreasingly seeking health information and gathering input toguide health decisions. By enlisting clinical decision supportsystems to deliver evidence-based information to both clini-cians and families, help families express their preferences andgoals, and connect families to the medical home, clinicaldecision support may ultimately be most effective in improvingoutcomes.

Curr Probl Pediatr Adolesc Health Care 2011;41:60-88

I mproving the safety and quality of health carehas become a priority for our health system.1,2

In this context, health information technology(HIT) is often cited as a necessary foundation forimproving health care delivery.2 This prioritizationas recently had a dramatic effect on public policy. Askey element of efforts to improve the health system,

he American Recovery and Reinvestment ActARRA) of 2009 provided $19bn to promote the

From the Pediatric Research Consortium, Center for BiomedicalInformatics, Center for Pediatric Clinical Effectiveness, Policylab, andthe Division of General Pediatrics at the Children’s Hospital ofPhiladelphia, and the Department of Pediatrics at the University ofPennsylvania School of Medicine, Philadelphia, PA.This research was supported by award Number K23HD059919 fromthe Eunice Kennedy Shriver National Institute of Child Health andHuman Development. The content is solely the responsibility of theauthors and does not necessarily represent the official views of theEunice Kennedy Shriver National Institute of Child Health and HumanDevelopment or the National Institutes of Health.Curr Probl Pediatr Adolesc Health Care 2011;41:60-881538-5442/$ - see front matter© 2011 Mosby, Inc. All rights reserved.

doption of electronic health records (EHRs).3,4 TheseARRA funds are targeted to increase the use of EHRsby pediatric primary care practices from current ratesof nearly 20% and expand opportunities to use HIT toimprove medical decision-making.5,6 With the growthof HIT, clinical decision support systems (CDSS),tools to deliver intelligently filtered, appropriatelytimed, and actionable information to patients or clini-cians, have become a centerpiece of efforts to use HITto improve health care and outcomes.7,8

By reviewing the details of systems functioning indistinct contexts, the goal of this review is to providean understanding of the role of HIT in pediatricdecision-making with a focus on the primary caresetting. Reviews of HIT-based CDSS have docu-mented that effective decision support must be care-fully designed to fit within clinical workflows,9,10 andthese workflows depend on how clinicians and fami-lies approach decision-making and how they choose tointeract. Given this complexity, considering CDSS ina narrow context may limit the examination of unin-tended consequences and not fully capture the impact
of HIT-based CDSS on the decision-making pro-
Curr Probl Pediatr Adolesc Health Care, March 2011

dtwchimraacweedtidast

cess.11 Therefore, we consider the role of HIT broadlyin the setting of different patterns of decision-making.Specifically, we consider 3 prototypes for decision-making in medicine identified through prior research:(1) paternalistic models in which practitioners maketreatment decisions and communicate them to fami-lies; (2) informed patient-models in which patientsreach their own healthcare decisions with informationfrom clinicians or other sources; and (3) models ofshared decision-making (SDM), which depend on theexchange of both medical and personal informationbetween clinicians and families as they arrive at adecision.12,13 As these descriptions illustrate, the flowand content of information that needs to be presentedto or captured from clinicians, patients, and theirfamilies differ in each model.

This review considers the requirements for andpotential benefits of CDSS in the context of these 3distinct styles of decision-making. A conceptualmodel outlining the relationship between styles ofdecision-making, decision support, and health out-

FIG 1. Framework for how decision-making interactions and dthe doctor–patient--parent interaction as well as decision supporpractice guidelines, and patient data, gathered from the famihealth record (EHR) systems, are needed to guide decision supthe decision-making model and the clinical context. Workflowfamilies to access the systems all must be considered. Multipldecision making and are highlighted in the model.

comes is shown in Figure 1 and discussed in more i


etail in the following section. We use this frameworko guide our discussion. By drawing examples fromork in computerized physician order entry (CPOE),

hildhood immunization, asthma, and attention deficityperactivity disorder (ADHD), we highlight promis-ng approaches for using HIT to support decision-aking as well as future directions for research. This

eview addresses the importance of practice guidelinesnd discrete clinical data to drive decision supportlgorithms, explores barriers to gathering these data,onsiders issues in health system design, includingorkflows, usability, and access to CDSS, and then

xplores how the clinical context may shape theffectiveness as well as unintended consequences of aecision support intervention. Throughout, we suggesthat clinical decision support may be most likely tomprove outcomes if it is matched to the model ofecision-making favored by clinicians and families inparticular situation. Developing a growing under-

tanding of these models in distinct pediatric condi-ions and their implications for CDSS will be an

n support impact health outcomes. This figure highlights howy impact health outcomes. Medical knowledge, organized intod past history and captured and organized within electronic. Decision support systems also need to be designed to matche usability of these systems, and the ability of clinicians andtors influence how both clinicians and families participate in

ecisiot maly anports, th

e fac

mportant focus of future research.

61

stc

ripdcwBalicnb

oFwpflunad

pclswuToabhs

A Framework for Thinking Aboutthe Role of HIT in SupportingDecision-Making in Pediatrics

Through family, professional contacts, peers, and themass media, we are frequently exposed to a widerange of new technologies. While it is tempting tofocus on these individual tools and their potential rolesin specific clinical settings, the framework for thisarticle instead focuses on the decision-making processas a whole. Prior reviews of clinical decision supportsystems in the office or hospital setting and theirimpact have widely discussed the importance of work-flows.9,10 For decision-making in the primary careetting, the workflow is shaped by the extent andiming of involvement of the clinician, parent, andhild.14 For these reasons, in the framework presented,

the clinical decision-making model, whether paternal-istic, informed, or shared, is the foundation for the HITsupported decision-making process (Fig 1). Based onthe decision-making model adopted in a particularclinical context, decision support needs differ. Infor-mation may need to be directed at the clinician, parent,child, or all the above. Distinct types of tools havebeen developed in varied settings to meet these needsand are addressed later in this review. Regardless ofthe type of HIT tools used, integrating the system intoclinician workflows, ensuring the usability of thesystem for clinicians and families, and, especially forfamilies, guaranteeing access to the system are criticalto the success of the intervention.

For clinical decision support systems to be effective,they depend on both patient information and clinicalguidelines.15 Therefore, practice guidelines that followa logic that can be adapted for computers as well aspatient data that are organized into retrievable data-bases are both fundamental for functioning CDSS.Challenges inherent in both of these processes arereviewed in sections that follow.

As reflected in Fig 1, the core of any decision-making process in pediatrics is the interaction of theclinician, parent, and child. Given this multipartyprocess, challenges for CDSS can include raisingawareness of the need for a test or procedure, deliv-ering adequate information to address decision-mak-ing needs (eg, information on options and their risksand benefits), effectively sharing knowledge, or help-ing all involved identify and communicate their treat-ment preferences and goals. To meet these needs, HIT
in this framework is presented as surrounding the c
62

interaction between the doctor, patient, and parent, capa-ble of supporting any of these decision-making needs andreaching the clinician, parent, or child (Fig 1).

Prior research has shown that the doctor, patient,parent interaction is not static, but varies across healthconditions, decisions, and contexts.16 Doctors maytake and parents may delegate control over decision-making to the physician in certain situations whileretaining it in others. In addition, patterns of decision-making often vary among clinicians, parents, or chil-dren/adolescents for a particular decision.16 As aesult, the health condition and context are highlightedn Fig 1 as an important input to the decision-makingrocess. Similarly, parents vary in the extent that theyelegate decision-making authority to children, andhildren vary with age and maturity in the extent tohich they participate.17,18 Of note, the Committee ofioethics of the American Academy of Pediatrics hasdvised that, whenever possible, decision-making re-ated to the treatment of older children and adolescentsn routine practice should include the assent of thehild in addition to parental permission.19 These dy-amics influence whether decision support may beest directed at parents, clinicians, or both.For families and clinicians, the literature has dem-nstrated many factors that influence decision-making.or clinicians, knowledge, attitudes, and beliefs asell as the influence of other professionals shape howarticular decisions are approached. Similarly, foramilies, the influence of their own social networks,evels of health and general literacy and numeracy (thenderstanding of quantitative concepts), emotionaleeds related to having a child with a medical concern,nd the child’s maturity level help to shape medicalecision-making.Finally, the framework illustrated emphasizes im-roving health outcomes as the goal of CDSS. Byiting findings from our work and the publishediterature, we highlight examples of how decisionupport may succeed or fail in different contexts. Ase discuss outcomes, we will also focus on potentialnintended consequences of HIT based interventions.hroughout our review of the impact of these systemsn outcomes in pediatrics, we highlight the need fordditional study to better define how these systems cane implemented to improve children’s health andealth care. Historically, the emphasis of decisionupport systems has often been on health care pro-
esses as opposed to health outcomes.

t

etvelmhm

Models of Decision-Makingin Pediatrics

Knowing how best to use HIT to improve deci-sions depends on understanding patterns of deci-sion-making. Patterns of decision-making in pediat-rics fall along a spectrum based on the extent towhich decision-making authority is centered withthe clinician, parent/family, or both (Fig 2). In the“paternalistic model,” clinicians retain control andcommunicate the plan to patients. Traditionally,paternalistic approaches have been favored whenthe medical evidence indicates the clear superiorityof 1 option over another.20 For example, bacterialmeningitis requires prompt antibiotic treatment.Lengthy discussion in that setting increases morbid-ity and mortality. Another setting where paternalis-tic models may be preferable is when patients orfamilies prefer a passive role. While many patientsprefer to be actively engaged in decision-makingand the literature has increasingly focused on pro-moting patient-centered care,21 a range of studiesindicates that, for some, active participation may beunwelcome or have negative consequences. Althoughthe issue is not well studied in pediatrics, breast cancerstudies in the adult setting suggest that, for some, theprospect of participating in decision-making may in-crease distress and anxiety.22,23

In contrast to the paternalistic model, the “informed”model of decision-making places the patient/family incharge of selecting the preferred treatment. Thegrowth of this approach has paralleled the broadertrend toward “consumerism” and the increasing

FIG 2. Patterns of information exchange, participants in delibemodels of decision-making.

availability of electronic resources empowering pa- f


ients/families to take ownership over decisions.24

Those favoring this approach argue that a patientand his or her family may be most invested inexploring all options and may be best positioned tounderstand how available options coincide withtheir values. However, this approach has pitfallssince the medical expertise of the clinician may nolonger act as a filter to separate evidence-based fromill-founded information.15 A variety of cognitiverrors may also impair judgment and lead familieso make choices that are inconsistent with theiralues or not in the best interest of their child. Forxample, testimonials from other patients or fami-ies, even if they do not match the medical evidence,ay guide decisions.25 In pediatrics, the potential

arms of consumerism and “informed” decision-aking are reflected in the growing number of

n and decision-makers in paternalistic, shared, and informed

FIG 3. Components of shared decision-making. (Adapted fromCharles et al.12)

ratio

amilies adopting alternative vaccine schedules that

63

hhdq

bkf

ccmptm

lsgaStrtcac

ssoc

delay timely immunization receipt.26 Although theyave vocal supporters, decisions to delay or with-old vaccination frequently contradict medical evi-ence with potentially life-threatening conse-uences.Unlike the paternalistic and informed models, SDMrings clinicians’ medical expertise as well as parents’nowledge of a child’s specific situation and theamilies’ values to the decision-making process (Fig 3).27

As a result, both the medical evidence and the personalinformation are prioritized and exchanged.12 As mostommonly defined, SDM involves both families andlinicians actively participating in the decision-aking process, the sharing of information, each

arty expressing decision-making preferences, andhe family and clinician jointly arriving at a treat-ent decision.While research on SDM in pediatrics has been

imited, results do suggest potential benefits. Atudy of otitis media presented parents with vi-nettes that featured either paternalistic or sharedpproaches to decision-making.28 Parents in theDM group were less likely to use antibiotics in

hese hypothetical scenarios and more likely toeport being satisfied with the care they received. Inhe behavioral health setting, interventions to helplinicians improve communication with familiesround behavioral health improved outcomes, espe-ially for minority families.29 Based on these results

in pediatrics and the results of trials largely in theadult setting, SDM has increasingly become a fa-

FIG 4. Barriers to physician adherence to practice guidelinesCabana et al.41 Copyright 1999 by the American Medical A

vored model of decision-making.13,30

64

Health Information Technology andDecision-Making

Translating Research into Practice: TheArgument for Health Information Technology-Based Decision Support Systems

Regardless of the decision-making model used, theprocess of integrating research findings into dailymedical practice represents an ongoing challenge.Even under ideal circumstances, the literature oninnovation suggests that it may take 5-10 years ormore for research findings to be disseminated.31,32

Fortunately, pediatricians have been leaders in thedevelopment of guidelines to support care. Althoughclinical trials continue to bolster the evidence base, asof 2005, pediatricians had 215 available practiceguidelines.33 In addition, after reviewing a subsetusing the Appraisal of Guidelines for Research andEvaluation instrument, researchers showed that pedi-atric guidelines, especially those published or en-dorsed by the American Academy of Pediatrics (AAP)or registered in the National Guideline Clearinghouse,were of a higher quality than those in other fields.33

Within this group, guidelines for asthma, acute otitismedia,34 otitis media with effusion,35 ADHD,36,37

febrile seizures,38 and obstructive sleep apnea39 areamong the most prominent in primary care. Overall,guideline development in pediatrics has been facili-tated by policies from the AAP Steering Committee onQuality Improvement and Management that provide

lation to behavior change. (Reproduced with permission fromiation.)

in re

standards for this process.40


TatAelgcistgwilntust

gisnmptbsoglflmt

niandebcmpi

ast

go

However, obstacles to disseminating research find-ings in clinical practice are reflected in the literature onguideline implementation and have implications forthe design and implementation of CDSS. In their 1999review, Cabana and coworkers evaluated barriers toguideline adoption in the context of physician behav-ior change and found that barriers exist in terms ofclinician knowledge, attitudes, and behavior41 (Fig 4).

he volume of new information, limited clinician time tossimilate new information, and a potential lack of accesso or awareness of guidelines present knowledge barriers.mong clinicians, attitudinal barri-

rs to following guidelines include aack of agreement with guidelines ineneral, lack of agreement with spe-ific guidelines, a sense that follow-ng guidelines will not lead to de-ired outcomes, a lack of confidencehat it will be possible to followuidelines in practice, and comfortith the status quo. When implement-

ng guidelines, clinicians can be chal-enged by patient preferences that doot correspond to guidelines, difficul-ies in applying guidelines to a partic-lar situation, and practical barriers,uch as the time and resources neededo implement the guideline.Given these barriers, the rates ofuideline adoption in pediatrics, asn other fields, are low. Findingsuggest that many pediatricians doot view guidelines as helpful anday not even be aware of many

reventive guidelines and that fewerhan 40% change their behaviorased on guidelines.42 A nationalurvey found that a maximum of 27%f pediatric clinicians use any givenuideline.43 Those that were simple and flexible were mostikely to be used. Results such as these have prompted callsor improved strategies to help clinicians adopt guide-ines.44,45 Clinical decision support systems are among the

ost prominent and well-studied approaches for achievinghis goal.

Clinician-Focused Decision SupportClinician-focused decision support systems are well

Findings sugpediatricianguidelines a

not even bmany p

guidelines, athan 40%

behavioguidelines. Rthese have pfor improve

help cliniguidelines. C

support samong the m

and weapproaches

this

matched to situations when decision-making is pater- c


alistic or it is technically most feasible to delivernformation to the clinician (as opposed to the family)t the point of care. Although additional research iseeded in this area, CDSS may also be helpful whenecisions are shared if they prompt clinicians tongage families in discussion or motivate families toe active participants. If carefully implemented, clini-ian-focused systems have been shown to overcomeany of the barriers to guideline adoption. For exam-

le, the team that develops the system and timesnformation delivery to the clinician removes the

burden of guideline awareness anddetailed knowledge of a particularguideline from the physician. Witha well-designed CDSS, the infor-mation simply appears at the ap-propriate time. If the local imple-mentation of the CDSS is plannedwith respected clinical experts andadministrative leaders, doubtsabout the quality of the recom-mendations may be mitigated.Also, if the system provides toolsto facilitate implementation of therecommendations, such as patientinstructions, clinician concernsabout families’ ability to follow-through on recommendations mayalso be reduced.

Strategies for ImplementingEffective Clinical DecisionSupport Systems

While CDSS have many poten-tial benefits, clinicians may resistthe implementation of these sys-tems for many of the same reasonsthey resist practice guidelines.Several reviews have highlighted

effective strategies for implementing decision supportsystems that overcome common barriers and effec-tively change clinician behavior. These strategies, asdescribed by Bates and colleagues in 2003 based ontheir experience with multiple decision support sys-tems (Fig 5),9 include (1) Ensuring that remindersppear without delays that slow workflows. Studiesuggest that clinician users prioritize speed more thanhe ability of a system to improve quality or reduce

t that manyo not viewelpful, mayaware ofentivethat fewernge theirsed on

ults such asmpted callstrategies tons adoptical decisionems aret prominenttudiedr achievingal.

gess ds herevndchar baesro

d scialinystosll-sfo

osts; (2) Anticipating the needs of clinician users and

65

dfrtpw

mpwd

bbaasate

be

delivering information without requiring the clinicianto search for it. System designers should recognizethat these needs may be “latent” or not yet consciouslyrealized; (3) Fitting decision support into users’ work-flows; (4) Prioritizing usability. Experience has dem-onstrated that seemingly small changes in how infor-mation is delivered can dramatically alter howclinicians respond; (5) Avoiding interventions thatrequire clinicians to stop. For example, if canceling anorder is recommended, suggest an alternative; (6)Encouraging clinicians to change direction instead ofstopping is more effective. This type of approach maybe particularly useful when a given drug should beordered at a different dose or fre-quency; (7) Emphasizing simplicity.If the presentation of a guideline oralert is complex, it is more likely tobe overlooked; (8) Whenever possi-ble, avoid asking for additional in-formation. Experience suggests thatthe more data elements that are re-quired from the clinician, the lowerthe likelihood that the decision sup-port will succeed; (9) Monitoringimpact, getting feedback, and re-sponding. This approach is espe-cially important immediately fol-lowing the implementation of adecision support system to ensurethat problems are recognized andaddressed before clinicians bypassthe system; (10) Managing andmaintaining the system. Althoughthis aspect of decision support im-plementation is often overlooked,ongoing maintenance is critical toensure that a decision support sys-tem functions as intended. For example, decision supportrules often need to be revised when updates to guidelinesare released.

In 2005, Kawamoto and colleagues conducted asystematic review of 70 research studies to furtherdefine optimal approaches for CDSS implementa-tion.10 Their work confirmed findings from Bates andistilled 4 rules for successful decision support. Theyound that decision support was most effective whenecommendations were automatically provided withinhe context of the workflow, recommendations wererovided instead of just assessments, decision support

While CDSpotential ben

may rimplementasystems forsame reaso

practice guidreviews hav

effective simplemensupport s

overcome coand effect

clinician

as provided at the time and location of decision- p

66

aking, and when decision support was computer vsaper-based. The importance of automated promptsas confirmed in a second systematic review con-ucted at the same time.46

Creating Effective Implementation Teams

Teamwork between different technical, administra-tive, and clinical experts is needed to effectivelyimplement CDSS and should also be prioritized in thedevelopment process (Fig 6).7 Input from practicingclinicians is necessary to understand needs, integrateCDSS into workflows, understand subtle points thatmay increase the effectiveness of the system, and,

overall, to ensure that the imple-mentation is consistent with styleof practice used in a given setting.These needs likely explain whyCDSS evaluated by clinician ex-perts who were also involved intheir implementation have provenmore effective.46 Depending onthe type of system under develop-ment, involvement of medical as-sistants, nurses, nurse practitio-ners, or physicians may also beneeded. Furthermore, inclusivityearly in the process may limit re-sistance to adoption once the inter-vention is developed.

Administrative expertise furthercontributes to the success ofCDSS. Administrators can helpdevelopers best align a given in-tervention with health system pri-orities, with which they are oftenmost familiar. In this context, ad-ministrators can help frame CDSS

as a vehicle to achieve a clearly recognized goal asopposed to a burden or threat to clinician autonomy.For example, EHR alerts improved billing for pulseoximetry performed in a primary care setting.47 Theilling problem had been identified by administratorsut was relevant to clinicians as well. In addition,lthough clinical guidelines may endorse a range ofpproaches in a given situation, the development ofimple and effective CDSS often depends on thebility to suggest a single recommendation. Based onheir distinct training and perspective, administrators,specially if they also have clinical expertise, may be

ave manyts, cliniciansst then of theseany of thethey resistnes. Severalighlighted

tegies fordecision

ems thaton barriers

ly changehavior.

S hefiesitiom

nselie htratingystmmive

articularly helpful in building consensus among


msn

Soigitgimeroeci

stakeholders in these situations. Finally, as the exam-ple above illustrates, the implementation of a CDSSoften has either explicit or implicit economic conse-quences. Being mindful of these implications duringthe planning stage may help ensure the long-termsustainability of a project.

Many of the tenets for the effective development ofCDSS depend on technical expertise. Questions ofspeed, developing strategies to integrate alerts intoworkflows, programming the system so that decisionsupport can be followed without stopping clinicians,simplifying content delivery, monitoring the successof the intervention, and maintaining the system allrequire technical skill. Increasingly, universities aredeveloping departments of medical informatics, a fieldthat integrates information science, computer science,and health care. Compared with computer science, thisfield emphasizes the application of technology tohealth problems as opposed to focusing exclusively oncomputers or other pieces of hardware.48 Experts in

edical informatics, who are often clinicians them-elves, are valuable partners in achieving these tech-ical goals.

Building Clinician-Focused Decision SupportSystems: Where to Begin?

Clinicians, administrators, and medical informaticsexperts considering implementing decision supportsystems often identify target conditions to addressfrom a list of competing priorities. In their text,“Improving Outcomes with Clinical Decision Support:An Implementer’s Guide,” Osheroff and colleagues

FIG 5. Ten commandments for effective clinical decision sup-port. The 10 commandments refer to evidence-based ap-proaches for effectively implementing clinical decision support.(Data from Bates et al.9)

present an approach for clarifying this decision7 (Fig 7).


pecifically, they measure the value of a project basedn patient impact, organizational impact, clinicianmpact, the number of patients positively affected, theap between ideal and actual behavior pertinent to thentervention, the difficulty associated with addressinghe objective, and the cost involved in meeting theoal. If 1 of these elements has an overwhelmingmpact, the authors argue that it should be weighedore heavily. In addition, the relative importance of

ach of these elements may differ if the goal isesearch, quality improvement, or achieving a businessbjective. In any context, reviewing each of theselements should help project teams more systemati-ally determine the likely benefit from creating andmplementing a particular decision support system.

Prerequisites for Health InformationTechnology-Based Decision Support:Accessible, Organized Patient Data andLogically Structured Practice Guidelines

In addition to an understanding of how decision-making unfolds in a particular clinical situation andthe creation of an effective implementation team, thesuccess of HIT-based decision support depends onaccessible and actionable patient data as well asguidelines that summarize medical evidence in amanner that can guide programmable logic (Fig 8).Guidelines provide the evidence-based rules to specifyhow care should be delivered and define the patientinformation needed to determine a recommended ac-tion. Complexities in both the structure of guidelinesand the organization of patient data present challengesto HIT-based decision support. These complexities aswell as strategies to address them are highlighted inthe paragraphs below.

The Organization of Patient Information

We begin with a discussion of the organization andcapture of patient data in electronic systems and focusfirst on data within EHRs, a frequent source of patientinformation to inform CDSS. The special require-ments for data within pediatric EHRs were highlightedin a 2007 publication from the AAP Council onClinical Information Technology.49 These include theability to record and manage immunization data, theability to track growth, the support of pediatric weight-based medication dosing, the inclusion of pediatricdata norms, and the protection of private information
for special populations, such as adolescents. In advo-
67

ttmop

ecisi

pfrt(

cating for these requirements for any pediatric EHRsystem, the group highlighted the additional benefit ofthese features for many nonpediatric populations.These functions also provide a foundation for creatinga standard data set to support decision-making formany pediatric conditions.

Despite this guidance and a widespread recognitionof the benefits of standardization in the capture ofpatient data, the use of EHRs has generated someconcern among physicians who feel that the computer

FIG 6. Teamwork necessary to effectively implement clinical d

FIG 7. Factors affecting the desirability of implementing aarticular clinical decision support intervention. These areactors that should be considered by an organization oresearch team when considering 1 or multiple possible con-exts for implementing a clinical decision support system.Adapted from Osheroff et al.7)

may compromise doctor–patient–family communica-

68

ion. The historical context is useful in understandinghis perspective. The present day problem-orientededical record (POMR) has its roots in efforts to

rganize patient information in a way that facilitatesatient care, research, and education.50 Prior to the

development of the POMR, medical data were enteredin charts with a focus on the visit and little attention totracking problems over time. Before the eventualwidespread recognition of its value, the POMR wasinitially met with skepticism by many clinicians whowere reluctant to adopt a new approach to organizing

on support systems.

FIG 8. Inputs for effective clinical decision support. Both usefulpractice guidelines and organized patient data are needed fora clinical decision support system to effectively function inproviding clinically useful, patient-specific recommendations.

medical information.


o“miFoCttIaadsnti

The implementation of the EHR within practicesrepresents another major transition in how patientinformation is captured and retrieved. In the EHR, thePOMR in many ways has been translated into elec-tronic form as tools often exist to review chronicproblems, laboratory results, or diagnoses that spanmany encounters. However, although many praise theability of these systems to capture discrete data thatcan be used to measure and improve the quality ofcare, critics argue that the ability to enter an unstruc-tured narrative in the note can be crucial to adequatelyconveying a patient’s condition.51

As this tension highlights, a balance is neededbetween structured and unstructured data. In the EHR,clinical data may be unstructured and entered as freetext, entered as text in response to open-endedprompts, entered using a series of multiple-choiceoptions, or hard-coded “discrete data” entered from afinite list of options and then transferred from the noteinto a database. Because of this variability, a funda-mental step in the design of any CDSS is reviewinghow the data needed to provide decision support for aparticular problem are structured within the EHR.Unfortunately, for developers of CDSS and cliniciansor researchers seeking to improve the quality of care,the capture of information often varies across institu-tions and clinicians who may develop their own stylesof EHR use, even if they use the same EHR system. Inthe ideal scenario, the data fields needed to address aparticular problem would be captured as discrete datain mandatory fields. As the required data become lessand less structured, “text mining,” a process of search-ing the narrative for key phrases, becomes much moreimportant.52 When data are not captured reliability orin a fashion that can be extracted, fields may need tobe custom built for a project to proceed. As will bediscussed in more detail later, personal health records(PHRs), patient portals, office-based kiosks, or formswithin the EHR to facilitate data capture may beparticularly helpful tools in settings where EHR dataare inadequate to inform CDSS.

Data Quality

Any discussion of data quality and capture warrantsconsideration of common data errors that can under-mine CDSS. Inaccuracies in the EHR arise through avariety of mechanisms. Human error is a commoncause of mistakes. For example, a clinician mightindicate a normal cardiac examination in the physical
examination section of a patient’s chart, but elaborate

n a murmur in the assessment and plan. Defaults ofnormal” in the EHR can facilitate this type ofistake. Limitations in the range of choices available

n the EHR may also contribute to poor data quality.or example, a clinician may want to code “left acutetitis media” as a diagnosis, but the Internationallassification of Diseases, 9th Edition, does not permit

he designation of laterality.53 While alternative sys-ems for entering more specific diagnoses, such asntelligent Medical Objects (Northbrook, IL), arevailable, transitions between coding systems maylso introduce data inconsistencies that complicate theevelopment of CDSS. Finally, health systems them-elves may merge, delete, or update medical recordumbers. CDSS systems need to be responsive tohese updates to effectively support decision-makingn “real-time.”

Synthesizing the Medical Evidence: ClinicalPractice Guidelines

Inaccurate, disorganized, or inaccessible patient datacompromise any effort to effectively support decision-making using HIT. Similarly, medical evidence mustbe available and organized to effectively trigger anydecision support system. Clinical practice guidelinesoutline the path to follow in a given clinical situationand therefore function as the roadmap for logic em-bedded in any CDSS. As stated above, a substantialnumber of guidelines already exist in pediatric prac-tice. However, many guidelines have not been writtenwith the logical framework needed to inform computer-based algorithms.

For example, in the case of the AAP acute otitismedia practice guideline, the first recommendationreads: “To diagnose acute otitis media the clinicianshould confirm a history of acute onset, identify signsof middle-ear effusion, and evaluate for the presenceof signs and symptoms of middle-ear inflammation.”34

To develop a rule based on this recommendation,definitions are needed for such terms as “acute,”“middle-ear effusion,” and “middle-ear inflamma-tion.” While these rules must be understood by clini-cians, the logic must also be programmable for HIT-based decision support to be implemented. Only oncethese definitions are established and rules outlined cana computer system use patient data to determine ifcriteria for a given condition or treatment are met.

To address the gap between the language in guide-
lines and the requirements of computer-based rules,
69

tt

tsfinaooicplndht

experts in CDSS now participate in many committeesgenerating clinical guidelines. In addition, scholarshave created frameworks for adapting guidelines forcomputer-based decision support. Among the mostprominent is the Guideline Elements Model developedby Karras and coworkers and Shiffman and cowork-ers.54,55 In this system, a hierarchical template is usedthat includes markers for the different elements neededto operationalize the guideline. This system drawsheavily on the original text of the guideline. Elementsconsidered include the target clinical user, the in-tended patient population, and the required knowledgeelements. Once these are outlined, the required dataelements, conditions, and actions, effectively a seriesof “if . . . then” statements, are specified. Through thisprocess, the practice guideline is decoded into logicalelements in preparation for programming. Of note,clinical expertise is needed as these steps proceed toensure that the result of this process is consistent withstandards of care and the final product remains clini-cally relevant.

Computerized Physician Order Entry:A Prototype Proving the Effectiveness ofClinician-Focused Decision Support

The emphasis on clinician-focused decision supportsystems is warranted since HIT-based CDSS havealready improved clinical practice. Early work dem-onstrating the value of clinician-focused decision sup-port systems evaluated CPOE, the process of enteringorders for medication or procedures into an EHRequipped with logic to identify potential errors. CPOEwas and remains a prime target for clinician-focuseddecision support since, although both families and clini-cians may be involved in choosing a preferred course oftreatment, clinicians are responsible for ordering medi-cations. In addition, medication errors are common,potentially fatal, and preventable.1

Given how well the clinical context in CPOE ismatched to clinician-focused decision support, thesuccess of CDSS in this area is not surprising.

Reviews of studies of CPOE indicate significantrates of error reductions with these systems, with areduction of �50% in serious medication errors ininpatient settings.56,57 In the adult primary care set-ing, researchers found that advanced prescribing sys-ems could prevent 95% of adverse drug events.58

While the overall literature on decision support em-
phasizes simplicity, researchers in this setting found
70

hat more basic systems that only suggested certaintandard doses and required completion of mandatoryelds, such as drug, dose, quantity, and duration, wereot effective. Rather, more complex systems with dosend frequency checking were needed to improveutcomes. Additional work in the field has also dem-nstrated that computerized prescribing advice canncrease the initial medication dose, increase serumoncentrations of medications, reduce times to thera-eutic stabilization, reduce the risk of toxic drugevels, and reduce the length of hospital stay.59 Ofote, the benefit of CPOE in pediatrics has also beenemonstrated. Studies suggest that 93% of pediatricospital adverse drug events could be preventedhrough CPOE.60 However, the impact of these sys-

tems may be reduced in pediatrics if systems are notcustomized to address the unique challenges in select-ing and dosing medications for children.61

Clinician-Focused Decision Support inPediatric Primary Care

As the example of CPOE illustrates, clinician-fo-cused decision support systems can transform thedelivery of pediatric care in high-risk inpatient set-tings. However, most pediatric care is delivered in theprimary care setting. For pediatricians in primary carepractice, CDSS has the potential to improve everydayprocedures, such as immunization delivery as well asthe management of chronic illness. In the followingparagraphs, we review work in the fields of immuni-zation, asthma care, and ADHD as examples of thestrengths and limitations of outpatient pediatric CDSS.In this discussion, the data elements needed to informthese systems, the challenges in developing systemslikely to change behavior, as well as the impact of usingthis technology on health outcomes are highlighted.

Pediatric Immunization Delivery

Like CPOE, pediatric immunization delivery is anideal setting to examine clinician-focused decisionsupport. These systems provide decision support in theform of clinical alerts, which notify the clinician whenvaccines are due. Many aspects of immunizationdelivery argue for the use of technology to enhancecare. Both the facilitators and the barriers to imple-menting immunization decision support are high-lighted in Fig 9. First, thanks to the efforts of the
AAP’s Council on Clinical Information Technology,

otdaAtsanesoa

a1mudanmpaoiecTpdansd

immunization information is generally captured asdiscrete data within the EHR.49 When vaccines arerdered within the EHR, they automatically populatehe system’s database. In addition, clinicians prioritizeocumenting immunization data because it is vieweds 1 of the most important functions of primary care.s a result, pediatric offices routinely develop systems

o enter data on immunizations administered in otherettings, for example, in the case of influenza vaccinesdministered at schools or when children move to aew practice, into the EHR. Furthermore, thanks to thefforts of pediatricians and others, local, regional, ortatewide immunization information systems (previ-usly known as registries) are available in many areasnd can help fill gaps in vaccine records.62 The annual

publication of a standard vaccination schedule alsoprovides rules to guide the implementation of immu-nization decision support so the adaptation of immu-nization rules for CDSS is simplified. Finally, vaccinesare delivered in the office setting, making it possiblefor clinician-focused decision support to directlychange practice and improve outcomes.

Challenges to effectively implementing clinicalalerts, even in the setting of an ideal prototype forclinician-focused decision support, such as immuniza-tion, highlight the complexity of delivering effectivedecision support in primary care. For example, evenwith a nationally standardized vaccine schedule, thelocal implementation of the schedule often varies. The

FACILITATORS• Immuniza�on informa�on is captured

as discrete data within the EHR

• Clinicians priori�ze documen�ng immuniza�on dataimmuniza�on data

• Local, regional, or state-wide immuniza�on informa�on systems (previously known as registries) are

il bl i d h l fillavailable in many areas and can help fill gaps in vaccine records

• Annual publica�on of a standard vaccina�on schedule provides rules to pguide the implementa�on of immuniza�on decision support

• Vaccines are delivered in the office se�ng so providing decision support tose�ng so providing decision support to clinicians can directly change clinician prac�ce and improve outcomes

FIG 9. A case study: facilitators and barriers to implementing ionline.)

2010 schedule specified that the diphtheria, tetanus,


nd pertussis vaccine is recommended between 15 and8 months of age and may be given as early as 12onths of age.63 However, within this range, individ-

al practices and even particular clinicians may chooseifferent ages to begin administering the vaccines. Inddition, individual practices may use distinct combi-ation vaccine products. As mentioned above, theost effective decision support systems recommend 1

articular course of action. However, to determine thege to begin to alert for a given vaccine and the rangef combination products to consider, a local consensuss needed. Ideally, this discussion should take placearly in the development of the system and shouldombine clinical, administrative, and technical experts.his teamwork is especially important in the setting ofediatric vaccination since the vaccine schedule is up-ated at least annually and sometimes more frequentlynd shortages may arise, prompting temporary updates toational recommendations. The follow paragraphs de-cribe results from the evaluation of 2 immunizationecision support systems developed by our team.

Vaccine Alerts for Routine Immunizations inYoung Children in the Urban Setting

Of all vaccines, those for infants and toddlers aregenerally most highly prioritized by pediatricians as ameans to prevent life-threatening illness. Despite thisfocus, the National Immunization Survey continues to

• Even with a na�onally standardized

BARRIERS• Even with a na�onally standardized

vaccine schedule, local implementa�on of the schedule o�en varies

• To determine the age to begin to alert for a given vaccine and the range of combina�on products to consider, a local consensus is needed

• Vaccine shortages may emerge quickly with li�le no�ce

nization decision support. (Color version of figure is available
mmu
report rates of coverage for all due vaccines at less

71

Ho

tittv

civbgea

t

than 70% at 2 years of age.63 Low rates promptedealthy People 2010 to designate immunization as 1f the 10 leading health indicators for the nation.64

The problem of improving rates in impoverished,urban settings is especially challenging for multiplereasons, including frequently missed appointmentsthat result in vaccine delays for young children.65

Beginning in September 2004, our group focused onaddressing this problem by designing a system thatwould alert nurses or clinicians when any vaccinedelayed child under 24 months of age arrived in theoffice and prompt a specific recommendation regard-ing due vaccines. In response to high rates of missedwell-child visits, we worked with clinicians, adminis-trators, and our technical team to develop a system thatwould alert at the earliest possible opportunity. Alertsappeared prominently at the top of the screen when-ever a chart was opened in the EHR (Fig 10). Inaddition, the alert system was designed to notify theuser of invalid doses, such as vaccines given at tooyoung an age or too close to the preceding dose (Fig 11).This approach was meant to minimize the risk that thesystem would confuse clinicians, leading to distrust inthe system.

The project impacted care on a large-scale andresulted in a clinically meaningful increase in vacci-nation rates. Two urban practices were randomized toa control group and received usual care through theEHR. Two practices received the alerts. At the 2intervention practices, alerts appeared at 15,928 visitsover a 1-year period.66 Rates of captured immuniza-ion opportunities, defined as at least 1 vaccine givenf vaccines were due at a visit, increased from 78.2%o 90.3% at well visits and nearly tripled from 11.3%o 32.0% at sick visits (Fig 12). Overall up-to-date

FIG 10. Clinical alert in the EHR. Specific recommendations arecommended when indicated. The system is also designed tothe order entry screen. Links to guidelines and to information f(Color version of figure is available online.)

accination rates for children in the population in- b

72

reased from 81.7% to 90.1% from the control tontervention period. In addition, children in the inter-ention group became up-to-date on all vaccines dueefore 2 years of age earlier than those in the controlroup. The success of the intervention prompted us toxplore the impact of the alerts in other populationsnd for other conditions.

Vaccine Alerts for Influenza Vaccine AmongChildren with Asthma

In 2006-2007, the Centers for Disease Control andPrevention encouraged the medical and public healthcommunities to develop plans to improve influenzavaccine delivery.67 Concurrently, the AAP highlightedthe importance of improving vaccination rates amongchildren with high-risk conditions.68 In response to thisguidance, we conducted a clinical trial to improve vac-cination rates for children with asthma in a multi-stateprimary care practice-based research network.69 At thatime, universal vaccination with influenza vaccine had

FIG 11. Vaccine alert systems can provide detailed informationon prior invalid doses that were given too early or too closetogether. Invalid doses are listed to minimize confusion regard-ing recommendations that may prompt clinicians to doubt thevalidity of recommendations. This level of transparency may beespecially helpful early in the implementation process for aclinical decision support system. (Color version of figure isavailable online.)

rovided. Combination products available at a given site arelitate order entry by providing links that direct the clinician tomilies also help clinicians to more effectively deliver vaccines.

re pfaci

or fa

een recommended for children �60 months of age, but


iihhte

rsmmtptot(taiombaas

i

not among older children or adolescents. Although influ-enza vaccine was recommended for all high-risk chil-dren, we focused on those with asthma because theywere readily identified on the basis of ICD-9 codeswithin the local EHR. Given the success of vaccine alertsamong infants and toddlers, the prioritization of influenzavaccination in high-risk populations nationally, and rec-ognition of the need to vaccinate children with asthmaagainst influenza, we hypothesized that alerts wouldimprove rates and lead to increased protection. Still, werecognized that in a different clinical context (influenzavs early childhood vaccination), the impact of vaccinealerts might differ.

Ultimately, the study found that influenza alerts in thispopulation had only a minimal impact on vaccinationrates.69 Since influenza vaccination rates were known tovary across practices and seasons, we used a pre-postdesign in which the performance of each practice wascompared with the performance of the same practiceduring the prior year. The study involved a population ofnearly 12,000 children in the intervention year andincluded over 20,000 visits by children or adolescentswith asthma who were eligible to be vaccinated. Rates ofcaptured immunization opportunities increased from14.4% to 18.6% at intervention sites and from 12.7% to16.3% at control sites, a 0.6% greater improvement thatwas not statistically significant. Overall, influenza vacci-nation rates standardized for the characteristics of thepopulation improved only 3.4% more at interventionthan control sites, again a nonstatistically significant

FIG 12. Percent of immunization opportunities captured for rousupport for children under 2 years of age. A control period wasintervention was associated with significant increases in c(Reproduced with permission from Fiks et al.66 Copyright 2007s available online.)

difference. The change in rates for each of the 20


nvolved practices between the baseline (control) andntervention year is shown in Fig 13. Of note and againighlighting the importance of context, the system didave a significant but modest effect when the effect ofhe intervention among only the urban practices wasvaluated.A published commentary on the study considered the

esults in the context of best practices for decisionupport implementation.70 The authors of the com-entary argued that substantially improving outcomesay take more than reminders alone. In broad terms,

hey highlighted the importance of communicating theurpose of the intervention to stakeholders, ensuringhat the intervention is acceptable, ensuring that theptimal intervention technique is used, and monitoringhe progress of the intervention to ensure successFig 14). Of note, they emphasized the importance tohe success of a CDSS implementation of ensuring thecceptability of the intervention for patients and fam-lies. In the case of the influenza vaccine study, a lackf acceptance of the vaccine among families likelyinimized the impact of the intervention. Strategies to

etter engage families in decision-making and encour-ge them to accept recommended interventions areddressed later in our discussion of clinical decisionupport for SDM.

ADHD Decision Support

ADHD is the most common neurodevelopmental

childhood vaccines. Results of a trial of immunization decisionwed by education then activation of the decision support. The

red vaccination opportunities at both sick and well visits.the American Academy of Pediatrics.) (Color version of figure

tinefollo

aptuby

disorder in children and adolescents with prevalence

73

hohcaAbdm

nline

estimates between 3% and 16% depending on thesample and measurement techniques used.36,71 Ac-cording to the National Survey of Children’s Healthconducted by the Centers for Disease Control, nearly4.5 million US children between 4 and 17 years of agehad ever been diagnosed with the condition as of2003.72 Children with ADHD rate significantly lowerin health-related quality of life in all psychosocialareas. Those with comorbidities, such as oppositional

FIG 13. Difference between the study and baseline years in thPractices are divided into intervention and control sites. Each bteaching practices, all located in urban settings. (Reproduced wAcademy of Pediatrics.) (Color version of figure is available o

FIG 14. Questions to consider before the implementation of a remedical context. (Adapted from Sittig et al.70)

defiant conduct, internalizing, and learning disorders, A

74

ave even greater deficits.73 In addition to its impactn the daily lives of children, ADHD impacts theealth and functioning of families, schools, and theommunity. Primary care pediatricians have assumedgrowing role in the treatment and management ofDHD, in line with national guidelines disseminatedy the AAP. Nevertheless, training in managing chil-ren with ADHD has historically been limited andany clinicians remain uncomfortable with managing

rcentage of patients up-to-date for influenza vaccine, by site.presents a different site. Sites labeled with “T” are the residentermission from Fiks et al.69 Copyright 2009 by the American.)

e, point-of-care clinical decision support system in a particular

e pear reith p

al-tim

DHD.


tdAowAto

csAvsdcspb

udnstattficvwis

loc

While ADHD is an ideal prototype for studyingdecision support, especially in instances when SDMis the preferred model of decision-making and bothfamilies and clinicians use the system, publishedstudies to date have only evaluated ADHD decisionsupport focused on the clinician. In this context, thegoal was to improve clinician management ofADHD, not decision-making between the clinicianand family. A 2010 study in the journal Pediatricsreported results from a cluster randomized trial of asystem that included (1) on-screen clinician remind-ers to assess ADHD symptoms every 3-6 monthsand (2) an ADHD note template with structuredfields for symptoms, treatment effectiveness, andadverse medication effects.74 Using this approach,he intervention sought to improve (1) the quality ofocumentation of the assessments of children withDHD in the medical record and (2) the proportionf children with visits during a 6-month period inhich the status of child’s ADHD was assessed.s reflected in the choice of outcomes, the focus of

he study was on process measures, not patientutcomes.The study did show significant improvements in the

are delivered to children with ADHD at the studyites. During the study period, 70.9% of children withDHD at intervention sites had an ADHD-focusedisit as opposed to 53.9% at control sites. ADHDymptoms and treatment were documented more oftenuring well visits for patients in the intervention thanontrol group (78 vs 63%), but this difference was nottatistically significant. Of note, only 33% of eligiblehysicians ever used the ADHD template developedy the research team.The researchers combined their quantitative eval-ation with focus groups that addressed barriers toecision support adoption, an approach that is ofteneeded to better understand the reasons for theuccess or failure of an intervention. They foundhat clinicians often forgot the templates were avail-ble, preferred to use templates they created forhemselves, and found the use of documentationemplates to be cumbersome and inefficient. Thesendings underscore the importance of involvinglinicians in the design of a decision support inter-ention, of minimally interfering with existingorkflows, and of only trying to gather additional

nformation from clinicians that is essential to the
uccess of a CDSS intervention.

Asthma Decision Support

HIT-based, clinician-focused decision support hasalso been explored in asthma. Asthma is the mostcommon chronic physical condition in childhood andaffects the health of more than 6 million US children.National guidelines exist to inform care and guidedecision support.75 Nevertheless, despite these guide-ines, epidemiologists have highlighted the problemsf rising asthma prevalence, hospitalization, andost.76 Since office-based procedures for asthma as

well as asthma medications are ordered by clinicians,clinician-directed decision support is a reasonableapproach to improve adherence to asthma guidelinesand the quality of primary care for asthma. Thefollowing paragraphs describe 2 examples of pediatricclinician-focused asthma decision support.

In 1 study, handheld computers were used to deliverguideline-based decision support to office-based pedi-atricians.77 The intervention was delivered using aNewton Message Pad (Apple Computer, Cupertino,CA) and, similar to the ADHD study, provided astructured documentation template. In addition, thesystem dynamically generated recommendationsbased on 1994 guidelines from the AAP, providedassistance with the calculation of the predicted peakexpiratory flow rate and medication doses, and printedencounter summaries and prescriptions. As with theimmunization and ADHD interventions, clinicianswere free to accept or decline recommendations fromthe decision support system. However, unlike theADHD study, both process measures and patientoutcomes were considered. To prepare subjects, thestudy team provided on-site training to clinicians inthe intervention.

The study was conducted as a “before-after trial,”sometimes called a pre-post design. The researchersfound that the intervention resulted in more frequentmeasurements of peak expiratory flow and oxygensaturation, and a greater likelihood that inhaled ste-roids would be administered. However, cliniciansfound that the system too often recommended thatpatients receive supplemental oxygen and ignoredthose recommendations 90% of the time. Results alsodemonstrated that the length of visits increased withthe intervention and that the fees associated with visitsincreased from $103.11 to $145.61. Patient outcomes,recorded through follow-up phone calls, were assessed7 days following asthma office visits. No differences
were noted in the proportion of children requiring an
75

pataawm

cw

Emergency Department visit or hospitalization. Inaddition, no differences were observed in missedschool days or caretaker missed work associated withthe asthma exacerbations. Overall, then, adherence toprocess measures improved, but measured patientoutcomes were not affected during the assessmentperiod. The increased costs of care in the absence ofimproved patient outcomes illustrate the importance ofconsidering both intended and unintended conse-quences of CDSS interventions, an area consideredlater in this review.

In contrast to the handheld computer-based study, a2010 study of asthma decision support from our groupfocused on improving asthma care using EHR-baseddecision support guided by family-reported symp-toms.78 For asthma, detailed knowledge of dailysymptoms is needed to classify severity.75 Withknowledge of severity, EHR-based CDSS can then beprogrammed to provide recommendations for specificclasses of medication treatment, especially inhaled ste-roids. Multiple strategies have been used in differentsettings for gathering patient-entered data, including theuse of kiosks79 or hand-held computers,80 paper-basedforms that can be scanned into the EHR,81,82 patientortals, or personal health records. Often, the desiredpproach is guided by practical considerations, especiallyhe availability of these different systems. For our study,

12-practice cluster randomized trial, our team creatednd implemented an asthma control assessment toolithin the EHR based on a previously published instru-ent (Fig 15).83

Given the potential for an increase in demands onclinical staff to undermine a CDSS intervention, theresearch team used multiple approaches to promote theintegration of this tool into office workflows. First,practices were encouraged to have the tool completedin whatever way minimally disrupted care. In some,nurses completed the tool while in others clinicianspreferred to complete the instrument while taking thehistory. To prevent the duplication of effort, once theform was completed, a clinician could use a simpleshortcut to bring a summary of the information cap-tured using the asthma control tool into the visit note.Practices also received a small financial incentive forsurpassing a benchmark rate of completion of the tool.Finally, intervention practices received onscreen re-minders based on evidence-based guidelines, the in-formation in the asthma control tool, and a child’sprior history of medication use. With this approach,
rates of completion of the tool at both intervention and
76

ontrol practices approached 70% at visits by patientsith chronic asthma (Fig 16). As Fig 16 shows, 6

months were required to achieve high levels of adop-tion. This timeline should be considered by thoseimplementing decision support systems, especiallythose that require adaptations of existing workflowsand that address complex clinical issues or processes.

The goal of the intervention, which began with aneducational presentation to all involved sites, was to

FIG 15. Asthma control tool. The asthma control tool providesa standardized way to record the recent asthma history with aseries of discrete, evidence-based questions. (Color version offigure is available online.)

FIG 16. Asthma control tool adoption by clinicians at visits bypatients with chronic asthma. Intervention refers to practicesthat had the benefit of decision support. Control practicesreceived the tool to assist with documentation, but received nodecision support on the basis of using the tool. With a smalleconomic incentive, both intervention and control practicesadopted the tool and achieved rates of 70% for tool use ateligible visits. (Color version of figure is available online.)

increase the proportion of children with persistent


em

lfhb

asthma who had at least 1 prescription for a controllermedication, an up-to-date asthma care plan, and acompleted office-based spirometry to measure lungfunction. Once again indicative of the importance ofcontext for any decision support intervention, theimpact of the CDSS was different at urban practicesserving a largely Medicaid-insured population vs sub-urban practices that cared for a largely privatelyinsured group of families. Results showed that theintervention significantly increased the number ofprescriptions for controller medications over time by6% more at intervention vs control practices in theurban setting. At the suburban practices, baseline ratesof controller medication use were lower and theyincreased at both intervention and control practices. Inthe suburban setting, the filing of an up-to-date asthmacare plan improved 14% and spirometry improved 6%,both significant changes. Urban practices, with higherbaseline rates of asthma control tool completion, hadlittle room to improve.

Summary: Clinician-Focused Decision Support

The preceding paragraphs highlight successes andchallenges in projects targeting clinicians to improveguideline-based care. Our work in immunization dem-onstrates the importance of aligning decision supportwith the goals of both clinicians and families. Thesystem worked well for young children in the urbansetting whose parents readily accepted vaccines, butwas less effective in the setting of influenza, a vaccineless widely accepted by families. Studies in ADHDand asthma illustrate the potential of using CDSS toimprove the care of children with chronic illness.Using clinician-focused decision support, these studiesfocused primarily on process measures that are respon-sive to the actions of pediatricians in the office setting.These examples also illustrate the limitations of datacaptured as part of routine care as inputs into decisionsupport systems. The asthma control tool was createdso that recommendations could be based on assessedsymptoms which were not captured in a standardformat in the EHR. As a result, before the decisionsupport could be effectively implemented, clinical andfinancial incentives were used to promote adoption ofthe tool. These examples illustrate the complexity ofdelivering decision support to clinicians in primarycare pediatrics. An overlapping, but in many casesdistinct, set of challenges exists in engaging families athome or in the office in decision-making using family-
directed decision support. p

The Promise of Decision Support ThatIntegrates Families

This review thus far has focused primarily on decisionsupport directed at clinicians. However, while the deci-sion support systems we have previously reviewed ad-dress decisions reached or actions implemented primarilyby clinicians, supporting families is especially importantin the setting of informed and shared models of decision-making in which families take more responsibility forchoices regarding their health. The importance of usingHIT to help structure care around patient preferences inaddition to practice guidelines, a goal of SDM, has alsobeen emphasized.84 Even when the responsibility for thedecision is primarily the clinician’s, families still provideinput, including medical and personal information abouttheir specific child. The following discussion brieflysummarizes the potential benefits of SDM, reviewsbarriers to implementing CDSS for SDM, and exploresthe potential role of HIT in engaging families as moreactive participants in decisions related to their child’shealth.

Based on findings primarily from adult health care, theattention of researchers and policymakers, including theInstitute of Medicine and the World Health Organization,has increasingly focused on SDM, an approach to engagefamilies as active partners in their health care.85,86 Asmentioned earlier, SDM is supported by studies thatdemonstrate that improving provider-patient communi-cation is directly linked to satisfaction, adherence, andhealth outcomes.87 In addition, since sociocultural differ-nces between clinicians and patients may impair com-unication and decision-making if not addressed,88-91

explicitly discussing values in the context of medicaldecisions is also likely to improve care for minoritiesunderrepresented in the health professions. Of impor-tance, although there are ethical reasons to involvefamilies in a wide range of health decisions, SDM ismeant for conditions and circumstances in which fami-lies specifically must choose between multiple evidence-based choices.

Despite these potential benefits, many barriers exist toimplementing SDM in pediatric practice. Prior researchhas documented that families often play a passive role inpediatric encounters.92 In addition, although their fami-ies may benefit most from participation, we found thatamilies of US children with ADHD and asthma withigh levels of impairment in either their general orehavioral health were less likely than those of unim-
aired children to report a high level of participation in
77

icnmf

kioacteiemcc

c

lt

dsdfbawdapFsOesspfmr

baapah

SDM.93 Low health literacy and quantitative skills in thegeneral population, highlighted in Fig 1, also hinderSDM since facility in these areas is needed to under-standing concepts of risk, benefit, and outcomes. This isespecially concerning since research on adult literacyindicates that 90 million US adults have basic or below-basic literacy skills and that 110 million have basic orpoor quantitative skills.94 Participation in SDM may alsoncrease stress for some families already confronting thehallenges of caring for a child with special health careeeds. In that setting, emotional support in addition toedical information may be needed and desired to help

amilies effectively participate in decision-making.95 Fi-nally, regular access to their health providers for familiescaring for a child with a chronic illness may bolsterSDM. Specifically, we found that families of childrenwith ADHD and asthma who had difficulty reachingtheir clinician by phone were at least 26% less likely toreport a high level of SDM.93

What Families Need to Effectively Participatein Health Decisions

The most comprehensive discussion of the decisionsupport needs of families making child health decisionscomes from a 2008 systematic review that consideredresults from studies across such varied areas as prenatalgenetic screening, immunization, cancer care, and avariety of surgical decisions.96 The review identified 3ey needs: (1) information (including suggestions aboutts content, delivery, source, and timing); (2) talking tothers (including concerns about pressure from others);nd (3) feeling a sense of control over the process, whichould be influenced by emotionally charged decisions,he consultation process, and structural or service barri-rs. Of importance, information alone was generallynadequate to meet the needs of families who also neededmotional support and to take a role in how the decision-aking process unfolded. These needs transcended spe-

ific health conditions and were generally poorly met bylinicians.

Decision Aids: Tools to Support Families’Decision-Making Needs

While decision support for clinicians has increas-ingly been linked to EHRs, the evolution of decisionsupport for patients and families has centered on thedevelopment of decision aids. Decision aids include awide range of tools developed using a standard meth-
odology that help families reach informed choices t
78

onsistent with their values.30,96-98 They include pa-per-based forms, educational videos, and sometimesinteractive computer-based programs. Demonstratingthe validity of this approach, meta-analyses of random-ized trials of decision aids have shown that these toolsimprove the quality of health care decisions and reducethe use of options that patients do not value.30,99 Specif-ically, the use of patient decision aids results in greaterknowledge, lower uncertainty, increased participationin decision-making, and fewer people who remainundecided.30 Also, those completing decision aidswere less likely to elect invasive surgery. In general,more complex decision aids have more impact thansimpler ones. However, less is known regarding deci-sion aids for pediatrics since the vast majority of trialsof decision aids have been in the adult setting. In themost recent Cochrane review of decision aids, only 2included trials focused on pediatric decisions: 1 forcircumcision100 and 1 for infant vaccination.101 Theack of pediatric trials underscores the need for addi-ional research in this area.Although few randomized trials have been con-ucted of decision aids in pediatric healthcare, deci-ion aids are increasingly available for pediatric con-itions. Pediatric clinicians and families may benefitrom decision aids that address such topics as ADHD,irth control, depression, diabetes, enuresis, head-ches, smoking cessation, thyroid disease, tonsillitis,arts, and weight control.102 In addition, certainecision aids provide a format that can be adopted for

variety of health conditions. An example of aaper-based, family-focused decision aid, the Ottawaamily Decision Guide developed by Margaret Law-on and others at the Children’s Hospital of Easternntario, Canada, is shown in Appendix A. As this

xample shows, decision aids generally address thepecific needs of families participating in health deci-ions: information is provided, personal values im-acting the decision are elicited, supports or pressuresrom others that could potentially bias decision-aking are assessed, and an individual’s or family’s

eadiness to reach a decision is documented.Through the use of decision aids, many of thearriers to SDM in practice can be addressed. Decisionids often include information on risks and benefits infashion that can be readily understood, thus at least

artially mitigating the effect of low health literacynd numeracy on participation. For example, researchas demonstrated that frequencies (3 of 10 people on
his medication experience a side effect) are generally

sacttotndoroattlef

wrtcwf

sfcbbcp

asttt

pdmaradagie

tem

more meaningful than percentages (there is a 30%chance of a side effect),103 and decision aids generallypresent risks and benefits in terms of frequencies. Inaddition, decision aids provide tools to help familiesprepare for medical office visits, a strategy likely toincrease participation. In addition, decision aids mayhelp families identify needs for emotional supportsthat may or may not already be acknowledged. Onceaware of these needs, clinicians can help familiesconnect with counselors or family support groups whocan help them manage stress and move forward withthe decision-making process.

Health Information Technology and DecisionSupport for Shared Decision-Making

Despite their many benefits, decisions aids havecertain limitations that may potentially be overcomeby embedding them within a HIT-based decisionsupport system. Although decisionaids have been found to improvedecision-making, meta-analyseshave not demonstrated a consistentbenefit in terms of patient out-comes.30 One reason is that deci-ion aids are generally used whenny 1 of 2 or more options haveomparable outcomes. In that con-ext, the goal is often to improvehe decision-making, not healthutcomes.30 However, as the au-hors of the Cochrane review aptlyote,30 an important question is whether the use ofecision aids can help individuals or families achieveutcomes they most highly value. Given the passiveole of families in many encounters and work from ourwn group in ADHD that suggests that families’ goalsre often not assessed,27 HIT-based systems may helpo more systematically capture and track progressoward families’ goals, a process that may increase theikelihood that these goals are actually achieved. Forxample, preferences and goals may be elicited fromamilies through surveys as part of routine care.HIT-based systems may have additional benefits. Asith clinical practice guidelines, clinicians are often

eluctant to adopt decision aids and families some-imes access these tools through the Internet withoutlinician involvement. By implementing decision aidsithin a HIT-based framework that engages both the

Decisions aidlimitation

potentially bembeddingHIT-based de

sys

amily and the clinician, family access to these tools


hould be increased and, if the system is implementedollowing the principles described for clinician-fo-used decision support, clinician acceptance may alsoecome more widespread. With this approach, a HIT-ased system to promote SDM may support bothlinicians and families so that they are both betterrepared to participate.In addition, many decision aids have been developed

nd evaluated for 1-time decisions, such as circumci-ion. However, pediatric clinicians are increasinglyreating chronic conditions, such as ADHD or asthmahat involve revisiting treatment decisions overime.104 For families of children taking daily medica-

tions, decisions are revisited constantly. In this setting,keeping families engaged may be even more importantto treatment adherence and outcomes then with so-called “one and done” decisions. HIT provides a toolto deliver decision support to families at home as they

negotiate the care of chronically illchildren each day. Systems couldprovide recommendations to fam-ilies that reinforce the continuationof guideline-based care. Finally, inmany settings, it is difficult forfamilies to communicate with theirclinicians on a regular basis or forclinicians to know when familiesmodify the treatment plan on theirown.104 As a result, HIT may playan important role in alerting clini-cians when families are not adher-ing to the treatment or follow-up

lan. Overall then, HIT may provide a means toeliver evidence-based recommendations for treat-ent and follow-up to both clinicians and families to

chieve goals that are most salient for families. As aesult, the approach of using HIT to engage cliniciansnd families, share information, facilitate ongoingecision-making, and improve chronic disease man-gement has been heralded as a goal for the nexteneration of decision support.105 Despite this prom-se, little work has been done to date to use HIT tongage families in SDM.

Technical Strategies for Delivering DecisionSupport to Families Using Health InformationTechnology

The medical literature is increasingly filled with

ave certainhat mayvercome bym within aion support.

s hs te othecis

discussions of how different technologies may be

79

eIvutnIhm

saaIp

mcc

aapdm8aic

dadapus

DM

implemented to enhance health care.106 This focus iswarranted since many use HIT to play a more activerole in their health care. For example, the Pew Internetand American Life project found that 60% of adultsand 80% of Internet users have looked online forhealth information.107 These approaches are not meantto replace verbal communication, patient handouts,print media, or even audio or videotapes, but tocomplement their strengths. While the quality ofInternet-based information remains a concern, thisinformation does allow families to increasingly reachinformed choices.

To be most effective, the technology implementedmust be well-matched to the targetpopulation and situation. Fortu-nately, research from Pew hasshown that Internet use continuesto climb among all age groups andthat 93% of adolescents and youngadults are now online.107 How-ver, a lack of familiarity with thenternet does not necessarily pre-ent patients and families fromsing HIT in the health care set-ing. In adult diabetes, Internet-aive patients were willing to useT tools if they felt that it wouldelp enhance communication andanagement of their condition.108

For those with access to the In-ternet, e-mail is among the mostwidely studied and used of HITapproaches to support family-cli-nician interaction. A 2002 studyfound that parents were interestedin using e-mail to schedule visits,get information related to test re-sults, or discuss specific health concerns.109 In thattudy, clinicians were less enthusiastic than familiesbout e-mail use, citing worries regarding confidenti-lity.109 Gerstle and the AAP Task Force on Medicalnformatics then provided guidelines for the “appro-riate” use of e-mail in the office setting.110 They

prioritize the need for well-established policies that areunderstood by both clinical and nonclinical office staffand families to ensure confidentiality. In addition, datafrom a pediatric rheumatology clinic indicated thate-mail communication can save time compared withtelephone care and improved access. Despite these

The approacto engage

families, shafacilitate ong

making, achronic

managemheralded asnext generat

support.promise, li

been doneHIT to enga

S

positive findings, the potential impact of e-mail com-

80

unication on the decision-making process and out-omes for children with chronic illness remains in-ompletely understood.PHRs, defined as computerized applications to storeperson’s health information, have similarly gener-

ted considerable interest. They have the potential toool information from different settings to inform keyecision-makers and, in some cases, to help familiesanage chronic illness. Reflecting these strengths,

7% of respondents to a national survey felt that usingPHR would improve their ability to keep doctors

nformed of their health status and 81% felt that theseould help them get treatments tailored to their health

status.111 Still, only about 3% ofthe population is using personalhealth records. As with e-mail,concerns regarding privacy andconfidentiality were the most com-monly cited reasons individualswere not using PHRs.111

The American Medical Infor-matics Association’s College ofMedical Informatics has priori-tized using PHRs to promote pa-tient self-management of illnessesand the linkage of PHRs withEHRs to improve their impact.112

Patient portals are tools that inte-grate PHRs with EHRs, allowingfamilies to view portions of theirchild’s health record and commu-nicate with the medical office.112

Patient portals also provide a routefor families to provide informationnot routinely captured in EHRs,for example, the level of symp-toms a child is experiencing at

home or a family’s preferences for treatment. Portalshave been successfully implemented in both the sub-specialty and the primary care setting and have, incertain settings, been accompanied by tools to managechronic disease.113-115 However, as with all toolsesigned to provide families with health informationnd elicit information from them, careful systemesign is needed to ensure usability. A recent study inpediatric setting found that even previously tested

ortals present difficulties to users in navigating,nderstanding medical language, and recovering fromystem errors.116 Ongoing research to improve the

f using HITicians andinformation,g decision-improve

iseasehas beenoal for theof decisionpite thiswork hasate to usefamilies in.

h oclinreoinndd

enta gionDesttleto dge

usability of these systems is warranted.


uac

firgshgu

eaSshS

uAmh“c

Social media have also become an increasingly prom-inent forum for the exchange of information.107 Sites,including “Patients Like Me” (http://www.patientslikeme.com), help individuals or families connect to others insearch of support or information. Such tools havepotential benefits in both informed and shared modelsof decision-making. However, well-publicized secu-rity breaches and a lack of oversight to ensure thatinformation is evidence-based have raised concernsabout the use of social media as part of health caredelivery.117 For example, a 2009 study found limitedseful clinical information and substantial incorrectdvice on an asthma MySpace site targeting adoles-ents.118

Additional approaches for using HIT to foster theinvolvement of patients and families should also beconsidered. Automated phone systems, often usinginteractive voice response, have been used for sometime to remind families of upcoming visits and canprovide or gather health information. These toolsfacilitate large-scale communication between familiesand the medical home. An application of this technol-ogy to support SDM is discussed below. From withinmedical offices, paper-based questionnaires may beadministered to families and scanned into the EHR asa starting point to SDM. Such systems have provenable to accurately provide information to clinicians81

and have been used to tailor care to families based onrisk factors identified through their responses.82 Of-

ce-based kiosks may also be particularly helpful foreaching families with tailored health information or inathering information on preferences, goals, or healthtatus from families as they seek care. Such systemsave been implemented in emergency room settings toather data on symptoms, medication use, and thenmet needs of children with asthma.119 Kiosks de-

signed to promote child health have also been success-fully placed in community settings, such as McDon-ald’s, and have prompted families to discuss healthconcerns in more detail with their clinicians.120

An Electronic Health Record-Linked TelephoneSystem to Support Shared Decision-Makingfor Asthma

Although the potential impact of CDSS on health inmany contexts may be greater with systems thatsupport SDM as opposed to targeting clinicians orfamilies alone, far less research has explored this area.
In addition, the systems that have been developed and

valuated in pediatrics have often targeted individualspects of SDM without addressing the entire process.till, although no decision aid was implemented, theystem highlighted in the following paragraphs revealsow HIT-based tools may be developed to supportDM in pediatric primary care.Asthma has been the focus of work in pediatrics tose HIT to integrate families into decision-making.mong systems that have been developed, the 1 thatost directly addresses SDM between families at

ome and clinicians in the primary care office isTLC-Asthma,” a system designed by Adams andolleagues at Boston University Medical Center.121

TLC stands for telephone-linked communications andthis system uses completely automated telephonecalls, an example of interactive voice response, tomonitor, educate, and counsel families with chronichealth conditions. The system communicates withchildren or their parents using digitized human speechand they respond using the telephone keypad or byspeaking into the receiver. Of note, calls into thesystem may be initiated by the clinician or family.Results of the calls are stored and then used to providedecision support to clinicians. The system was de-signed to maintain weekly contact with families andusability testing demonstrated its feasibility.122

Using this approach, the TLC system provides com-prehensive and ongoing symptom monitoring, asthmaeducation, and knowledge assessment and reinforcesself-management.121 All of these features have the po-tential to enhance decision-making. Specifically, themonitoring components include assessment of the corre-spondence between treatment and symptom severity,peak flow measures, and knowledge of and adherence tothe prescribed medication regimen. Educational featuresconsist of information related to symptom recognition,especially regarding flares, identification and mitiga-tion of potential asthma triggers, the management ofexacerbations, medication use, and appropriate use ofthe health system for routine and urgent care. Knowl-edge of asthma is tracked through a series of yes/noquestions asked each week. Based on responses, ad-ditional educational material is provided. The systemalso summarizes important points, provides positivereinforcement and reassurance, and reminds the child/family to call next week. TLC-Asthma is designed forchild participants between 5 and 16 years of age andtheir parents.

While focused on the child and family and support-
ing self-management in the home, the system also
81

http://www.patientslikeme.com

http://www.patientslikeme.com

rbtlscepu

lfiwis

pipcaEc

iitststasaizagoi

mfthttHmetDmCdbspbun

alerts clinicians to potential problems that may requiremedical intervention.121 For example, if a child reportseduced peak flow readings that are unresponsive toeta-agonist treatment, the system sends a messagehat prompts an immediate call to the clinician. Foress emergent issues, such as medication refills, theystem sends a message to a nurse who acts as a careoordinator and addresses the concern. Of importance,arly data suggest that the system is feasible to use inractice and that alerts derived from the system can besed to guide targeted follow-up.121 Results from the

full evaluation of the system, including system use,changes in adherence to asthma therapy, and patient-centered outcomes have not yet been published. Asthis example illustrates, additional research is neededin asthma as well as other health conditions to deter-mine how HIT may most effectively improve SDM,adherence, and outcomes. Of note, the TLC system isalso being evaluated in the management of obesity.

Unintended Consequences ofClinician-Focused Decision Supportin Pediatrics

As we have highlighted in multiple areas in thisarticle, the implementation of clinician-focused deci-sion support has sometimes been associated withunintended negative consequences. In the outpatientsetting, our group found that decision support toincrease vaccination at sick visits contributed to de-creased rates of subsequent preventive care.65 Fami-ies who received vaccines were less likely to returnor well visits. In the pediatric intensive care unit, themplementation of a CPOE system was associatedith increased mortality at 1 site.11 However, mortal-

ty was decreased across all inpatient settings at aecond pediatric academic medical center.123 As these

examples illustrate, in both the inpatient and theoutpatient settings, those implementing CDSS willbenefit from carefully considering the many differentbehavior patterns that a new system impacts.124 As theexample of immunization at sick visits illustrates forpediatric outpatients, impacts on the behavior of thefamily, not simply the clinician, must be weighed.

ConclusionsAs payers and credentialing boards increasingly require

clinicians and health systems to demonstrate that they

82

rovide high-quality and safe care, practitioners willncreasingly rely on tools to support evidence-basedractice. In this context, the historically prolonged pro-ess of translating research into practice will need toccelerate. Given the $19bn investment in promotingHRs included in the ARRA, HIT is likely to have aentral role in this process.This review was designed to provide an understand-

ng of the potential benefit of HIT to decision-makingn pediatrics, especially in the primary care setting. Forhis process to be successful, the implementation teamhould consider how to apply relevant guidelines tohe question of interest and whether existing data areufficient to inform computer-based recommenda-ions. Teamwork between clinicians, administrators,nd IT professionals who can assess workflows, de-ign usable systems, and ensure that systems areligned with broader goals will promote success. Evenn relatively simple clinical contexts such as immuni-ation, the team should be prepared to both proactivelyddress many concerns that will arise from investedroups as well as promptly confront problems internalr external to the system that develop followingmplementation.To maximize the benefits of CDSS, clinicians, ad-inistrators, and IT professionals will also benefit

rom considering these tools within the broader con-ext of the literature on medical decision-making andealth communication. Delivering the right informa-ion to clinicians in the correct format and at the rightime and place is sufficient in certain clinical contexts.owever, particularly in the outpatient setting, treat-ent success often depends on engaging the family to

nsure that a prescribed treatment is acceptable andhat families will adhere to the recommendation.ecision aids have proven to be effective in improvingany aspects of decision-making. Using HIT-basedDSS to support families as well as clinicians in theecision-making process, we will likely be able touild on the successes of both clinician-focused deci-ion support and decision aids, 2 sets of tools toromote quality health care that have historically noteen integrated. Scholarship is needed to help usnderstand how best to realize this goal by testing aew set of approaches for improving outcomes.

AcknowledgmentWe thank Cayce C. Hughes, MPH for his help in

manuscript preparation.


Appendix A. Ottawa Family Decision Guide

Ottawa Family Decision Guide For Families Facing Tough Health or Social Decisions

You will be guided through four steps:

Clarify the decision. What decision do you face?

What is your reason for making this decision?

When do you need to make a choice?

How far along are you with your decision?

Are you leaning toward a specific option?

Explore your decision.

Knowledge Below, list the options and main benefits and harms

you already know. Underline the benefits and harms that you think are most likely to happen.

Values Use stars ( ) to show how much each

benefit and harm matters to you. 5 stars means that it matters "a lot". No stars means "not at all".

Certainty Circle the option with the benefits

that matter most to you and are most likely to happen.

BENEFITSReasons to choose this

option

How much it mattersAdd s!

HARMSReasons to avoid this

option

How much it matters Add s!

Option #1

Option #2

Option #3

Which option do you prefer? #1 #2 #3 unsure

Support

What role do you prefer in making this choice?

I prefer to share the decision with

I prefer to decide myself after hearing the views of

I prefer that someone else decides. Who? Who else is involved? (name and relationship)

Which option does this person prefer? How can this person help? #1 #2 #3 unsure Child

Curr Probl Pediatr Adolesc Health Care, March 2011 83

Identify your decision making needs.

Knowledge Do you know the benefits and harms of each option? Yes No

Values Are you clear about which benefits and harms matter most to you? Yes No

Support Do you have enough support and advice from others to make a choice? Yes No

Certainty Do you feel sure about the best choice? Yes No

People who answer “No” to one or more of these questions have decision making needs. They are more likely to delay their decision, change their mind, feel regret about their choice or blame others for bad outcomes.

Plan the next steps based on your needs.

Decision making needs Things you would like to try…

KnowledgeIf you feel you do NOT have enough facts

Find out more about the options and the chances of benefits and harms.

List your questions and note where to find the answers (e.g. library, health professionals, counsellors):

ValuesIf you are NOT sure which benefits and harms matter most to you

Review the stars in the balance scale to see what matters most to you.

Find people who know what it's like to experience the benefits and harms.

Talk to others who have made the decision.

Read stories of what mattered most to others.

Discuss with others what matters most to you.

SupportIf you feel you do NOT have enough support

Discuss your options with a trusted person (e.g. health professional, family, friends).

Find help to support your choice (e.g. funds, transport, child care).

If you feel PRESSURE from others to make a specific choice

Focus on the opinions of others who matter most.

Share your guide with others.

Ask another person involved to complete this guide. Find areas of agreement. When you disagree on facts, agree to get more information. When you disagree on what matters most, respect the person’s opinion. Take turns to listen to what the other person says matters most to them.

Find a neutral person to help you and others involved in the decision.

Other factors making the decision DIFFICULT

List anything else you need:

CHEO Decision Services © 2009 Lawson, Saarimaki, Kryworuchko, Barnes, Children’s Hospital of Eastern Ontario, Canada.
Web site: www.cheo.on.ca/en/DecisionServices Adapted from Ottawa Personal Decision Guide © 2006 O’Connor, Jacobsen, Stacey, University of Ottawa, Ottawa Hospital Research Institute, Canada.
References

1. Kohn LT, Corrigan J, Donaldson MS. To err is human:building a safer health system. Washington, DC: NationalAcademies Press; 2000.

2. Institute of Medicine (US). Committee on Quality of HealthCare in America. Crossing the Quality Chasm: a new healthsystem for the 21st century. Washington, DC: NationalAcademies Press; 2001.

3. H.R.1. 111th Congress: American Recovery and Reinvest-
ment Act of 2009. In: http://GovTrack.us (database of federal
84

legislation). 2010 from: http://www.govtrack.us/congress/bill.xpd?bill�h111-1.

4. Blumenthal D. Stimulating the adoption of health informa-tion technology. N Engl J Med 2009;360:1477-9.

5. DesRoches CM, Campbell EG, Rao SR, Donelan K, FerrisTG, Jha A, et al. Electronic health records in ambulatorycare: a national survey of physicians. N Engl J Med2008;359:50-60.

6. Hsiao C-J, Beatty PC, Hing ES, Woodwell DA, RechtsteinerE, Sisk J. Electronic medical records/electronic health record
use by office-based physicians: United States, 2008 and

http://GovTrack.us

http://www.govtrack.us/congress/bill.xpd?bill=h111-1

http://www.govtrack.us/congress/bill.xpd?bill=h111-1

preliminary 2009. In: Health E-Stat. Atlanta (GA): NationalCenter for Health Statistics; 2009.

7. Osheroff JA, Pifer EA, Teich JM, Sittig DF, Jenders RA.Improving outcomes with clinical decision support: animplementer’s guide. Chicago (IL): Healthcare Informationand Management Systems Society; 2005.

8. Shojania KG, Jennings A, Mayhew A, Ramsay CR, EcclesMP, Grimshaw J. The effects of on-screen, point of carecomputer reminders on processes and outcomes of care.Cochrane Database Syst Rev 2009;3:CD001096.

9. Bates DW, Kuperman GJ, Wang S, Gandhi T, Kitter A, VolkL, et al. Ten commandments for effective clinical decisionsupport: making the practice of evidence-based medicine areality. J Am Med Inform Assoc 2003;10:523-30.

10. Kawamoto K, Houlihan CA, Balas EA, Lobach DF. Improv-ing clinical practice using clinical decision support systems:a systematic review of trials to identify features critical tosuccess. BMJ 2005;330:765.

11. Han YY, Carcillo JA, Venkataraman ST, Clark RS, WatsonRS, Nguyen TC, et al. Unexpected increased mortality afterimplementation of a commercially sold computerized physi-cian order entry system. Pediatrics 2005;116:1506-12.

12. Charles C, Gafni A, Whelan T. Shared decision-making inthe medical encounter: what does it mean? (or it takes at leasttwo to tango). Soc Sci Med 1997;44:681-92.

13. Charles C, Gafni A, Whelan T. Decision-making in thephysician–patient encounter: revisiting the shared treatmentdecision-making model. Soc Sci Med 1999;49:651-61.

14. Revere D, Dunbar PJ. Review of computer-generated outpa-tient health behavior interventions: clinical encounters “inabsentia”. J Am Med Inform Assoc 2001;8:62-79.

15. Eysenbach G, Diepgen TL. The role of e-health and con-sumer health informatics for evidence-based patient choice inthe 21st century. Clin Dermatol 2001;19:11-7.

16. Kiesler DJ, Auerbach SM. Optimal matches of patientpreferences for information, decision-making and interper-sonal behavior: evidence, models and interventions. PatientEduc Couns 2006;61:319-41.

17. Miller VA, Nelson RM. A developmental approach to childassent for nontherapeutic research. J Pediatr 2006;149:S25-30.

18. Weithorn LA, Campbell SB. The competency of children andadolescents to make informed treatment decisions. Child Dev1982;53:1589-98.

19. Kohrman A, Clayton EW. Informed consent, parental per-mission, and assent in pediatric practice. Pediatrics 1995;95:314.

20. Bauchner H. Shared decision making in pediatrics. Arch DisChild 2001;84:246.

21. Denboba D, McPherson MG, Kenney MK, Strickland B,Newacheck PW. Achieving family and provider partnershipsfor children with special health care needs. Pediatrics2006;118:1607-15.

22. Levy SM, Herberman RB, Lee JK, Lippman ME, d’AngeloT. Breast conservation versus mastectomy: distress sequelaeas a function of choice. J Clin Oncol 1989;7:367-75.

23. Hack TF, Degner LF, Dyck DG. Relationship between
preferences for decisional control and illness information

among women with breast cancer: a quantitative andqualitative analysis. Soc Sci Med 1994;39:279-89.

24. Keselman A, Logan R, Smith CA, Leroy G, Zeng-Treitler Q.Developing informatics tools and strategies for consumer-centered health communication. J Am Med Inform Assoc2008;15:473-83.

25. Ubel PA. Is information always a good thing? Helpingpatients make “good” decisions. Med Care 2002:40:V39-44.

26. Offit PA, Moser CA. The problem with Dr Bob’s alternativevaccine schedule. Pediatrics 2009;123:e164-9.

27. Fiks AG, Gafen A, Guevara J, Hunter K, Tweedie A, PaxsonM, et al. Culturally effective shared decision making inADHD. In: Pediatric Academic Societies Abstracts. Balti-more (MD): Pediatric Academic Societies; 2009.

28. Merenstein D, Diener-West M, Krist A, Pinneger M, Cooper LA.An assessment of the shared-decision model in parents of childrenwith acute otitis media. Pediatrics 2005;116:1267-75.

29. Wissow LS, Gadomski A, Roter D, Larson S, Brown J,Zachary C, et al. Improving child and parent mental health inprimary care: a cluster-randomized trial of communicationskills training. Pediatrics 2008;121:266-75.

30. O’Connor AM, Bennett CL, Stacey D, Barry M, Col NF,Eden KB, et al. Decision aids for people facing healthtreatment or screening decisions. Cochrane Database SystRev 2009;3:CD001431.

31. Rogers EM. Diffusion of innovations. 5th Edition. NewYork: Free Press; 2003.

32. Lomas J, Sisk JE, Stocking B. From evidence to practice inthe United States, the United Kingdom, and Canada. MilbankQ 1993;71:405-10.

33. Boluyt N, Lincke CR, Offringa M. Quality of evidence-basedpediatric guidelines. Pediatrics 2005;115:1378-91.

34. American Academy of Pediatrics. Subcommittee on Man-agement of Acute Otitis Media. Diagnosis and managementof acute otitis media. Pediatrics 2004;113:1451-65.

35. American Academy of Family Physicians, American Acad-emy of Otolaryngology-Head and Neck Surgery and Amer-ican Academy of Pediatrics. Subcommittee on Otitis Mediawith effusion. Otitis media with effusion. Pediatrics 2004;113:1412-29.

36. American Academy of Pediatrics. Clinical practice guideline:diagnosis and evaluation of the child with attention-deficit/hyper-activity disorder. Pediatrics 2000;105:1158-70.

37. American Academy of Pediatrics. Subcommittee on Atten-tion-Deficit/Hyperactivity Disorder and Committee on Qual-ity Improvement. Clinical practice guideline: treatment of theschool-aged child with attention-deficit/hyperactivity disor-der. Pediatrics 2001;108:1033-44.

38. Steering Committee on Quality Improvement and Manage-ment, Subcommittee on febrile seizures. American Academyof Pediatrics. Febrile seizures: clinical practice guideline forthe long-term management of the child with simple febrileseizures. Pediatrics 2008;121:1281-6.

39. American Academy of Pediatrics. Clinical practiceguideline: diagnosis and management of childhood obstruc-tive sleep apnea syndrome. Pediatrics 2002;109:704-12.

40. Steering Committee on Quality Improvement and Manage-
ment, American Academy of Pediatrics. Classifying recom-
85

mendations for clinical practice guidelines. Pediatrics2004;114:874-7.

41. Cabana MD, Rand CS, Powe NR, Wu AW, Wilson MH,Abboud PA, et al. Why don’t physicians follow clinicalpractice guidelines? A framework for improvement. JAMA1999;282:1458-65.

42. Christakis DA, Rivara FP. Pediatricians’ awareness of andattitudes about four clinical practice guidelines. Pediatrics1998;101:825-30.

43. Flores G, Lee M, Bauchner H, Kastner B. Pediatricians’attitudes, beliefs, and practices regarding clinical practiceguidelines: a national survey. Pediatrics 2000;105:496-501.

44. Bergman DA. Evidence-based guidelines and critical path-ways for quality improvement. Pediatrics 1999;103:225-32.

45. Steering Committee on Quality Improvement and Manage-ment, American Academy of Pediatrics. Toward transparentclinical policies. Pediatrics 2008;121:643-6.

46. Garg AX, Adhikari NKJ, McDonald H, Rosas-Arellano MP,Devereaux PJ, Beyene J, et al. Effects of computerizedclinical decision support systems on practitioner performanceand patient outcomes: a systematic review. JAMA 2005;293:1223-38.

47. Biggs L, Grundmeier RW, Luberti AA. Effectiveness of aclinical/billing alert system vs education alone in assistingphysicians to correctly bill for asthma nebulization treat-ments in the pediatric office setting. AMIA Annu Symp Proc2007;11:872.

48. Shortliffe EH, Cimino JJ. Biomedical informatics: computerapplications in health care and biomedicine. 3rd edition. NewYork: Springer; 2006.

49. Spooner SA. Special requirements of electronic health recordsystems in pediatrics. Pediatrics 2007;119:631-7.

50. Weed LL. The problem oriented record as a basic tool inmedical education, patient care and clinical research. AnnClin Res 1971;3:131-4.

51. Armstrong-Coben A. The computer will see you now. NYTimes March 6, 2009. Sect. A27.

52. Meystre SM, Savova GK, Kipper-Schuler KC, Hurdle JF.Extracting information from textual documents in the elec-tronic health record: a review of recent research. Yearb MedInform 2008;1:128-44.

53. International Classification of Diseases. Ninth Revision, Clini-cal Modification (ICD-9-CM). National Center for HealthStatistics (National Center for Health Statistics), 2007 October20, 2010 from: ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Publications/ICD9-CM/2007/Readme08.txt.

54. Karras BT, Nath SD, Shiffman RN. A preliminary evaluationof guideline content mark-up using GEM—an XML guide-line elements model. Proc AMIA Symp 2000:413-7.

55. Shiffman RN, Karras BT, Agrawal A, Chen R, Marenco L,Nath S. GEM: a proposal for a more comprehensive guide-line document model using XML. J Am Med Inform Assoc2000;7:488-98.

56. Bates DW, Leape LL, Cullen DJ, Laird N, Petersen LA,Teich JM, et al. Effect of computerized physician order entryand a team intervention on prevention of serious medicationerrors. JAMA 1998;280:1311-6.

57. Kaushal R, Shojania KG, Bates DW. Effects of computerized
physician order entry and clinical decision support systems
86

on medication safety: a systematic review. Arch Intern Med2003;163:1409-16.

58. Gandhi TK, Weingart SN, Seger AC, Borus J, Burdick E,Poon EG, et al. Outpatient prescribing errors and the impactof computerized prescribing. J Gen Intern Med 2005;20:837-41.

59. Durieux P, Trinquart L, Colombet I, Nies J, Walton R,Rajeswaran A, et al. Computerized advice on drug dosage toimprove prescribing practice. Cochrane Database Syst Rev2008;3:CD002894.

60. Kaushal R, Bates DW, Landrigan C, McKenna KJ, ClappMD, Federico F, et al. Medication errors and adverse drugevents in pediatric inpatients. JAMA 2001;285:2114-20.

61. Walsh KE, Landrigan CP, Adams WG, Vinci RJ, ChessareJB, Cooper MR, et al. Effect of computer order entry onprevention of serious medication errors in hospitalized chil-dren. Pediatrics 2008;121:e421-7.

62. Yasuda K. Immunization information systems. Pediatrics2006;118:1293-5.

63. Recommended Immunization Schedules for Persons Aged 0through 18 years-United States, 2010. MMWR Morb MortalWkly Rep 2010;58:1-4.

64. Department of Health and Human Services. Healthy people2010. In: Conference ed. Washington, DC: US Departmentof Health and Human Services; 2000.

65. Fiks AG, Hunter KF, Localio AR, Grundmeier RW, Ales-sandrini EA. Impact of immunization at sick visits onwell-child care. Pediatrics 2008;121:898-905.

66. Fiks AG, Grundmeier RW, Biggs LM, Localio AR, Alessan-drini EA. Impact of clinical alerts within an electronic healthrecord on routine childhood immunization in an urbanpediatric population. Pediatrics 2007;120:707-14.

67. Prevention and Control of Influenza: Recommendations ofthe Advisory Committee on Immunization Practices (ACIP).Atlanta, GA: US Department of Health and Human Services,Centers for Disease Control and Prevention (CDC), 2006.

68. Committee on Infectious Diseases, American Academy ofPediatrics. Prevention of influenza: recommendations forinfluenza immunization of children, 2006-07. Pediatrics2007;119:846-51.

69. Fiks AG, Hunter KF, Localio AR, Grundmeier RW, Bryant-Stephens T, Luberti A, et al. Impact of electronic healthrecord-based alerts on influenza vaccination for children withasthma. Pediatrics 2009;124:159-69.

70. Sittig DF, Teich JM, Osheroff JA, Singh H. Improvingclinical quality indicators through electronic health records:it takes more than just a reminder. Pediatrics 2009;124:375-7.

71. Green M, Wong M, Atkins D. Diagnosis of Attention-Deficit/Hyperactivity Disorder: Technical Review. 3. Rock-ville (MD): Department of Health and Human Services;1999. Agency for Health Care Policy and Research publica-tion 99-0050.

72. Mental health in the United States: prevalence of diagnosisand medication treatment for attention-deficit/hyperactivitydisorder—United States, 2003. Morb Mortal Wkly Rep2005;54:842-7.

73. Klassen AF, Miller A, Fine S. Health-related quality of
life in children and adolescents who have a diagnosis of

ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Publications/ICD9-CM/2007/Readme08.txt

ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/Publications/ICD9-CM/2007/Readme08.txt

attention-deficit/hyperactivity disorder. Pediatrics 2004;114:e541-7.

74. Co JP, Johnson SA, Poon EG, Fiskio J, Rao SR, Cleave JV,et al. Electronic health record decision support and quality ofcare for children with ADHD. Pediatrics 2010;126:239-46.

75. National Institutes of Health. National Lung and BloodInstitute, National Asthma Education and Prevention Pro-gram. Expert panel report 3: Guidelines for the diagnosis andmanagement of asthma. NIH Publication 085846. Bethesda(MD): National Health, Lung, and Blood Institute; 2007.

76. Moorman JE, Rudd RA, Johnson CA, King M, Minor P,Bailey C, et al. National surveillance for asthma—UnitedStates, 1980-2004. MMWR Surveill Summ 2007;56:1-54.

77. Shiffman RN, Freudigman M, Brandt CA, Liaw Y, NavedoDD. A guideline implementation system using handheldcomputers for office management of asthma: effects onadherence and patient outcomes. Pediatrics 2000;105:767-73.

78. Bell LM, Grundmeier R, Localio R, Zorc J, Fiks AG, ZhangX, et al. Electronic health record-based decision support toimprove asthma care: a cluster-randomized trial. Pediatrics2010;125:e770-7.

79. Porter SC, Cai Z, Gribbons W, Goldmann DA, Kohane IS.The asthma kiosk: a patient-centered technology for collab-orative decision support in the emergency department. J AmMed Inform Assoc 2004;11:458-67.

80. Olson AL, Gaffney CA, Hedberg VA, Gladstone W, DuganS, Mathes R, et al. The healthy teen project: tools to enhanceadolescent health counseling. Ann Fam Med 2005;3(suppl 2):S63-5.

81. Biondich PG, Anand V, Downs SM, McDonald CJ. Usingadaptive turnaround documents to electronically acquirestructured data in clinical settings. AMIA Annu Symp Proc2003:86-90.

82. Biondich PG, Downs SM, Anand V, Carroll AE. Automatingthe recognition and prioritization of needed preventiveservices: early results from the CHICA system. AMIA AnnuSymp Proc 2005:51-5.

83. Zorc JJ, Pawlowski NA, Allen JL, Bryant-Stephens T,Winston M, Angsuco C, et al. Development and validation ofan instrument to measure asthma symptom control in chil-dren. J Asthma 2006;43:753-8.

84. Brennan PF, Strombom I. Improving health care by under-standing patient preferences: the role of computer technol-ogy. J Am Med Inform Assoc 1998;5:257-62.

85. Institute of Medicine (US). Committee on ComparativeEffectiveness Research Prioritization. Initial National Prior-ities for Comparative Effectiveness Research. Washington,DC: National Academies Press; 2009.

86. O’Connor A, Stacey D. Should patient decision aids (PtDAs)be introduced in the health care system? Copenhagen: WHORegional Office for Europe; 2005.

87. Stewart M, Brown JB, Boon H, Galajda J, Meredith L,Sangster M. Evidence on patient–doctor communication.Cancer Prev Control 1999;3:25-30.

88. Eisenberg JM. Sociologic influences on decision-making byclinicians. Ann Intern Med 1979;90:957-64.

89. Cho J, Solis B. Healthy Families Culture and Linguistics
Resources Survey: a physician perspective on their diverse

member population. Los Angeles (CA): LA Care HealthPlan; 2001.

90. Stevens GD, Shi L, Cooper LA. Patient-provider racial andethnic concordance and parent reports of the primary careexperiences of children. Ann Fam Med 2003;1:105-12.

91. Smedley BD, Stith AY, Nelson AR, Institute of Medicine.(U.S.). In: Committee on Understanding and EliminatingRacial and Ethnic Disparities in Health Care, editor. UnequalTreatment: Confronting Racial and Ethnic Disparities inHealth Care. Washington, DC: National Academies Press;2003.

92. Cox ED, Smith MA, Brown RL. Evaluating deliberation inpediatric primary care. Pediatrics 2007;120:e68-77.

93. Fiks AG, Localio AR, Alessandrini EA, Asch DA, GuevaraJP. Shared decision-making in pediatrics: a national perspec-tive. Pediatrics 2010;126:306-14.

94. Kutner M, Greenberg E, Jin Y, Boyle B, Hsu Y. Literacylevels in everyday life: results from the 2003 nationalassessment of adult literacy. In: United States Department ofEducation. Washington, DC: National Center for EducationStatistics; 2007.

95. Fiks AG, Gafen A, Hughes CC, Hunter KF, Barg FK. Usingfreelisting to understand shared decision making in ADHD:parents’ and pediatricians’ perspectives. Patient Educ Couns(in press).

96. Jackson C, Cheater FM, Reid I. A systematic review ofdecision support needs of parents making child health deci-sions. Health Expect 2008;11:232-51.

97. Elwyn G, O’Connor A, Stacey D, Volk R, Edwards A,Coulter A, et al. Developing a quality criteria framework forpatient decision aids: online international Delphi consensusprocess. BMJ 2006;333:417.

98. O’Connor AM, Stacey D, Barry MJ, Col NF, Eden KB,Entwistle V, et al. Do patient decision aids meet effective-ness criteria of the International Patient Decision Aid Stan-dards Collaboration? A systematic review and meta-analysis.Med Decis Mak 2007;27:554-74.

99. O’Connor AM, Wennberg JE, Legare F, Llewellyn-ThomasHA, Moulton BW, Sepucha KR, et al. Toward the “tippingpoint”: decision aids and informed patient choice. Health AffMillwood 2007;26:716-25.

100. Herrera AJ, Cochran B, Herrera A, Wallace B. Parentalinformation and circumcision in highly motivated coupleswith higher education. Pediatrics 1983;71:233-4.

101. Dunn RA, Shenouda PE, Martin DR, Schultz AJ. Videotapeincreases parent knowledge about poliovirus vaccines andchoices of polio vaccination schedules. Pediatrics 1998;102:e26.

102. The Children’s Hospital of Eastern Ontario. Decision Services,August 2 2010. Available from: http://www.cheo.on.ca/en/decisionaids. Accessed August 2, 2010.

103. Gigerenzer G. Calculated risks: how to know when numbersdeceive you. New York: Simon & Schuster; 2002.

104. Brinkman WB, Sherman SN, Zmitrovich AR, Visscher MO,Crosby LE, Phelan KJ, et al. Parental angst making and revisitingdecisions about treatment of attention-deficit/hyperactivity disor-der. Pediatrics 2009;124:580-9.

105. Grant RW. Next generation of health information tools:
where do we go from here? Mayo Clin Proc 2008;83:745-6.
87

http://www.cheo.on.ca/en/decisionaids

http://www.cheo.on.ca/en/decisionaids

1

1

1

1

1

1

1

106. D’Alessandro DM. Challenges and options for patient edu-cation in the office setting. Pediatr Ann 2010;39:78-83.

107. Fox S, Jones S. The social life of health information. ThePew Internet and Amerian Life Project 2009, October22 2010 from: http://pewinternet.org/Reports/2009/8-The-Social-Life-of-Health-Information.aspx.

108. Watson AJ, Bell AG, Kvedar JC, Grant RW. Reevaluatingthe digital divide: current lack of internet use is not a barrierto adoption of novel health information technology. DiabetesCare 2008;31:433-5.

109. Kleiner KD, Akers R, Burke BL, Werner EJ. Parent andphysician attitudes regarding electronic communication inpediatric practices. Pediatrics 2002;109:740-4.

110. Gerstle RS. E-mail communication between pediatriciansand their patients. Pediatrics 2004;114:317-21.

111. Westin AF. Connecting for Health: Americans Overwhelm-ingly Believe Electronic Personal Health Records CouldImprove Their Health. New York: Markle Foundation; 2008.

112. Tang PC, Ash JS, Bates DW, Overhage JM, Sands DZ.Personal health records: definitions, benefits, and strategiesfor overcoming barriers to adoption. J Am Med InformAssoc 2006;13:121-6.

113. Dorr D, Bonner LM, Cohen AN, Shoai RS, Perrin R, ChaneyE, et al. Informatics systems to promote improved care forchronic illness: a literature review. J Am Med Inform Assoc2007;14:156-63.

114. Earnest MA, Ross SE, Wittevrongel L, Moore LA, Lin CT.Use of a patient-accessible electronic medical record in apractice for congestive heart failure: patient and physicianexperiences. J Am Med Inform Assoc 2004;11:410-7.

115. Riegel B, Carlson B, Kopp Z, LePetri B, Glaser D, Unger A.Effect of a standardized nurse case-management telephoneintervention on resource use in patients with chronic heart
failure. Arch Intern Med 2002;162:705-12.
88

116. Britto MT, Jimison HB, Munafo JK, Wissman J, RogersML, Hersh W. Usability testing finds problems for noviceusers of pediatric portals. J Am Med Inform Assoc2009;16:660-9.

117. Du Bois S. Electronic Medical Records: Great, but Not VeryPrivate. October 2010 from: http://money.cnn.com/2010/10/06/technology/electronic_medical_records_safety.fortune/index.htm.

18. Versteeg KM, Knopf JM, Posluszny S, Vockell AL, BrittoMT. Teenagers wanting medical advice: is MySpace theanswer? Arch Pediatr Adolesc Med 2009;163:91-2.

19. Porter SC, Forbes P, Feldman HA, Goldmann DA. Impact ofpatient-centered decision support on quality of asthma care inthe emergency department. Pediatrics 2006;117:e33-42.

20. Thompson DA, Lozano P, Christakis DA. Parent use oftouchscreen computer kiosks for child health promotion incommunity settings. Pediatrics 2007;119:427-34.

21. Adams WG, Fuhlbrigge AL, Miller CW, Panek CG, Gi Y,Loane KC, et al. TLC-Asthma: an integrated informationsystem for patient-centered monitoring, case management,and point-of-care decision support. AMIA Annu Symp Proc2003:1-5.

22. Luo RQ, Adams WG, Fuhlbrigge AL, Friedman RH. Usagepatterns and clinical impact of an automated asthma man-agement system for children and their parents. AMIA AnnuSymp Proc 2005;1038.

23. Longhurst CA, Parast L, Sandborg CI, Widen E, Sullivan J,Hahn JS, et al. Decrease in hospital-wide mortality rate afterimplementation of a commercially sold computerized physi-cian order entry system. Pediatrics 2010;126:14-21.

24. Sittig DF, Ash JS, Zhang J, Osheroff JA, Shabot MM.Lessons from “Unexpected increased mortality after imple-mentation of a commercially sold computerized physician
order entry system.” Pediatrics 2006;118:797-801.

http://pewinternet.org/Reports/2009/8-The-Social-Life-of-Health-Information.aspx

http://pewinternet.org/Reports/2009/8-The-Social-Life-of-Health-Information.aspx

http://money.cnn.com/2010/10/06/technology/electronic_medical_records_safety.fortune/index.htm



designing computerized decision support that works for ... · designing computerized decision...

Documents