enhanced care management: improving health for high need ... · enhanced care management: improving...
TRANSCRIPT
EnhancedCareManagement:ImprovingHealthforHighNeed,HighRiskPatientsin
EstoniaEvaluationReportofthe2017EnhancedCareManagementPilotinEstonia
DraftVersion
WorldBankGroup
Tallinn,October25th,2017.
ContentsListofAcronyms .......................................................................................................................................... 1
Acknowledgements ..................................................................................................................................... 2
1. Introduction......................................................................................................................................... 3
2. Background–MakingtheCaseforECMinEstonia ............................................................................. 3
3. Preparation–IdentifyingBestPracticesinECM ................................................................................. 5
Risk-Stratification-TargetingPatientsAmenabletoCareManagement ................................................ 6
CareManagementPlans.......................................................................................................................... 9
ProactiveOutreachandTransitionsFollowUp ..................................................................................... 10
EnhancedCareManagementTeamApproachandResourceConnections .......................................... 11
4. Engagement–UtilizingaParticipatoryProcesstoDesignanECMPrograminEstonia .................... 11
InitialConsultationswithFamilyPhysicians-February2016................................................................ 12
1stFamilyPhysicianWorkshop-March2016 ........................................................................................ 12
2ndFamilyPhysicianWorkshop-November2016................................................................................. 17
5. Implementation–ConductingtheECMPilotinEstonia.................................................................... 21
6. Results–EvaluatingtheEstonianECMPilotExperience................................................................... 26
Feasibility ............................................................................................................................................... 26
Acceptability .......................................................................................................................................... 28
Process................................................................................................................................................... 28
Outcomes .............................................................................................................................................. 30
Facilitators(+)andBarriers(-) ............................................................................................................... 31
7. ConclusionsandPlanningforScale ................................................................................................... 32
Conclusions............................................................................................................................................ 32
AStrategyforScale................................................................................................................................ 32
PatientDashboardDevelopment .......................................................................................................... 33
ImprovedImplementationandUpdateofthePatientSelectionAlgorithm ......................................... 33
ScalableCoachingMethodology............................................................................................................ 33
TrainingProgramandMaterials ............................................................................................................ 34
LinkwithSocialCare .............................................................................................................................. 35
TowardPatient-level,Activity-basedCostingofECM ........................................................................... 35
Annex1:PerformanceMonitoringFramework......................................................................................... 38
Feasibility ............................................................................................................................................... 38
Acceptability .......................................................................................................................................... 38
Process................................................................................................................................................... 39
Outcomes .............................................................................................................................................. 40
FacilitatorsandBarriers......................................................................................................................... 40
Annex2:FrameworksofKeyInformantandPatientInterviews............................................................... 41
KeyInformantInterviews ...................................................................................................................... 41
PatientInterviews.................................................................................................................................. 42
Annex3:AdditionalTablesandFigures..................................................................................................... 43
TableA1:Listof50ChronicConditionsConsideredintheRisk-StratificationAlgorithm...................... 43
FigureA1:ConstructingtheConsolidatedPatientListsforEachPracticeList ...................................... 45
TableA2:OverviewofIndicatorsandTracersUsedin“TheStateofHealthCareIntegrationinEstonia”
............................................................................................................................................................... 45
References ................................................................................................................................................. 46
1
ListofAcronyms
ECM EnhancedCareManagementEHIF EstonianHealthInsuranceFundFPA FamilyPhysiciansAssociationEHR ElectronicHealthRecordMOSA MinistryofSocialAffairsQBS QualityBonusSchemeNCDs Non-CommunicableDiseasesSOPs StandardOperatingProceduresWBG WorldBankGroup
2
AcknowledgementsThepilotofenhancedcaremanagementinEstoniadescribedinthisreportwasplannedandcarriedoutby a team led by Christoph Kurowski (TTL). Team members included Elyssa Finkel, Kaija Kasekamp,MarvinPloetz,HannahRatcliffe(AriadneLabs),andAsafBitton(AriadneLabs).
Theteamexpresses itsgratitudetothefamilyphysicians(KerstiVeidrik,RuthKalda,DianaIngerainen,KatrinMartinson, Kerli Jaagosild, Tatjana Ruže, Anne Kaldoja, Helve Kansi, Elle Mall Keevallik, Marje
Oona,KatrinKurg,KatrinLätt)andnurses(AveAbermann,KairiÖöpik,AnneAger,NataljaIvanova,EveliJoost, Laura Keidong, Irina Võšegorodtseva, Terje Rannala, Tiina Tamberg, Sandra Soosaar, NataljaVabarna)thatjoinedtheworkinggroupandparticipatedintheenhancedcaremanagementpilot.
Furthermore,theteamwouldliketothankEllenKamenik,HelenaGrauberg,JeremyVeillard,WalterP.Wodchis,GeraintLewis,LaylaMcCay,EmilyWendell,LisaHirschhorn,andMichaelMusharbashfortheir
contributionsatdifferentstagesofthiswork.ThepilotworkevaluatedinthisreportwasguidedbyaNationalSteeringCommittee,whosemembers
includedTiiuAro(HealthBoard);DianaIngerainenandKatrinMartinson(FamilyPhysician’sAssociation);Ruth Kalda (Department of Family Medicine, University of Tartu); Marju Past (Society of DisabledPeople); TriinHabicht,Maris Jesse andAgris Koppel (Ministryof SocialAffairs);Natalia Eigo (National
Institute for Health Development); Urmas Sukles, Priit Tampere (Hospital Association); Gerli Liivet(NursingAssociation),AngelaEensalu-Lind(TallinnHealthCareCollege);andTanelRoss(EstonianHealthInsurance Fund). The findings, interpretations, and conclusions expressed in this report do not
necessarilyreflecttheviewsoftheSteeringCommittee.TheteamexpressesitsappreciationtothestaffoftheEstonianHealthInsuranceFund,inparticulartoTanelRoss,SiljaKimmeland JekaterinaDemidenko for their inputsandclosecollaborationduring thepreparationandimplementationofthepilot.TheteamalsogratefullyacknowledgestheexcellenteditorialsupportbyAmitChandra.TheEstoniantranslationofthisreportwasnotcreatedbyTheWorldBankandshouldnotbeconsideredan officialWorld Bank translation. TheWorld Bank shall not be liable for any content or error in this
translation.
3
1. IntroductionIn2014theWorldBankGroup(WBG)partneredwiththeEstonianHealthInsuranceFund(EHIF)to
evaluate health system challenges related to the coordination of health care across different caresettings. Data related to select tracer conditions was assessed in relation to the prevention andtreatmentof chronicdiseases,withparticular attention to the role and functioningof primaryhealth
care. Primary carewas selected as the focus of the project as the result of significant evidence thatcountrieswith strongprimary health systemsperformbetter in termsof improvedpopulationhealthand health expenditure growth mitigation (50). The results revealed care coordination gaps and a
potential for primary care system strengthening utilizing an “enhanced care management” (ECM)approach.ThisconclusionformedthebasisofasubsequentECMpilotproject,launchedin2015,whichisdescribedinthisreport.
“Caremanagement”aimsto improvehealthaccess,quality,andvaluethroughproactiveoutreachfortargetpatientpopulations.TheECMapproachreliesontheleadershipandenergyofprimaryhealthproviders.ECMhasthepotentialtoimproveintegrationofcareacrosscaresettings,enhanceindividual
patientoutcomes,andraisethevalueofhealthinterventionsprovidedatalllevelsofthehealthsystem.
This paper presents the pilot project on ECM in Estonia implemented between 2016 and 2017.Section2providesabriefoverviewofthehealthcaresystemofEstonia,presentsanin-depthanalysisof
theresultsofthe2014precursorstudy,andmakesthecaseforECMasapotentialsolutiontoEstonianhealthsectorchallenges.Section3reviewsECMexperiencesandbestpractices.Section4describesthe
participatory ECM program design process. Section 5 presents the experience of programimplementation. Section 6 enumerates the results of the pilot project basedon an analysis of healthinsurance claims, stakeholder interviews, a pre- and post-pilot provider survey and themonthly pilot
monitoringreportspreparedbythelocalcoordinator.Section7outlinesthenextstepsandplanningforscale.
2. Background–MakingtheCaseforECMinEstonia
EstonianHealthSystemOverviewThe Estonianhealth systemhas beenhailed for its track recordof cost efficiencywhile achieving
goodoutcomesatlevelssimilartoEUaverages.LifeexpectancyatbirthisclosetotheEUaverage(77.5
versus80.9years in2014)andunder-5childmortality rate is slightlybetter than theEUaverage (3.4versus4.4per1000livebirthsin2014)(53).Despitetheserelativeoutcomeindicators,Estoniaspendssignificantly lessthantheEUaverageonhealth(6.4percentversus10.1percentofGDPin2014).The
principal source of health financing is public, constituting approximately 76 percent of total healthexpendituresin2015andoperationalizedthroughEHIF,thesinglepayerentity(54).Thesecondlargestremaining source of financing is out-of-pocket expenditures, comprising approximately 23 percent of
total health expenditures. These are mostly concentrated on co-payments for medicines and dentalservices,suggestingthatthefinancingsystemiseffectiveinprotectinghouseholdsagainstcatastrophichealthexpenditures.
4
Estoniahas alreadymadegreat strides to addresshealth carequality and integration suchas theintroductionofaqualitybonusscheme(QBS)forproviders,ane-consultationsystem,andvariousother
quality assurance mechanisms. Nevertheless, the rise in non-communicable diseases (NCDs), carefragmentation,alongwiththerisingcostofmedicalproductsandtechnologiesarealsocurrentlydrivingincreases in national expenditures, while the shrinking working-age population has resulted in
decreasing revenues to finance health care. These threats to the social health insurance systemmotivatedthecountry’sinitialengagementwiththeWBGonthistopicin2014.
WorldBankAnalysis2014-2015
ThequantitativeandqualitativeanalysisperformedbytheWorldBankin2014-20151demonstratedthat the Estonian health care system faces considerable challenges with respect to healthcare
integration, especially with respect to the prevention and treatment of chronic diseases. Specificfindingsofthestudyincluded:
1. Alargeproportionofacuteinpatientcarecouldbeavoidedbyshiftingcaretomoreappropriate
primarycaresettings.2. Alargeshareofspecialistvisitscouldbeavoided.3. Lowcoverageofpreventiveservicesfordiabetesandhypertensionpatientswasnoted.
4. Patientsoftenbypassprimarycareanddirectlyaccessedspecialistcareeventhoughthesespecialistsaddedlittlevalueintermsofthecareofchronicconditions.
5. Coordinationchallengesacrosslevelsofcareexistbeforeandafteracuteinpatientcare
episodes,asevidencedbythesignificantshareofunnecessarypre-operativetestsandinadequatefollowupcarebyprimaryhealthproviders.
6. Manyoftheseoutcomeswerealsoshowntodiffersignificantlyacrossspecificpatient/populationgroups(e.g.avoidablehospitaladmissionswerehigheramongthepoor,men,ruralresidentsandpatientswithdepression).
Theseresultsrevealedanopportunitytoimprovethemanagementofspecificpatientsub-groupsat
the primary care level. Primary care strengthening could reduce avoidable hospital admissions andspecialist visits, increase the provision of preventive services, and improve the flow of information
betweenprimarycareprovidersandspecialists.Thiscaresystemstrengtheningapproachwouldrequiretheexpansionoftheroleoffamilyphysicianstoprovideasetofhealthmanagementinterventionstoasubsetofhigh-needpatients,hereafterreferredtoasenhancedcaremanagement(ECM).
EnhancedCareManagementtoStrengthenPrimaryCare
Primarycareisdefinedas“essentialhealthcarebasedonscientificallysoundandsociallyacceptable
methods,universallyaccessibletoindividualsandfamilieswiththeirfullparticipationatacostthatthecommunityandcountrycanaffordinaspiritofself-relianceandself-determination”(55).Primarycare
1DescribedinWorldBank,2015:TheStateofHealthCareIntegrationinEstonia.SummaryReport,WorldBank.PleaseseeAnnex3foranoverviewofindicatorsandtracersusedinthestudy.
5
is usually the first level of care, where patients make their first contact with a health provider, andwhere the majority of diagnosis and therapy occurs. Primary care was designated as the principal
mechanismtoachieve“healthforall”intheWHO’sAlma-Atadeclarationin1978.
The four functional pillars ofmodern primary care include: 1) first-contact access; 2) longitudinal
continuity over time; 3) comprehensiveness,with capacity to provide care for themajority of health
problems;and4)coordinationofcarewithotherpartsofthehealthcaresystem(2).Allfourpillarsaredifficult toachieve incomplex,high-needpatients,particularly insub-populationswhofacesignificantsocial barriers to care. Patients with complex socio-medical conditions typically make up a small
percentage of the overall patient population but account for a disproportionate burden of illness,utilization, and cost. These individuals often have multiple chronic conditions, face significantsocioeconomicchallenges,and/orhaveco-existingbehavioralhealthcomorbidities.Failuretoeffectively
managethecareofthesepatientscanleadtodeteriorationsintheirhealth,testduplication,medicationconflicts,andmedicalerrors(3,49).
Enhancedcaremanagement(ECM) isaneffectivetooltoachievecarecoordinationforhigh-need,
complexpatientsandaddresschallengesrelatedtoall4pillarsofprimarycare.ECMinvolves focusedandproactiveoutreachtoasmallnumberofpatients(typically5-10%ofadoctor’spatientlist)whoareathighriskofhealthstatusdeteriorationorincreasedutilization.TheexpressgoalofECMplatformsis
to target these complex and high-need individuals in order to improve their health and reduce theirneedforreactivemedicalservices(4,5).ECMmayinclude:followupduringcaretransitions(e.g.followuphospitaldischarges);trackingtestresultsandreferrals;ensuringthatquality-of-caretargetsaremet
(suchastheQBSinEstonia);ensuringmedicationreconciliationandadherence;andpatientmonitoringbetween scheduled visits. ECM can improve care coordination and patient outcomes and areincreasinglybeingimplementedacrosshealthcaredeliverysystemsworldwide.
3. Preparation–IdentifyingBestPracticesinECM
EnhancedCareManagement(ECM)Defined
Definition:“Asetofactivitiesdesignedtoassistpatientsandtheirsupportsystemsinmanagingmedicalconditionsmoreeffectively.”(41)
Objectives:
! Increasecarecoordinationforhigh-need,complexpatientsacrossdifferentprovidersandlevelsofthehealthsystem
! Addresspatientneedswithexplicitgoalsofimprovedpatienthealthandreducedneedformedicalservices(4,5)
! Improvepatientengagementbyelicitingpatienthealthgoals,promotingpatientself-management,andestablishingpatientcareplans
6
Inlate2015theWorldBankengagedinaconsultationprocesswithexpertsonrisk-stratificationandenhanced care management from Canada, the United Kingdom and the US in order to identify thepotential of carrying out a risk-stratification and care management pilot in Estonia as part of itscontinuedcooperationwiththeEHIF.Eventually,theWorldBankhiredateamofexpertsfromAriadneLabstohelpwiththeimplementationofsuchapilot.Fourkeyelementsofprogramswereidentifiedasaframeworkofanalysis:
1. Risk-stratificationtotargetpatientsmostlikelytobenefitfromcaremanagement.2. Caremanagementplans to proactively respond to changes in patient’s conditions, anticipate
specificfutureproblems,andpromotebetterself-managementofcurrentconditions.3. Proactiveoutreachandtransitionsfollowupwithallprofessionalsinvolvedinpatients’care.4. Team approach and resource connections with patients and their caregivers to
comprehensivelyassessandaddressmedicalandsocialneeds.
In addition to the key elements described above, a set of system characteristics or conditionsprecedenttoasuccessfulcaremanagementpracticewereconsidered.Theseincludeduniversalhealthcoveragewithapaymentenvironmentsupportiveofcaremanagementinterventions,motivatedmulti-disciplinary teams, and electronic health records (EHR) systems that allow for reporting on qualityoutcomes(seeFigure1).Figure1:KeyElementsofEnhancedCareManagement
Source:OwnElaboration.
Risk-Stratification-TargetingPatientsAmenabletoCareManagement
The first step to identifying high-risk patients for any given care management program is
definingthetypeof“risk”thatthecaremanagementprogramisseekingtomitigate.Risk-stratificationismost often conducted to identify those patients at risk of high utilization frequency, high utilization
7
costs,avoidablehospitalizations,and/oravoidablemorbidityormortality.Thesedifferentrisktypesarecorrelatedamongeachother,butthesub-typeofriskthat isofhighestpriority inaprogramwillvaryacross contexts and is dependent on target population characteristics, health system priorities, andbottle-necks within service delivery pathways. Rising health care costs have been an overarchingconcern inmanyhealthcaresystems(i.e. intheUnitedStates),andthemajorityofcaremanagementprogramshavebeenimplementedtoreduceexcessutilizationandcostexpenditure(6).However,caremanagementeffortsareincreasinglyfocussedonpromotingpatienthealthandwell-beingthroughthefacilitationof self-managementandengagementwith careproviders. This shift away fromshort-termcost-cutting efforts towards health promotion requires a corresponding shift in the risk-stratificationapproach.
Conventional patient selection tools, when narrowly focussed on high-utilization or high-costpatients,selectpatientswithawidespectrumofhealthissues—fromhigh-riskpregnancytosubstanceabusetosevereheartdisease—eachofwhichwouldrequireadifferentcaremanagementapproach.Asanalternative,itisusefultoconsiderbuildingarchetypesofpatientsthattheinterventionshouldtarget.These archetypes are determined by the category of risk that a caremanagement program seeks tomitigate.Patientarchetypes,arelativelynewconcept, involvethe incorporationofdesignthinking2todefine,group,andsolveaproblemfromtheperspectiveoftheuser,i.e.thepatient(8).Bydevelopingcommon classifications of patient groups that incorporate both medical diagnoses and psychosocialneeds,careprogramscanbeorientedtowardgoalsthatareachievableanddesirabletothesepatients.From the provider’s viewpoint, patient archetypes help simplify the design of care managementprograms.By creatingapatientdefinitionwithanassociatedneedsprofile (seeFigure2below), caremanagementprogramscanbedesignedwiththepatientarchetypeinmindratherthanconsideringthewholeuniverseofpotentialriskfactorsandscenarios.Figure2:ArchetypesofHighUsersbyPatternandTypeofNeed
Source: Vaillancourt, S. (2014), Using Archetypes to Design Services for High Users of Healthcare, Healthcarepapers.
2Designthinkingisaproblem-solvingapproachusedbydesignerstointegratecreativityandsocialconsiderationstoproductandprogramdesign.
8
Onceaprogramhasdefinedthetargetoutcomeandidentifiedthepatientarchetypestotarget,arisk-stratificationapproachisemployedtoidentifyaspecificcohortofpatientstoincludeinthecaremanagement program. The three primary approaches for risk-stratification are: 1) algorithm basedtools,primarilyrelyinguponhealthutilizationandclaimsdataorclinicaldataabstractedfromtheEHR,2) clinician referral, primarily relying on provider intuition, or 3) a hybrid approach, which utilizes acombinationofthequantitativeandqualitativeapproachesdescribed.Emergingevidencesuggeststhatthebestmethodforidentifyinghigh-riskpopulationsisahybridapproachthatcombinesanalgorithm-basedtoolandclinicianintuition(3).
Hybridapproachesaresequential,utilizingastheirfirststepanalgorithm-basedriskprediction
tool to analyze available clinical data (claims or EHR) in order to identify a subset of patients at thehighestrisk fortheoutcomeof interest,orpatientswhofit thearchetype.Therearemanyalgorithm-basedrisk-stratificationtools thathavebeenwellvalidatedfor identifyingasubsetofpatientsathighrisk forexperiencingspecificoutcomes.However, therearenodefinedstandardsforrisk-stratificationtoolsandmanyareproprietaryalgorithms(e.g.,theJohnsHopkinsACGsystem).Furthermore,nosingletoolstandsoutassuperiortotheothers.Algorithm-basedmodelsingeneraldonotexplainmorethanhalfoftheobservedvariabilityinpatientoutcomes.Inpart,thismaybeduetothefactthatalgorithm-based methods are limited in their ability to assess important psychosocial considerations that mayimpactbothapatient’sneedforandabilitytobenefitfromcaremanagement(9).Furthermore,patientsgroupedintohighest-riskcategorieshavelargevariationsincareintensityandmorbidityyearoveryear,thereby obscuring attempts to cluster individuals into simple categories of high spending or highutilization.Whicheveralgorithm-basedapproachisused,thereisaneedtocloselyalignitsparameterswith the planned caremanagement interventions, and to consider both the patient’s need and theiramenabilitytocaremanagementparticipation(seeFigure3).
Figure3:VennDiagramofRelevantPatientTypesforEnhancedCareManagement
9
FigureAdapted from:GerardAnderson&ClaudiaSalzberg (2016), IdentifyingHighNeedHigh
CostIndividuals,JohnsHopkinsUniversity.Toimprovethepredictivevalueofanalgorithm-basedmethod,thehybridapproachincludesa
subsequentsteputilizingclinicalintuitionandknowledgeofpatientcontextualfactorsoftennotfoundwithinhealthrecordsystemsorclaimsdata.Inthisstep,thelistofpatientsidentifiedbythealgorithm-based tool is reviewedby the responsibleprimarycareprovideror theECMteam.Using their clinicaljudgmentandpersonalknowledgeofpatientswithintheircommunity,theycanremovepatientsfromthe listwho—forclinical (i.e. terminaldiagnosis),social,orbehavioralreasons—areunlikelytobenefitfromthecaremanagementprogram.Additionally,practitionerscanaddpatients to the listwhowerenot initiallycapturedbythealgorithm-basedmethodbutwhohaveastrongpotential tobenefit frominvolvement in the caremanagementprogram.The roleof clinical intuition in thehybrid approach ispowerful, particularly in addressing psychosocial considerations not easily captured in clinicaldocumentation or billing data. Clinicians can leverage their personal relationships with patients toconsider–characteristicssuchasapatient’shealthliteracy,copingskills,physicalvulnerabilities,existinglinkagestoothercareproviders,andsocialcontextorhomeenvironment(10).
Apatient’ssocialandbehavioralcharacteristicscanprofoundlyaffecttherelevanceandutilityofcaremanagementprograms.Forexample,anelderlypatientwholivesaloneandisincreasinglyfrailmay benefitmore from a caremanagement program than an elderly patient with the same diseaseburdenwho liveswitha familymemberwho is involved in theirday-to-day care.Other social factorssuchaspoverty,homelessness,andunemploymentmayalso increasetheneedthatpatientshave forcare management programs (13). Behavioral health issues such as substance abuse, alcoholism, ormentalhealthdisordersoftenincreaseapatient’scarecomplexityinwaysthatincreasetheirpotentialtobenefitfromcaremanagementprograms(14).However, insomecasessocialandbehavioral issuescan also reduce the likelihood of a patient benefiting from a particular care management program.Severementalhealthdisordersorseveresubstanceabuse, forexample,mayrequiremorespecializedattentionthancanbeprovidedthroughaprimarycare-basedprogram.Whethersocialandbehavioralissues make it more or less likely for patients to benefit from care management programs dependsgreatlyonthetypesof resourcesavailable to theprogram.Forexample,multidisciplinary team-basedprogramsthatemploysocialworkersandmentalhealthexpertsmaybeabletobenefitpatientswhomsinglephysician-nurseteamsdonothavethecapacitytohandle(13,17).
CareManagementPlansOnce patients are selected for the ECM, they will each need to undergo a comprehensive
evaluationby the care team inorder to design an appropriate caremanagementplan anddevelop atrusting relationship. This comprehensive assessment is needed to determine an individual patient’sneedssituatedwithinanawarenessoftheirsocialandcontextualenvironments.Thisassessmentshouldconsider clinical history, gaps in care, barriers they face accessing and receiving care, behavioral andsocial needs, functional status, and baseline level of patient activation3 (5,20). Building trustingrelationshipswithhigh-riskpatientsisacornerstoneofsuccessfulcaremanagementprograms.Thebestwaycaremanagementteamscandothis isbyunderstandingapatient’scontextandaddressingtheirunmetsocialneeds.Ithasbeenestimatedthatupto70%ofthefactorsthatimpactapatient’sabilitytostayhealthyaresocialandenvironmental,whileonly10%aredirectlyrelatedtomedicalcare(31).
3Patientactivationreferstotheknowledge,skills,andconfidencethatapatientneedsinordertomanagetheirownhealthandhealthcare.
10
Care plans should build upon a comprehensive assessment of patient needs, values, and
preferences.Thedesignandcontentofany individualcareplanwillnecessarilydependupontherisksbeing targeted, theoutcomes theprogramhopes toachieve, thestaffing resourcesavailable,and theavailable modalities for care team-patient interactions (5). Successful care management programstypically promote patient and family engagement in self-care (20).Onemechanism throughwhich toachieve this outcome is to design “dual-facing care plans,” or care plans that are jointly designed byprovidersandpatientsor caregivers. Topromoteusebypatientsandproviders, careplans shouldbekeptsimpleandorganized inawaythatmaximizesclarityandrelevancetopatients’dailyexperience.Careplansshouldincludean“actionplan”forpatientswhichoutlineconcretestepsforthemtotaketomakeprogresstowardsthecareplangoals.Theyalsotypicallyincludepointsofcontactforthepatientandactionsthecareteamshouldtakeinresponsetocriticalevents(e.g.hospitalizations).Finally,careplansshouldbedesignedtomeetthetechnologicalandliteracylevelsandcapabilitiesoftheintendedusers.
There is a distinction between a care plan (which is static) and the activity of care planning(whichincludesacontinuousprocessofreassessmentoftheplanandrealignmentofcareprovided).Forcareplanstobedynamic,careteamsmustconsiderthefollowing,especiallyforpatientsnotachievingtheircareplangoals:
● Wheredoesourcurrentcarealignornotalignwithourpatient'scareplanandgoals?● Whyisourcareplannotworking?● Whatisthevaluethatweprovidetothispatient?● Whatisthesinglemostimportantthingthatneedstohappentopreventthispatientfrom
deterioratingortoaligntheircaretotheircareplan?● Whatspecificallyistheproblemwearetryingtosolve?Isitmedical,social,orboth?● Reflectingonanswerstothequestionsabove,howdoweplanourdailywork?
It is important to note that no “gold standard” exists for either risk-stratification or care
managementplanconstruction. Instead, thedesignofanycaremanagementprogramwillnecessarilybe dependent on the type of risk the program is trying to ameliorate and the target outcomes theprogram intends to change. Itmustbebuilt and continually refinedbasedon theexperienceofbothpatients and providers to ensure that it meets the goals of improved care, smarter spending, andimprovedhealthoutcomes.ProactiveOutreachandTransitionsFollowUp
A primary goal of a care management program is to coordinate the care and services thatpatients receive, both inside andoutsideof the clinical system.Doing so requires that the care teamestablishes strongworking relationshipswith hospitals, nursing facilities, and other clinical specialtiesnotincludedinthecareteam(5).Giventhisrequirement,coordinationshouldfactorintothedesignofcare management program to encourage harmonization across potential stakeholders (37).Coordinationwithclinicalprovidersenablesappropriate followupduringcaretransitions, tracking labtests and referrals, ensuring medication reconciliation and adherence, and proactive outreach andmonitoringbetweenscheduledvisits(38).Establishingregularschedulesofcontactbetweencareteamsandpatientsalsoenablesproactiveoutreachandmonitoring.Theseschedulescanvarybasedonpatientneedandcare teamcapacity, ranging fromdaily,weekly,ormonthly,and incorporating tools suchastext messages, phone calls, and health coach visits. Successful care management programs alsocoordinatewithsocialserviceproviderstoconnectpatientswithresourcesoutsideofthedirectpurview
11
of thehealth system thatmay impact apatient’s health status, suchashousing, foodaid, education,eldercare,andtransportation.
A critical component of coordinating care is monitoring changes in patient health status toensuresafecaretransitionsacrosslevelsofthehealthsystem.Doingsorequiresstrongcommunicationandcoordinationbetweenprimary,secondary,andtertiarycareservicesasoutlinedabove.Whenusedappropriately,healthinformationtechnologycanfacilitatethiscoordination.Technologycanallowforshared documentation, timely communication between care providers and with patients, real-timealerts,andremotemonitoring(5).Technologyshouldbeemployedtoenabletimelynotificationofkeyeventssuchashospitaladmissions,transfers,anddischarges.Theseeventsarecriticalmomentswhencaremanagementisparticularlyimportanttoensurethetimelyfollowupwiththeprimarycareteam,totriggerconversationsaboutneededchangestopatientcareplans,andtoassesskeysafetyconcernssuchasmedicationfulfilmentandreconciliation(5).Forexample,caremanagementprotocolsintheUSregularlyestablishinformationlinkagesbetweenprimarycarepracticesandhospitals.Whenapatientisseenattheemergencyroomoradmittedtothehospital,caremanagersarenotifiedandstayintouchwithinpatientteams.Onceapatientisdischarged,anexpectedintervalforfollowupisestablished.Foremergencyroomvisitsnotresultinginadmission,caremanagersareoftenexpectedtocallthepatientwithin72hoursofdischarge.Forpatientsdischargedfromaninpatientsetting,theycallwithin24or48hoursofdischargetoschedulea followupvisitwithinthefollowingweek. At this followupvisit, theprovidercanperformamedicationreconciliationandmonitorforsignsofclinicaldeterioration.EnhancedCareManagementTeamApproachandResourceConnections
Caremanagementprogramscanbeledbyavarietyofdifferentorganizationsincludingpayers,hospitals,orthirdparties,butthemosteffectiveprogramsare ledbyprimarycareteamsand locatedwithin thewalls of their practices (3,22).Whilemultidisciplinary care teamsare a vital componentofmany successfulprograms (5), therearealsoexamplesof small teamsofonly twoproviders that canachievesignificantimpactonpatientcare(24).Atanyrate,thecompositionofcareteamsvariesacrossprogramsandshouldbematchedtomeettheneedsofenrolledpatients.
Teamstypicallyincludeattheircoreadedicatedcaremanager,oftenanurse,socialworker,or
communityhealthworkerworking inpartnershipwithaprimarycarephysician.As caremanagementprogramsmature, care teamsmay grow to includeother provider types based on the needs of eachprogram’spatientpopulation.Forexample,socialworkers,communityhealthworkers,andbehavioralhealthspecialistsmaybeincluded.Theinfrastructureandcultureofcareteamscansignificantlyimpactclinical and operational performance. Identifying habits of effective care teams can therefore beinstrumentalincreatingsuccessfulcaremanagementprograms.Co-location,face-to-facemeetings,anduseofsharedITplatformsbetweenalllevelsofproviderscanimproveefficiencyandqualityofcare,andpromote a cohesive team culture (28). Another key practice is the clear assignment of roles andresponsibilitiesacrosstheteam,alignedwiththeoverallteammission.Finally,buildinganenvironmentof support, cohesiveness, and reflection can contribute an effective team atmosphere conducive toachievingthecaremanagementprogram’sgoals.
4. Engagement–UtilizingaParticipatoryProcesstoDesignanECMPrograminEstonia
Thischapterdescribesinitialstepsmadetoimplementanenhancedcaremanagementpilot in
Estoniaandhowthemainstakeholderswereengagedintheprocessofdevelopingarisk-stratification
12
model and care management program. Drawing from the international lessons learned and goodpractices related to risk-stratification and care management described in Section 3 of this report, aWorldBanktaskteam-withthesupportofAriadneLabs–coordinatedwithEstonianstakeholderstodevelopECMprogram.The initialphase involved thedesignofa risk-stratificationapproach to targetpatientswiththemostrelevanthealthrisksinEstoniaandanassociatedinterventionpackagefeasibleforimplementationinthecontextoftheEstonianhealthcaresystem.InitialConsultationswithFamilyPhysicians-February2016
In February 2016, the EHIF hosted an initial consultation with family physicians to initiate adiscussionon thedevelopment andpilotingof a patient at-risk registry. This registrywasmeant to i)fully harness existingdata sources available in the country and ii) serve as a decision-making tool for
family physicians in their work related to patients with selected chronic diseases. Following theconsultation, it was decided to convene a dedicated working group of family physicians with themandate to assist in the design and piloting of themethodology for the registry. Thisworking group
aimed to i)determine theobjectivesof thepatientat-risk registryandoperationalparametersof thepilot, ii) provide input on the development of a claims-based risk-stratification methodology andguideline to incorporate family physician knowledge and intuition, iii) participate in the pilot of the
patient at-risk registry and share experiences, and iv) participate in a final workshop to assess thefindingsofthepilotanddiscusstheregistry’sbroaderimplementationplan.
1stFamilyPhysicianWorkshop-March2016
ThefirstworkinggroupworkshoptookplaceinMarch2016.Theirmainaimwastoagreeonthekey design features of a primary health care-based caremanagement program in Estonia.WBG andAriadne Labs representatives presented concepts and models of ECM to a group of fourteen family
practitionersfromthroughoutEstoniaandstafffromtheMinistryofSocialAffairs(MOSA)andEHIF.Thefamily practitioners participatingwere selected togetherwith the Family Physicians Association (FPA)and EHIF. This group included previous members of the FPA management board, faculty physicians
involved in training Family Physicians in Estonia, and resident physicians in process of training anddevelopingtheirpatientlists.FamilyphysicianswereinvitedbytheEHIFtojointheworkinggroup.Themain selection criteria for the participation in the ECM pilot were English language skills and a high
intrinsicmotivation. At the same time, the EHIF ensured that theworking group of family physicianswouldbe representativeofEstonian familyphysicianpracticesasawholeand theirdifferentworking
conditions (e.g.practice locations inbothruralandurbanareas,groupaswellas solopractices,etc.).Figure 4 shows the distribution within Estonia of family physician practices (whether solo or multiproviderpractices)representedinthepilot.
13
Figure4:GeographicalDistributionofFamilyPhysicianPracticesParticipatingintheECMPilot
Source:OwnElaboration.
In order to agree on the methodology for the risk-stratification of patients underlying a patient
registry,theWBGtaskteamfirstpresenteddataontheburdenofdiseaseinEstoniainordertoidentifypotential target groups for the care management program. To further define criteria for a risk-stratificationapproach,theworkshopparticipantsdiscussedthefollowingquestions:
• Whichclinicalcharacteristicsaremostrelevant?
• Foreachclinicaldiagnosiscategory,whichadverseeventsshouldbeavoided?
• Whichotherco-morbidconditions impairapatient’s likelihoodtobenefit fromenhancedcaremanagement?
• Which other clinical, behavioral, social/economic factorsmake patientsmore or less likely tobenefitfromenhancedcaremanagement?
• Which patient “archetypes” (based on combinations of clinical, behavioral and socialcharacteristics)aremostlikelytobenefitfromenhancedcaremanagementattheprimarycarelevel?
14
Consensuswas reached that the objective of the caremanagement pilot in Estonia should be toimprove patient engagement and health outcomes for patientswith cardio-vascular, respiratory, and
mentalconditions.
As a consequence, thequestionofhow to identify the specific groupofhigh-need (andhigh-utilization)patientsthatareamenabletoECMwassubsequentlyaddressedduringtheremainderofthe
workshop.Afteragreeingonasubsetofdiseasegroups that representa largeshareof theburdenofdisease, yet are amenable to care management interventions, the workshop concentrated ondeveloping archetypes of patients most likely to benefit from care management. These archetypes
formed the basis of the risk stratificationmodel subsequently developed by the working group. Thepatientarchetypeapproachallowedparticipantstofocusnotonlyonpatientswhoarethesickestandmostatriskofanadverseevent,butalsoacohortofpatientswhoareslightlyhealthierandforwhom
caremanagementcouldpreventfurtherdiseaseprogression.
Themetabolic triad (hypertension, diabetes, and hyperlipidaemia)was chosen as the startingpoint or first filter for the risk-stratification algorithm. Themetabolic triad diseases were considered
important both in terms of their share of the burden of the disease and also in terms of theircontribution to theprogressionofotherchronicdiseases (e.g.cardiovasculardiseases). Inorder tobeincludedintheECMpilot,apatientmusthaveatleastoneofthemetabolictriadconditions(seeFigures
5and6).
Thedevelopedrisk-stratificationapproachgroupspatientsnotintermsoftheirpastutilizationofhealthservices,butratherintermsofthechronicconditionsthattheysufferfrom.Atthesametime,
the selection mechanism does not use a simple count of chronic conditions, but rather considers achroniccondition’stype,severityandrelationshipwithotherchronicconditionsincreatingamultilevelor hierarchical risk-stratification model. Non-triad conditions are counted in order to determine the
numberoftotalchronicconditionsthatapatientsuffersfrom.Thisnumber,whenitexceedsacut-offlevel,canbeusedtoexcludepatientsfromtheECMprogram.Theseotherconditionsarenotexplicitlyconsideredasacriterionfordefiningthepatientarchetypes.
PatientsthatcannotsufficientlybenefitfromECMarenotincludedinthepatientlistsproducedby the algorithm. One reason why a patient might not be able to benefit from enhanced care
EnhancedCareManagementinEstonia
Improvehealthoutcomesforpatientswithcardio-vascular,respiratory,andmentaldisease
Pilotobjectives:
! AssessfeasibilityofimplementingenhancedcaremanagementinEstonia
! Understandimpactofpilotoncaremanagementprocessesandselectedpatientoutcomes
! Learnfromexperiencetoinformpossiblescale-upthroughoutthecountry
15
management is thattheseverityofoneorsomeoftheirconditionsmaybetooadvanced.Hence,thepatientselectionalgorithmaimstoidentifyingthosepatientsthatsufferfromadefinedrangeofchronic
conditions (i.e. at least two but notmore than 7 conditions) and can benefit from the interventionsofferedbytheECMpilot.Theseinterventionsincludesecondarydiseasepreventionandimprovedandintegratedmanagementoftheirchronicconditions.
The risk-stratification approachexcludespatientswithnoor limitedpotential to benefit fromcaremanagement at the primary health level.Whether a patient can potentially benefit from a caremanagementinterventionornotisproxiedbyfourdifferentcriteria.Firstofall,patientsthathaveany
diagnosisofacutecancer (cancer in treatment),schizophrenia,dialysisduetorenal failure,congenitalmalformationsrequiringspecializedcare,andrarediseasesareexcludedfromthepatientlistproducedasa resultof the risk-stratificationalgorithm. Likewise,patientswithmore than twoofa selectionof
non-triad chronic conditions are excluded due to the resulting complexity of managing all of theirmedicalconditions.Patientswithmorethanoneofaselectionofmentalconditionsconsideredarealsoexcludedfromthepatientlistsforthecaremanagementpilot,giventhatafamilyphysicianisunlikelyto
beabletotakecareofthecaremanagementneedsofapatientwithtwodifferentmentalconditions.Finally,patientswhohaveatotalofmorethan7chronicconditionsareexcludedfromthepatientlistsandthecaremanagementpilotaswell4(seeFigure5).
Thepatientselectionprocessensuresthatpatientsareselectedwhoseneedsarecomplexbut
notsosevereorcomplicatedastooverwhelmtheECMteamresources.Followingselectionofpatientsvia the algorithm, ECM teams reviewed the list and removed or added names based on their clinical
experienceandintuition,followingthehybridapproachtoriskstratificationdescribedinsection3(SeeFigure6).Oneofthekeyinnovationsoftheagreeduponrisk-stratificationapproachisthatitdoesnotonlyidentifypatientsthathavebeenhighusersofthehealthcaresysteminthepast,butitalsoselects
and targets patients that have not been regularly in touch with their health care providers. Theseunknownpatientshaveriskfactors(i.e.aconditionfromthemetabolictriad)thatifpoorlymanagedcantoleadtoaworseninghealthstatus,diseaseprogression,andfuturehighhealthcareutilization.Given
thenatureofchronicdiseasesandtheirwayofprogressingifnotproperlytreated,patientsthatdonotregularly seek care and helpwith disease prevention from their family physicians constitute another
patientgroupof interestfortheproposedcaremanagement intervention.Thesepatientsarereferredto as unknown patients, because their doctors –while being assigned to them – are not necessarilyawareoftheirneedsandcurrenthealthstatus,becausetheytendtohaveveryfewornovisitsduring
theyear(seegroupmarked“targetgroupII)inFigure7below).
4Allthe45chronicconditionsthatdonotwarrantanimmediateexclusionduetotheirseverity/dominanceareconsideredinthecounttowardsthetotalnumberofchronicconditions.
16
Figure5:ThePatientSelectionAlgorithmfortheEstoniaECMPilot
Source:OwnElaboration.EssentialInclusionCriteria:
- Atleast1Metabolictriadcondition(hypertension,hyperlipidemia,diabetesmellitus)and- 1–4respiratoryand/orcardiovasculardisorders,where0-2ofAsthma,COPDand0-2ofischemic
heartdisease,stroke,congestiveheartfailure,atrialfibrillation.
OptionalInclusionCriteria:
- 0-1mentaldisorders(mooddisorders,alcoholabuse,substanceabuse,dementia)and- 0-2functionaldisorders(visionimpairments,hearingimpairments,frailty).
ExclusionCriteria:
- Morethan2cardiovasculardisorders/morethan1mentaldisorder.- Any diagnosis of certain disorders: Acute cancer, schizophrenia, kidney disorders requiring
hemodialysis,congenitalmalformationsandrarediseases5.- Morethanatotalof7comorbidities:Diagnoses fromthe inclusion list,plus listofmostcommon
chronicconditions(seeAnnex3).
5Thoughifyouwishtoincludethesepatients,youmaymakeanexception.
17
Figure6:VennDiagramofPatientTypesandTwoPotentialTargetGroupsofPatients
Figure Adapted from: Gerard Anderson & Claudia Salzberg (2016), Identifying High Need High CostIndividuals,JohnsHopkinsUniversity.
Followinginternationalexperienceswithcaremanagement,thepatientselectionalgorithmforEstonia identifies complex patients with multiple comorbidities and/or social and behavioral health
problemsastheidealcandidatestobeincludedinacaremanagementprogram.Thesepatientsaccountforadisproportionateburdenofillness,healthcareutilization,andcost.However,carecoordinationfor
thissegmentofthepopulationintheabsenceofanECMprogramcanbechallengingsincetheyoftenreceivecarefrommanydifferentproviders,takemultipleprescriptionmedicines,andhavehighratesoffunctional limitation. This tailoring of the risk-stratificationmodel to the available caremanagement
interventions,andtheinclusionofproviderintuitiontoremovepatientsselectedbythealgorithmandaddnewpatients,ensuresthatthepatientsenrolledinthecaremanagementprogramcanpotentiallybenefit from it. The key criteria of inclusion for any given patient is whether they have high unmet
potential needs andwhether they can benefit from the caremanagement interventions, rather thanwhethertheyhavehadhighcostsorserviceuseinthepast.Thisapproachallowsfortheinclusionofthe“unknown” future high risk patients as described above and focuses on the healthmaximization for
patientsincludedinthecaremanagementprogram.
2ndFamilyPhysicianWorkshop-November2016
Followingthefirstworkshop,thepatientselectionalgorithmwasimplementedandtestedusingEHIF claims data. The 2nd Family PhysicianWorkshop in November 2016 beganwith a review of the
18
selection approach. For instance, working group members were familiarized with the distribution ofcandidatepatientsforacaremanagementprogramacrossthedifferentarchetypes(medicalarchetypes,withoutconsideringsocialandbehavioralpatientcharacteristics,seeTable1)consideredinthepatientselection approach. For this workshop, family nurses joined the working group, attesting to theimportanceofateamapproachforthesuccessofthecaremanagementprogram.
Table1:DistributionofEstonianPatientsAcrossArchetypes(Medical)in2015
Archetypes(Medical) Absolute Percentage CVD&Resp.&Mental&Functional 382 0.20
CVD&Resp.&Mental 2,741 1.47
CVD&Resp.&Functional 1,946 1.04
CVD&Mental&Functional 1,541 0.83
Resp.&Mental&Functional 179 0.10
CVD&Resp. 15,491 8.31
CVD&Functional 8,432 4.52
CVD&Mental 13,441 7.21
Resp.&Mental 2,419 1.30
Resp.&Functional 1,072 0.58
CVD 87,637 47.01
Resp. 16,248 8.71
Anyoftheabove 151,529 81.28
UnknownPatients 34,898 18.72Source: World Bank team calculations.
Mirroring international best practices in caremanagement (see Section 3.), family physicians
and nurses attending the workshop were trained in the 4 key components of enhanced caremanagement,namely i) risk-stratification (utilizing the inclusion/exclusionalgorithmdefinedat the1stworkshop),ii)thepreparationofcaremanagementplans,iii)proactiveoutreachandtransitionsfollowup, and iv) team building with patients and caregivers. In addition, the group discussed overall pilotlogistics,thetimetable,andthemonitoring&evaluationframeworkoftheECMpilot.
Thepatients identifiedbythealgorithmpresentedastartingpointofdiscussion,butwerenotrequiredtobeincludedontheparticipatingphysician’sECMprogramlists.Sincetheprocessaimedtoidentify the patients most likely to benefit from care management (without being unmanageablyresource-intensive), family physicians were asked to refine the algorithm generated patient lists and
19
removepatientsunlikelytobenefit(byprovidingareasonfortheirdecisiontoexclude).Theywerealsoaskedtoaddpatientswhotheyfeltweremissedbythealgorithm.ExclusionandinclusionweredirectedbythegeneralguidancepresentedinTableXXbelow.Table2:Guidanceforcareteamswhenexcluding/includingpatientsforECMprogram
Bewareofmakingassumptionsthatpatientswon’tbenefit/wanttobeinvolved.Beware of biases: try not to exclude patients who elicit negative reactions fromprovidersunlessthereisagoodreason.Patientswithflaggedsocialrisksshouldbe includedunlessthereisaverysignificantexclusionreason.
Tendencies to exclude certainpatientsthatmayhavethepotentialtobenefitmost
Care management programs often benefit patients previously unengaged bytraditionalprimarycaremodels,e.g.poorpastadherencetotreatment,poorhealthliteracyorlackofengagement.If a patient’s disease is severe but likely to benefit from care management, theyshouldbeincluded;ifnot,donotinclude.
Practice capacity to deliver benefitfromcareprogram
Yourassessmentofwhetherparticularpatientswillbenefit fromcaremanagementmight depend on the composition of your care team, access to particular careproviders,capacity,etc.Existing relationships with other providers such as specialist physicians (e.g.oncologist), private care managers, or institutional care providers (group homes,assistedliving)maylessentheadditionalbenefitofcaremanagement.
Currentsupportlevels
Isolated patients may particularly benefit from proactive continuous outreach:absenceof familysupportmay limit theirability tonavigateandnegotiate thecaresystem.
Safetyconsiderations Donotselectpatientswhoarelikelytobeasafetyrisktopractitioners.
FamilyphysiciansinEstoniahave,onaverage,anassignedpatientlistofaround1,700people,andmanyworkinsolopracticeswithonlyonenurse.ConsideringthiscontextoftheEstonianprimaryhealth care system and the objective of evaluating the feasibility of a caremanagement program in
Estonia,apackageofinterventionswasdiscussedforinclusioninthecaremanagementpackage.Threestepswereagreedupontoguidetheimplementationofthecaremanagementpackage.
Thefirststepfortheenhancedcaremanagementonthehigh-riskpatientsisneedsassessment
and care planning. Care plans should be designed first and foremost with patient use in mind andtailoredtoeachpatient.Thefollowingkeyprinciplesshouldbeusedtoachievethisgoal:
● Co-Development:Careplans shouldbeco-developedwith thepatient, careprovider, and/orpatientfamilymembers.
20
● Keep It Simple: Care Plans should be organized in a format that maximizes clarity andpromotesusebyboththepatientandprovider.Itisessentialthatthepatientfeelthecareplanisaccessible.
● Remember the Overall Goals, and Continually Reassess the Work: There is a distinctionbetween a care plan (which is static) and the activity of care planning (which includes anevolvingassessmentbasedoncareprovidedandalignedwiththecareplan).
Table3:ComponentsofthefinalcareplanNeedsassessment
Asummaryofallactivemedicalproblemsandkey issuesthepatientwants toaddress.
Medication Alistofallmedicationsthepatientiscurrentlytakingincludingtimeswhentheyshouldtakethem.
Patientgoals 2-3patientgoalswrittenintheirownwords:whatdoesthepatientwantmostin terms of their health? e.g. improved health outcomes, self-careconsiderations,utilizationofcertain services,meetingpsychosocial challenges,etc.
Actionplan Identifyrelevanthealthissuesthatmightoccurandarticulatecontingencyplans(ifxhappens,thendoy).
Caretransitions Articulatewhat they should do if admitted to hospital (e.g. phone the familypracticetoalertthecaremanagementteam)
Contactinformation PatientandrelativecontactdetailsDoctorandnursecontactdetailsDayandeveningcontactdetails
The second step for the enhanced care management on the high-risk patients is care
coordinationandcommunicationwithotherhealthcareandsocialcareproviders.Thefamilyphysicianswere asked to keep track of the high-risk patients by ensuring compliance with national guidelines(current quality bonus system), reconcilemedication plans and improve adherence, follow up on thehigh-riskpatientsduringcaretransitions(e.g.followupcalls,visitsafterhospitaldischarges,etc.),tracklabtestsandreferrals,andmonitorofpatientsbetweenscheduledvisits.
Prior to the workshop, theWBG team conducted a survey of family physicians to evaluate their
familiaritywithavailablesocialservicesandbenefitsatthemunicipalityandstatelevel.Thesurveyalsodocumented their views on whether coordination with these entities was their responsibility and tosolicit feedback on their experiences interactingwith social services in the past. The survey revealedgaps and misunderstandings regarding the role of family physicians in interacting with social careproviders, particularly with respect to identifying which social resources were relevant in particularpatientcaresituations.
ThefamilyphysiciansinvolvedintheECMpilotwerethereforeaskedtoimproveinformationflows
betweenphysiciancareteamsandsocialcare(servicesavailable,servicesreceived).Themainaimwasto increase coordination with social workers and promote wider implementation of social needsscreeningbythefamilyphysicians.Connectingpatientsinneedwithrelevantsocialservicescanhaveasignificant impacton their qualityof life andability tobenefit fromcaremanagement activities. Thephysicians were asked to contact municipalities to identify available services and establish relevantcontactsperthefollowingprotocol:
1. Review their registry lists to identify themunicipalitieswhere patientswith social care needsreside.
21
2. Contactthemunicipalitiestocreatean inventoryofservicesofferedbyeachmunicipalitywithrelevantcontactinformationandreferralinstructions.
3. Identify relevant points of contact at the municipality to help coordinate referrals for state-providedservices/benefits.
The third step towards enhancing care management involves building an effective team in thepracticesenrolled in theprogram.The familyphysicianswereasked to restructure their currentworkflowsinordertooptimizeoutreachtopatientsenrolledintheECMpilot.Forinstance,nursesandfamilypractitioners involved inthecareplanpilotcouldmeetweekly forapproximately1hourdevotedtoacaremanagementdiscussion,duringwhichtimetheycouldreviewtheregistryanddiscussupdatestoindividualpatientcareplans.Thefamilyphysiciansandnursesinvolvedinthepilotwouldthencreateanactionplanwhereintheydescribeallthepilotrelatedactivitiesandhowthetasksaredividedacrosstheteam.
● Reviewpatientsthattheteamisconcernedaboutforanyreason(changeinhealthstatus,socialchallenge,inabilitytocontact,etc).
● Discusspatientsrecentlyadmittedorseenintheemergencydepartmentandensurethattheyhaveafollowupplanandfollowupvisitscheduled.
● Assesswhichpatientsneedacareplanrevisedorupdated,andoutlinehowthatwillhappen.● Assesswhoneedstobeconnectedwithsocialservicesorreferredtospecialtyphysicians,and
whowillmakethesereferrals.● Updateessentialelementsoftheregistry.● Identifyandassignkeyfollowuptasks(withduedates)toteammembers.● Ensurefollowupactionsareperformedincludingschedulingpatientoutreachorappointments,
establishing follow up plans with patients after hospitalization, updating care plans, andinteractionswithsocialservices,arrangementsforQBS-relatedlaboratorytests,orinitiationofappropriatemedicationsbasedonpatientconditions.
5. Implementation–ConductingtheECMPilotinEstonia
ThepilotwasofficiallylaunchedinJanuary2017,designedasafeasibilityandacceptabilitytestofthecaremanagementprogramandengagingalimitednumberofcareteamsandpatientstoenablearapidtestingandrefinementofthecaremanagementprocesstoensurethattheimplementationcouldbeappropriatelytailoredtothelocalcontextbeforebeingscaled-up.Overall,10familyphysiciansand1residentworkingona totalof9differentpatient lists joinedthepilot.Themainobjectiveof thepilotwas to assess the feasibility of implementing enhanced caremanagement in the primary health caresetting inEstonia, tounderstand the impactof thepilotoncaremanagementprocessesandselectedpatientoutcomesandtolearnfromexperiencestoinformapossiblescale-upthroughoutEstonia.Thecaremanagementprogramitselfwasdesignedwiththeobjectiveofimprovingthehealthoutcomesofcomplex patients (and potentially reducing their needs for healthcare utilization). However, givenexperience from other countries which suggests that fully realizing these outcomes can take severalyears,thissix-monthpilotwasnotexpectedtoresultinimprovedpatientoutcomes.
GuidelineforthePilotImplementation
Prior to thepilot start, participating familyphysiciansandnurses receivedawrittenguidelineexplaining the rationale of the pilot and describing the different pilot activities in detail. For each ofthreekeypilotactivities—applyingintuitiontopatientlists,buildingcareplans,andtheprocessofcarecoordination and dashboard maintenance—the task team had developed Standard Operating
22
Procedures (SOPs). These SOPs were drafted based on a review of the global literature on risk-stratification and care management and the experience of the task teammembers in implementingsimilarprogramsinothersettings.
Thepatientdashboard
Thepatientdashboardisthecentralmanagementtoolrelatedtothecaremanagementpilotforcare teams. It isaworkingdocument thathouseskey information forallpatientspotentiallyenteringthecaremanagementpilot.Thedashboardallows familycare teams to reviewpatient listsandmakeexclusions/inclusions,and–asimportantly-thepurposeofthedashboardisalsotoprovidecareteamswithimportant,standardized,up-to-dateinformationabouteachoftheenrolledpatientstosimplifythemonitoring of their care. There is functionality for patient care plans to be hyperlinked within thedashboard, and in general the dashboard is themain tool for improving patients’ care coordination.Maintaining the patient registry is critical to successful care management, since it allows to quicklyaccesspatients’ care status and currentneeds anddeterminewho is fallingbehind in their careplanand/orrequiresadditionalcareteamattention.However,thedashboarddevelopedbytheEHIFwassetuponlybymid-February,delayingtheprocessofreviewinginitialpatientlists.Alackofuser-friendlinesspreventedthedashboardfrombecomingthekeymanagementtoolforfamilydoctorsthatitpotentiallycouldbe.
CreationofPatientLists
In January, the EHIF sent each family practitioner from theworking group the list of patientsidentified by the selection algorithm for a potential inclusion in the caremanagement program. Theoriginalplanwasforeachfamilypractitionertoenrolatleast20patientspermonthbetweenFebruaryandMay.However,allfamilyphysicianpracticesencounteredsomedifficultieswhenenrollingpatientsduringthefirstmonthsofthepilot.Duetothedelayedreadinessofthedashboard,theEHIFmanuallyselected patients in each practice who fit the pilot algorithm and sent a list of eligible patients toproviders via email. The dashboard was ready by February, but some practices noted discrepanciesbetween their list received from EHIF via email and the one from the dashboard. These technicalchallengesdelayedpatientenrolment.Theoriginalpilot timelineestimated thateachsitewouldhaveenrolled50-75patientsbytheendofMay.Inreality,only40-60patientsfromeachlistwereenrolledbytheendofMay.MonitoringandSupportFunctions▪ A team of local coordinatorswas hired for the logistical support of the pilot. These coordinators
includedalocalWorldBankconsultant,acurrentEHIFstaffmember,andaphysician.Theirrolewasto help family practitioners execute the pilot, troubleshoot problems, and conduct fidelity andoutcomemonitoring. The local coordinators had an absolutely critical role for implementing andexecutingthepilotsuccessfully.
▪ TheEHIForganizedcommunicationandoutreachactivitiestokeepstakeholdersinformedabouttheworkstreamsandprogressmadeunderthecaremanagementpilot.
WebinarsandTraining
Throughoutthepilot,familypractitionersjoinedaseriesofwebinarstoreinforceandrefreshtheirinitialtraining.Thewebinarswereledbythelocalpilotcoordinator(WorldBankconsultant).
Table4:WebinarsandSeminars—TopicsandParticipationRates
23
Videoconferences Seminar
Date 20.January
3.February
15.February
28.February
17.March 5.April 28.April 21.May
Topic ReflectionsonBuildingTeamsinPrimaryCare
Coordinatingpatientcareafterhospitalization
Reviewofproviderintuitionandcareplans
Dashboardandfinalizingthelists
ElicitingPatientGoalsandPromotingPatientActivation
SocialNeedsAssessmentandResourceConnections
StatinsandMedicationReconciliation
SeminaronECMpatientstoriespresentedbythenursesandFP-s
ParticipationRate
91% 73% 73% 45% 73% 64% 45% 91%
To evaluate fidelity of the pilot implementation, the local coordinators conducted monthlymeetings with all family practitioners who were then evaluated qualitatively and quantitatively tomeasure their progress with respect to different aspects of the care management program. Table 5states the different evaluation criteria used. Family practitioners were graded on a 1 (poor) to 5(excellent) (see also Table 5 formore insights on the scale). Togetherwith qualitative data from themonthly meetings, the quantitative score was used to inform pilot implementation in real-time. Forexample,thelocalpilotcoordinatorsnoticedthatmanyofthefamilypractitionerswerehavingtroubledeveloping patient-friendly care plans and decided to hold a webinar about the topic. During thatwebinar,eachfamilypractitionerhadtopresentanexampleofapatient-friendlycareplanfromtheirownpracticeand then received feedback from theotherparticipants.Asa resultof thewebinar, thequalityanduser-friendlinessofcareplansimprovedrapidly.Table5:EvaluationCriteriaforPilotImplementationandMonthlyReportRubricEvaluationCriterion
Description 1 3 5
UnderstandingofPilot
Familiarityoffamilypractitioners
withthepilotgoals,tasks,and
materials.
Doesnotunderstandtheaimofthepilot
andisnotfamiliarwiththetasksormaterials.
Mayunderstandtheaimofthepilotbutis
familiarwithonlysomeofthetasksandmaterials.
Understandstheaimofthepilotandisfamiliarwithallthetasksandmaterials.
ActionPlan
Existenceofaformalplanfor
pilotimplementationdevelopedbythefamilypractitionerandtheirstaff.
Noactionplan.Actionplanis
missingcomponents.
Actionplanincludesallcomponents.
24
UseofIntuition
Understandingbycareteamsofhow
toincludeorexcludepatientsinthepilotbasedontheirintuition.
Doesnotknowhowtoapplyintuitionandnotabletoarticulaterationalforinclusionorexclusion.
Hasappliedintuitiontoincludeand
excludepatients,butreliesheavilyonEHRdata
and/orinappropriatelyexcludespatients
withcertainconditions.
Hasappliedintuitiontoincludeand
excludepatients,understandsthegoalsofintuition,
andmaycontinuetomodifylistasneeded.
TeamWork
Clearlydefinedrolesand
responsibilitiesbetweenphysiciansandnursesandaregularmeeting
timetodiscussthepilot.
Havenotdefinedroles
andresponsibilitieswithintheteamor
proceduresformanaging
patientsonthelist.Haveno
regularmeetingtime.
Haveestablishedsomerolesandresponsibilitiesandproceduresformanagingpatientsonthelist.Havenoorveryinfrequentmeetingsasa
team.
Haveagreedonrolesandresponsibilitieswithintheteamand
proceduresformanagingpatientson
thelist.Haveestablishedfrequentandregularmeeting
times.
PatientEnrolment
Thenumberofpatientsenrolledin
thepiloteachmonth.
Nopatientshavebeeninvitedtoparticipate.
Somepatientshavebeeninvitedtoparticipate.
Allpatientshavebeeninvitedtoparticipate.
CarePlans
Qualityofcreatedcareplans
(measurable,time-boundhealthgoals,aplantoachievethosegoals,contact
informationofthepatient,their
family,andfamilypractitioner).
Havenotestablished
patient-friendlycareplans,andinsteadusethedashboard
outline.Mostpatientgoals,andaction
plans,andcaretransitionsaretoogeneralanddonot
followtherulesofgoodcare
plans.
Haveestablishedpatient-friendlycareplanswhichcontainmostrequired
information.Mostpatient
goals,andactionplans,andcare
transitionsaretoogeneralanddonotfollowthe
rulesofgoodcareplans.
Haveestablishedpatient-friendlycareplanswhichcontain
allrequiredinformation.Mostpatientgoals,action
plans,andcaretransitionsfollowtherulesofgoodcare
plans.
EstablishedLinkwith
Familynurseshavecontactedhospitals
Havenotcontactedthe
Havecontactedthehospital,but
Havecontactedthehospitalandhavea
25
Hospitals toobtainaccesstothehospital’s
electronicmedicalrecordsorto
establishanotherwaytoregularlysharedata.
hospital. donothaveawaytoregularlyshare
data.
waytoregularlysharedata.
RegularCommunicationwithHospitals
Familynurses/careteamsknowaboutpatientadmissions,
discharges,ERvisits,and
ambulancecalls.
Donotreachouttothehospitalorcheckthe
hospitalEMRtodeterminewhether
patientshavebeenseenrecently.
SometimesreachouttothehospitalcheckthehospitalEMRtodeterminewhetherpatientshavebeenseen
recently.
FrequentlyreachouttothehospitalcheckthehospitalEMRtodeterminewhetherpatientshavebeen
seenrecently.
EstablishedLinkwith
SocialServices
Careteamestablishescontactwithsocialworkersandisawareofavailablesocial
services,knowsthemunicipalitieswhereapatient
lives.
Havenotcontactedasocialworker,donotknowwherepatientslive,andarenotfamiliarwithwhat
socialservicesareavailable.
Havecontactedasocialworkerandexplainedthepilot,are
somewhatfamiliarwiththeavailableservices,buthavenotagreedonacontactperson.
Havecontactedasocialworkerandexplainedthepilot,arefamiliarwiththeavailableservices,
andhaveagreedonacontactperson.
RegularConnectionwithSocialServices
Careteamscreensforsocialneeds
Donotseeaneedtoscreen
forsocialneeds.Rarelycontactasocialworkerwhena
needisdiscovered.
Mayknowtheirpatientswell,butdonotregularlyscreenforsocialneeds.Usuallycontactasocialworkerwhena
needisdiscovered.
Regularlyscreenpatientsforsocialneeds.Alwayscontactasocial
workerwhenaneedisdiscovered.
CoordinationofPatientCare
Regularreviewsofcareplanstogether
withpatients,arrangingofnextpatientvisitsin
advance,reviewedcareplanswith
patients,proactivetrackingofpatients’
Rarelyschedulesnextvisitorcontact
timewithpatients.
Rarelyreviewscareplansonfollowupvisits.Rarelycheckswhetherthe
Usuallyschedulesnextvisitor
contacttimewithpatients.
Usuallyreviewscareplansonfollowupvisits.Usuallycheckswhetherthepatientshave
Alwaysschedulesnextvisitorcontacttimewithpatients.Alwaysreviewscareplansonfollowup
visits.Regularlychecks
whetherthepatientshaveboughtthe
medicines
26
medicationadherence,followupswithpatientsrecentlydischargedfromthehospital.
patientshaveboughtthemedicinesprescribed.
Rarelycontactspatientsafterlearningaboutspecialistvisit
orhospitalization.
boughtthemedicinesprescribed.
Usuallycontactspatientsafterlearningaboutspecialistvisitorhospitalization.
prescribed.Alwayscontactspatientsafterlearningaboutspecialistvisitorhospitalization.
6. Results–EvaluatingtheEstonianECMPilotExperience
ThefollowingsectionprovidesasummaryoftheresultsfromtheECMpilotevaluationbasedonananalysisofhealthinsuranceclaims,stakeholderinterviews(keyinformantsandpatients),apre-andpost-pilotprovidersurveyandthemonthlypilotmonitoringreportspreparedbythelocalcoordinators.The monitoring & evaluation framework underlying this analysis as well as the frameworks of keyinformant and patient interviews can be found in Annex 2. The reference for the provider survey isstatedunderReferences.
Inordertotraceanimpactofthepilotusingclaimsdata,adifference-in-differenceapproachis
being employed. As part of the pilot preparations, the entire Estonian population was risk-stratifiedbased on themethodology developed for the ECM pilot in Estonia. As a consequence, the group ofpatients selected by the risk-stratification algorithm but assigned of non-pilot family physiciansconstitutesagoodcomparisongroupforthegroupofpilotpatients:Theyhadcomparablerisk-profilesandutilizationpatternsprior to thepilot, but only thepilot patientswereexposed to enhanced caremanagement. The claims data analysis employs data from February – August 2016 and 2017,respectively.
Feasibility
Thecaremanagementprogrammeteachof its implementationtargetareasandprovedtobefeasible. A total of 466 patients (197 men and 269 women) were enrolled (and not subsequentlyexcluded)inthecaremanagementpilotprogrambetweenFebruaryandAugust2017.Despitetheshortduration of the pilot, the ECM program was quickly adopted by participating practices, as animplementationsciencesummativematrixofperformance(seeFigure7)quicklyshows.Theadherencewithkeypilotactivitiesconsistentlyimprovedfromthestarttowardtheendofthepilot.
Finally, all the key informants stated that the pilot did meet the objective of proving thefeasibilityofcaremanagementinthecurrenthealthcareorganizationalmodel.Thepilotshowedthatitispossibletoimprovethecollaborationbetweensocialandhealthcareproviders.Challengesremaininthe collaboration with hospitals and social workers as well, but important improvements have beenmade.
27
Figure7:ChangeinPilotAdherence(AcrossallFPSites)
Source:OwnElaboration.
PatientEnrolment
NotalltheFPsreachedthegoalofenrollingmorethan50patientsandcreating50careplans.However, 92% of enrolled patients have a completed care plan. The last FP finalized her list on the
dashboardonly inmid-March. Initially,mostcareplanswerepoorlydesigned.Oneproblemwas that,instead of developing a custom plan with each patient, some family practitioners simply printed offgenerictemplatesfromthedashboard.Thesegenericcareplanswerenotpatient-friendly.Forexample,
theyincludedpharmaceuticalcodesunderthemedicationslistinsteadofthenameofthemedications.Another problem was that family practitioners and patients both found it difficult to come up with
measurable,time-boundgoals.Toimprovethequalityofthecareplans,thelocalcoordinatorsdecidedtoholdawebinaraboutthetopic.Duringthatwebinar,eachfamilypractitionerpresentedstoriesaboutat leasttwooftheirpatientsandgaveanexampleofthecareplansfromtheirownpractice.Thecare
teams also received a comprehensive template in Estonian on how to conduct care plans. After thiswebinar, the local coordinators found that thequality of care plans formany sites improved. Patientinterviewsshowedthatabouthalfof the interviewpatientswereawareof theexistenceof theircare
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
28
plans. In thosecases,patients report that their careplanswerealsobeingused tomanage theirowncareneeds.
LinkwithHospitalsNearlyeverypilotsitefounditdifficulttoworkwiththeirmunicipalhospitalatfirst.Mostsites
were not able to arrange for notification whenever a patient was discharged from the hospital.However,bytheendofthepilot,4ofthe9siteswereabletoregularlyaccessthehospital’selectronic
medicalrecordstoreceivedataabouttheirpatients.
Acceptability
By patients: At the start of the pilot, there were some concerns among stakeholders thatEstonian patients might not be very willing to enrol in a care management program. Also, key
informants mentioned that the enhanced care management model make patients even moredependant. However, less than 10% of the patients thatwere approachedwith the offer to join theenhancedcaremanagementprogram,actuallyrejectedtobeincludedinthepilot.Insomepractices,no
patientsdeclinedtoparticipate.Infact,thepatientinterviewsshowthatabouthalfoftheinterviewedpatientswouldrecommendthepilottobeextended.TheseareexactlythepatientsthatalsostatethattheyactuallyknowtheircareplanandwereproperlyintroducedtothepilotbytheirFPs.Theremaining
patientsare indifferent regarding thepilot,giventhat theydidnot reallynoticeanychange inservicedelivery.
Byfamilydoctors:Whilefamilydoctorsingeneralembracedthepilot,fewamongthemdidnot
think that their values as practitioners would fit the pilot well. In that regard, the recurrently citedtheme was patient responsibility. Several family practitioners expressed that patients should beresponsiblefornotifyingtheirphysicianafterbeingdischargedfromthehospital.
ProcessThe utilization of PHC services increased for pilot patients across the board relative to the
comparison group (Table 6). However, the relative increase in the use of physical visits of the familydoctorswasminimal.Incontrast,theincreaseintelephoneconsultations(bothwithnursesandFPs)aswell as the increase in thenumberofpilotpatient interactionswithnurses ingeneral isnotable.TheincreaseinphoneconsultationsofpilotpatientsisasignofthebettercarecoordinationofferedtoECMpilotpatients,whileatthesametimethefactthatin-personvisitswithfamilydoctorsdidnotincreasefor pilot patients underlines the feasibility of ECM (nomajor additional resources areneeded for thecareprovisiontopilotpatients).Table6:ChangesinPer-CapitaUtilizationofPHCservices
Changesinper-capitautilization
Pilot ComparisonGroup
Diff-in-DiffPercentagechange
29
1st-timeorfollowupvisitwithGP
(0.762) (1.042) 10%
ProphylacticvisitwithGP
0.026 0.011 53%
HomevisitbyGP 0.004 (0.001) 52%PhonecallwithGP 1.617 0.982 118%Consultationwith
nurse0.208 (0.021) 16%
Homevisitbynurse 0.026 0.001 296%Phonecallwith
nurse1.065 0.088 237%
Source: World Bank team calculations.
At the same time, the provision of laboratory diagnostic tests to pilot patients increasedconsiderably in comparison to comparisongrouppatients.Table7 statesa listofprocedures thatarepartoftheguidelinesfordiabetes/hypertensionpatientsandthatarealsoreflectedintheQBS(withtheexception of albuminuria testing which is no longer a QBS-relevant procedure as of 2017, see thecorrespondingimmensedeclineintheuseofthisdiagnosticstest).Table7:PercentageChangesinDiagnosticLabTestingforPilotandComparisonGroupPatients
%Change2017vs.2016
Pilot ComparisonGroup
Difference
Albuminuria -77.4% -88.1% 10.7%Cholesterol 38.5% -6.1% 44.7%Cholesterolfractions
42.9% -6.0% 48.9%
Creatinine 32.5% -10.3% 42.9%EKG 80.4% -4.5% 84.9%
Glucose 44.4% -4.3% 48.7%Glycated
Hemoglobin33.1% -2.2% 35.3%
Potassium 25.0% -6.6% 31.6%Source: World Bank team calculations.
30
OutcomesTables8and9statemajoroutcomesforpilotpatientsandattesttotheimprovedqualityofcare
delivered to pilot patients. The number of pilot patients with a prescription of statins increased by
almost 12 percentage points or about 30% from 2016 to 2017. At the same time, and reflecting theincreased number of phone conversations between pilot patients, the percentage of acute careadmissions that had a follow up within 30 days of discharge increased drastically by 20 percentage
points to more than 70% of all cases of discharged patients. The average time between a hospitaldischargeandthefollowupvisitdecreasedbyabouthalfadayforpilotpatients,mirroringthedecrease
forpatientsinthecomparisongroup.
Table8:%ofPatientsReceivingaPrescriptionofStatins
%ofPatientswithStatinPrescriptions
Pilot ComparisonGroup
Difference
2016 38.6% 31.5% -7.1%2017 50.6% 31.8% -18.8%Change 12.0% 0.3% 11.7%
Source: World Bank team calculations. Table9:Followupswithin30DaysafterAcuteCareDischarge.
%ofPatientswithPost-AcuteCareFollowupCallorVisit
Pilot ComparisonGroup
Difference
2016 52.4% 57.7% 5.3%2017 71.7% 56.4% -15.3%Change 19.3% -1.3% 20.6%
AverageTime(inDays)betweenDischargeandFollowup
2016 8.77 10.90 2.132017 8.32 10.46 2.14
31
Change -0.45 -0.44 -0.01Source: World Bank team calculations.
Regardingmedicationreconciliationandadherence,boththeprovidersurveyaswellaspatientinterviewssuggestthatfamilyphysicianshaveobtainedbettermeansofcommunicationswithspecialist
inordertocarryoutmedicationreconciliationandthatpatientshaveactuallynoticedchangesintheirmedication regime reflected in their care plans. Examples include patients that have received newmedications and others that have stopped taking superfluous medicines. While the interviews and
surveysshowstronganecdotalevidenceforapositiveimpactofthepilotonmedicationreconciliation,the available claims data does not clearly allow us to confirm that observation. At the same time,improvements in medication adherence cannot be traced down in claims data from the pilot
implementationperiod, given thatmanyprescriptionsmade in the lastmonthshavenot expired yet,eventhoughtheyhavenotbeenpickedup.Hence,noconclusivestatementcanbemadeonthebasisoftheanalysisofclaimsdata.
Facilitators(+)andBarriers(-)
DashboardReadinessandTechnicalDifficulties(-):Notallpreparationmaterialswerereadyontimefor
the pilot start. To keep the pilot on schedule and not slow down the enrolment process, the EHIFprovided lists of potential patients to each family practitioner in the form of MS Excel sheets. ByFebruary, the EHIF had successfully developed the dashboard, but some family practitioners found
discrepancies between their initial patient lists and the new lists from the dashboard,which createdconfusion among family practitioners and frustrationwith thedashboard. Several family practitionersrequestedguidelines for thedashboard. Inaddition, thecurrentdevelopmentof thedashboard isnot
very user-friendly and requires family physicians to perform a lot of extra and double reporting,effectivelydecreasingtheirmotivation.
Participatory Pilot Design (+): The pilot initiation did take more time than expected, but the
implementationprocesseshadbeenconceivedincooperationwithfamilyphysicians. Inparticular,thealgorithmwas developed together with the same family physicians and nurses who piloted the caremanagementprogram.Thefactthattheintuitionoffamilyphysiciansregardingtheirpatientsisapplied
intheselectionofpatientsincreasedtheirmotivationandhasprovedtobeakeysuccessfactorforthepilot.
LanguageBarriers(-):Oneearlybarrierwaslanguage.SincemostofthepilotactivitiesareinEstonianand English, there was a language barrier for parts of Estonia with many Russian speakers. ThedashboardandexamplecareplansareinEstonian,whichmadeitdifficultforRussian-speakingpatients
tounderstandthecareplans.
Involvementof familynursesand joint learning (+):One success factorexpressedby key informantswastheinvolvementofnurses.Thepilotinitiatedchangesinhowthefamilyphysiciansandnurseswork
together and the service delivery model of enhanced care management gave nurses more
32
responsibilities.Thishasenforcedchangesinthewaythenursesanddoctorsworktogetherwithintheirteams andhow they coordinate chronic patients care and share taskswithin their practice. Thepilot
designalso focusedon joint learning,bringingprimaryhealthcareproviders togetherandgiving themthe possibility to learn from each other’s experiences. Key informants highlighted the importance ofhavingthepossibilitytoconsultpatientcaseswithpeers.
7. ConclusionsandPlanningforScaleConclusions
This pilot study has demonstrated that enhanced care management is both feasible andacceptable in the Estonian health care system. All pilot activities—including the development of an
algorithmtoidentifypatients;theapplyingofproviderintuitiontofinalizepatientlists;theformationofteams of family doctors and nurses within primary care practices; the enrolling of patients; theestablishment of individual patient care plans; and building connections between primary care and
hospitals aswell as the social services sector—were completedonor aheadof schedule. ClaimsdataanalysisfromEHIFshowsimprovedratesof30daysfollowupvisitsafterhospitalization,improvedratesofappropriatestatinprescriptions,andimprovedcholesteroltesting.Therewasanon-significanttrend
towardimprovedspecialistvisitsandavoidableadmissions.
Basedonthisdemonstratedsuccess,theEHIFhascommittedtoscalingupECMandpreparedanactionplandetailingnextstepsandtargets.Establishedtargetsincludeexpandingthepilottoatleast
20 primary care practices in 2018 in order to receive data, analyze the progress and develop acomprehensive monitoring and evaluation framework. The future goal is to roll out a nation-wideenhancedcaremanagementprograminEstonia.Inordertomeetthistimeframeandsuccessfullyscale
theprogram,severalactionitemswillberequired,asoutlinedbelow.
AStrategyforScale
ScalingtheEnhancedCareManagementPilotfromninefamilyliststoall800+familylistsinthe
whole country is a laudable but ambitious target. Achieving this goal will require a clear, deliberatestrategyforscalethatconsidersperspectivesofallstakeholders,includingEHIF,familydoctors,nurses,patients, hospital representatives, and stakeholders from relevant social services. It will require an
ongoing commitment to technical assistance through both direct and web-enabled coaching andcoordination services.Most of the key informants actually stated that the scaling of ECM should beincremental.Thescalingstrategyshouldaddressimportantconsiderationssuchas:
• Theoptimalmethodandtimingofenrollingnewpracticesintotheprogram.• Thepaymentsandincentiveschemeprovidedtothesepracticesfornewactivities.• Otherstakeholderinvolvement(i.e.hospitalsandsocialworkers).
• OtherongoingorplannedchangestoprimarycaredesignandthehealthsystemoverallintheEstoniancontext.
33
For the next phase of the pilot, a number of resources are necessary to draw upon. The FPparticipants andEHIF staffwhowere involved in the first pilot are a critical group that couldprovide
coaching and advice around scaling the model. Focused resources directed toward coaching,educationalmaterials in Estonian, anddirect practice supportwill be required.Over the longer term,educationaltrainingwithingraduateandcontinuingmedicalandnursingeducationregardingthevalue
andpracticeofenhancedcaremanagementwillbenecessary.
The planned staged expansion of the Enhanced Care Management Pilot affords a valuableopportunity to strategically plan for a building and learning period in 2018 that employs rapid-cycle
testing in order to iterate and improve on program materials and procedures in preparation for anationwide scale-up. At the same time, already participating care teams can further increase thenumber of patients to be enrolled in the care management pilot. Currently, only around 3% of
empanelled patients joined the pilot on average across pilot practices, whereas a typical targetpercentageofpatientsforcaremanagementprogramsisaround6-7%ofallempanelledpatients.
PatientDashboardDevelopment
A common challenge faced by participants in the pilot was the limited functionality of thedashboarddevelopedtoallowproviderstotrackpilotactivities,storecareplans,andfacilitateproactivecaremanagement.Beforebeginningtheenrolmentofnewcareteamsintothepilot,itwillbeimportant
thatthispotentialbarriertoscaleismitigated.Thiscouldbeaccomplishedbyconductinganassessmentofchangesneededtothedashboardandusertestingofareviseddashboardtoensurethatthedesignisuser-friendlyandsupportiveofthecoreECMactivitiesthatcareteamsneedtoundertake.
ImprovedImplementationandUpdateofthePatientSelectionAlgorithm
Feedback from family doctors and nurses involved in the pilot activity suggests that thealgorithm to identify patients for pilot participation – as it has been implemented -may need to berevisedand/orupdatedtoensurethatthosepatientswhoaremostlikelytobenefitfromECMarebeing
targeted. Using data from the pilot period and in consultation with nurses and family doctors, animportantnextstepwillbetorevisitthepatientarchetypestargetedbyECMinordertosolicitspecific
care team concerns and identify areas for improvement. Following refinements of the patientarchetype, the algorithm used to generate patient lists will need to be updated. Implementing therandomselectionmechanisminordertobalancethesizeofpatientlistspassedontofamilyphysicians
fortheirreview(seeFigureA1inAnnex3)wouldbeastraightforwardbutsubstantialimprovementofthe implemented patient selection algorithm. Furthermore, the prioritisation of patients within theregistrylistbasedonbehaviouraldata(i.e.whetherthepatientshavefilledalltheirprescriptionsduring
pastmonths)andsocialpatientcharacteristics(e.g.whetherthepatientmaybesociallyvulnerable)stillneeds to be fully developed and then used, as currently this information is not used in the patientselectionprocess.
ScalableCoachingMethodologyThecoachingemployedduringthepilotperiodwascriticaltothepilot’ssuccess.Whileitmaybe
feasible to scale the coachingmethodemployedduring thepilot to at least 20 family physicians, the
34
eventual scale-up of ECM across all of Estonia will require a different model for providing coachingsupporttoensurefeasibilityandcompatibilitywithavailablepersonnelandfunding.Therefore,akeyto
successfully achieving the targets established by EHIF will be developing a sustainable coaching,mentorship, and problem-solving support system to all care teams involved in enhanced caremanagement.Thesupportsystemshouldbuildonthelessonslearnedaboutwhichcoachingfunctions
and activities were most helpful to pilot participants. A plan for expanding coaching activities to allmembers of the care team involved in care management activities, including nurses, should beprepared.
TrainingProgramandMaterialsTrainingonenhanced caremanagement forpilotparticipantswasalsoessential to thepilot’s
success. Training activities included two workshops as well as a series of six webinars focused onbuilding the knowledge and capacity of participating family doctors. Similar to coaching, in order tosuccessfully scale enhanced caremanagement to all family doctors in Estonia, itwill be necessary to
develop a scalable and sustainable training program and supporting materials for care teams. Thisprogramshouldaddresstrainingforthefollowinggroups:
ForPilotDoctors
Thetopicsaddressedthroughtrainingduringthepilotprimarilyinvolvedhowtogetstartedwith
enhancedcaremanagementactivities.However,asfamilydoctorscontinuetheseactivitiespastthesix-monthpilotperiod,anewsetofskillsandcapacitieswillbeneeded.Thetrainingprogramandmaterialsthatwillbedevelopedshouldaddressthecontinuingeducationneedsoffamilydoctorsandcareteams
to ensure that the full cycleof ECMactivities is coveredby available training.New topics that futuretrainingmayneedtoaddressinclude,amongothers:supportingcomplexpatientgoals,suchasweight
loss;ensuringadynamicpatientregistryovertime;andscreeningforsocialneedsandconnectingwithsocialservices.
FornewFamilyDoctorsThe training strategy should also address how to sustainably provide training—both on
foundational pilot activities as well as more advanced topics—at scale to a large number of familydoctors.The familydoctorsparticipating in theprogramhavehighlightedthatat thebeginningof thepilot, it was essential to have a more thorough training program about ECM. Training activities
conducted during the pilot periodmay provide a starting place for this program.However additionaltrainingmodalitieswillalsolikelyneedtobeexploredinordertoreachallfamilyphysiciansinEstonia,bothpriortotheirenrolmentinECMactivitiesandonanongoingbasisastheirengagementwithECM
progresses.
FornursesTrainingactivitiesinthepilotperiodwerelargelytargetedtofamilydoctors.However,asisclear
fromthepilotexperienceandresults,nursesareessentialcontributorstoenhancedcaremanagement
teams and successful scale will require fully capacitating and enabling nurses. Therefore, it will beimportantforthetrainingstrategytoestablishaplanfortrainingnurses(continuingandnew)alongsideorinadditiontotheirphysiciancounterparts.
35
LinkwithSocialCare
The assessment of social care needs should be further strengthened as part of the caremanagement program, and possibilities of working more directly with the social sector should be
explored.Asaresultofthepilot,participatingfamilyphysiciansbecameawareoftheopportunitiestohelp theirpatientswith their social careneeds.While theawarenessandacceptabilityof this activityincreasedamongpractitioners,theactualnumberofpatientsthatreceiveanykindofsocialcareservice
stillisminimal.
TowardPatient-level,Activity-basedCostingofECM
TodayEHIFlacksamechanismtoreimbursefamilyphysiciansparticipatingintherisk-stratified,
enhanced care management program for patients with multiple chronic conditions and social andbehaviouralrisks. Aspartoftheprocessofscalingupthepilotingprogram,apaymentmechanismtoreimburse participating family physicians for the costs incurred and to incentivize them to provide
enhanced care management should be developed. The international experience suggests differentoptions to pay for enhanced care management and coordination activities6. These payment choicesrangefromaflatrateforeachenrolledpatienttopaymentsforbundlesofor individualservicesorto
performance-basedpayments.
A first option is to compensate primary care providers in the form of a prospective, add-on
paymentforallpatientsenrolledintoanenhancedcaremanagementprogram.ForEstonia,thisseemsat least in the beginning the most appropriate choice. With the pilot, the EHIF adopted a risk-stratification approach, which identifies patients most likely to benefit from enhanced care
management.Nevertheless,lessonsaboutgoodperformancearestilllimited,evenattheprocesslevel.Because of that, it seems premature to tie payments for care management and coordination toperformance indicators and targets as part of the current quality bonus system. Yet, close provider
monitoring and other quality assurance measures will be required to ensure that patients receiveadequate care. The risk-stratification system itself provides sufficient mechanisms to ensure thatproviders do not ‘dump’ patients that are difficult and costly to manage. Paying providers per
performancemaybeanoptioninthefuture,wherethefixedcomponentofthecaremanagementandcoordinationcomponentshouldbepartofthecapitationpaymenttoavoidunnecessarycomplexityoftheprimaryhealthcarepaymentsystem.Thebestfirststeptostartwith,wouldbeanadd-onpayment
forallpatientsenrolledintoenhancedcaremanagementprogram.
Oneoftheaimsofthepilotwastoestimatetheresourcesneededtopermanently implement
enhancedcaremanagementinprimaryhealthcareinEstonia.
6WorldBank,2017:Towardgreaterintegrationofcareandimprovedefficiency-AcriticalreviewofEHIF’spaymentsystem.SummaryReport,WorldBank.
36
Currently amixed payment system including capitation, allowances, fee-for-service paymentsand quality bonus payments covers the costs of providers to deliver a defined set of services for
everyone. This includes chronic care management. Capitation payments cover the costs of labour,medicalandnon-medicalequipmentanddevices,medicines,aswellasofficespaceandadministrativeactivities. A monthly basic allowance covers the cost of premises, IT systems, transportation and
training.Sincethecostsforlabour,equipment,premises,labtestsetc.arecoveredalreadythroughthecurrentpayments system,estimating theextra time thatnursesand familyphysicians spendonpilot-relatedtasksisthemostrelevantquestion.
All primary healthcare teams that participated in the pilot kept track of the time spent on
different pilot-related activities to assess the resources necessary to conduct enhanced caremanagement. The time that nurses spent per patient and on the non-face-to-face activities (i.e.
activitiesrelatedtocarecoordination,registryandteambuilding)wasbeingkepttrackof.Thenurseswereaskedtoreportonthetimethatwasspentbyfamilyphysicians,nursesaloneorinteamsonpilot-relatedactivities.ThenursessubmittedthedataeverymonthtotheEHIF.
During the monthly monitoring visits, family physicians reported the average time they had
spentusingtheregistryorapplying intuitiontothepatient listsaswellas theaveragetimespentperpatienttocreateinitialcareplansandmakefollowupvisits.Familyphysicianswerealsoaskedtoreport
ontheregularityofteammeetingstodiscussthepilotpatients.Theinformationprovidedbythefamilyphysicianswascomparedtothedatareportedbythenurses.
By the end of the pilot, nine main activities covering all tasks related to enhanced caremanagementhadbeenidentified:
1. Invitingofpatients,receivingpatients’agreementtoenrolinthepilot.2. Preparationofthefirstpatientvisit.3. Firstvisits,creationcomprehensivecareplans.4. Phonecontacts.5. Followupvisits,updatesofcareplans.6. Reviewingthepatientlist,decisionstoinclude/excludepatients.7. Teammeetingstodiscusspatients.8. Reviewingandupdatingofthepatientregistry.9. Coordinationofpatienttransitions(socialorspecialistcare).
Thefirstfiveactivitiesarerelatedtooneindividualpatient’scaremanagement.Invitingpatients
to enrol in the program and receiving patients’ agreement has usually been a task of nurses in the
practicesparticipatinginthepilot.Thefamilyphysiciansandnursesarebothinvolvedinthepreparationofthepatientvisits.Thisusuallyincludesthereviewofpatients’medicalrecords,currentmedications,thegenerationofaplanforanalysesorneededtestsetc.Thefirstvisitwasusuallydonebythefamily
physicians, because often the patients needed changes in the treatment plan (newmedications or a
37
changingcurrenttreatmentplan).However,therewerealsopractices,wheredoingthecareplanswasthenurses’responsibility(twopracticesoutofnine).Thecostingproposalshouldtakeintoaccountthat
family physicians were the ones who mostly conducted the first care plans. Most of the follow upactivitieswere supposed to be the nurses’ responsibilities, including follow up phone calls and visits.Nevertheless, the claims data from the pilot period show that phone calls were still made by family
physicians.
The last four activities arenot exactly related to the caremanagementof individual patients.
These activities should take place every week or month to help to organizing enhanced care
managementintheentirepractice.Both,thenurseandthefamilyphysicianshouldreviewthepatientlistandinclusions/exclusionsofpatients.Inbiggergrouppractices,thismightalsoneedtheinclusionofotherstaffmemberswhohavecontactwiththehigh-riskpatients.Mostofthepractices(7outof9)did
setuparegularmeetingtimeforthenurseandfamilyphysicianwhentheycandiscusspatientcasesorcare management. Following up on and updating the patient registry has mostly been the nurses’responsibility. This includes following up on the info thatwasmade available through the dashboard
aboutwhetherpatientshaveboughttheirmedicinesandwhethertheyhavehadahospitaldischarge.Inaddition,nursesneededtokeeptrackofwhenfollowupcontactsweremadewithpatientsandwhenthetimeof followupvisitschanged. Inmostof thepractices,bothnursesandfamilyphysicianswere
involvedinthecoordinationofpatienttransitions(socialandspecialistcare).Usuallythefamilydoctorswere the ones who did get in touch with the social workers if a social need was discovered. Thepracticeswhohadaccesstohospitalmedicalrecordshadusuallyagreedthatthenurseswouldregularly
review the information on hospital admissions or emergency room visits. Inmore complicated cases,sometimesalsofamilyphysicianscheckedtheelectronicmedicalrecords.
Adraftactivity-basedcostingmodelforenhancedcaremanagementhasbeendeveloped,taking
into account the information provided by primary healthcare providers. The draft proposal will besharedanddiscussedwiththeEHIF.
38
Annex1:PerformanceMonitoringFramework
FeasibilityEvaluationQuestion Indicatorsforassessment DataSources
● Completedregistryprocesses o %ofpracticesapplyingintuition
topatientlistsandprovidinginclusion/exclusionrationale
o %ofpatients,bypractice,enrolledinprogram
o %ofcareplans,bypractice,completed
Monthly monitoring of thedashboardMonthlyreports
● Qualityofcareplans Monthlyqualitativereviewviamonitoringvisits
● %ofpracticesthatestablishlinkwithlocalhospitaltotrackpatientadmittances/discharges
Monthlymonitoringvisits
1. Wasthepilotfeasibleinthetimeallowed,withtheavailablestaff?
● %ofpracticesthatestablishlinkwithlocalsocialservicescoordinator/providertofacilitatecoordination
Monthlymonitoringvisits
AcceptabilityEvaluationQuestion Indicatorsforassessment DataSources
● Changesinprovidersatisfactionwithpracticingmedicine
Providersurvey
● Changesinperceptionofqualityofcareprovided
Providersurvey
2. Whatwastheexperienceoftheproviders?
● Changesinperceivedstressofjob Providersurvey
● Patientacceptance/enrolmentrate
Dashboard
● Patientexperienceofcoordination PatientFocusGroupDiscussions
3. Whatwastheexperienceofthepatientsinvolvedinthepilot?
● Patientconcernswithprogram PatientFocusGroupDiscussions
4. Whatwastheexperiencewiththe
● Didthepilotmeetthegoals? KeyInformantInterviewswithEHIFstaffandother
39
pilot? stakeholders
ProcessEvaluationQuestion Indicatorsforassessment DataSources
● Changestoteampracticeandfunction
o %ofteamsthatinstituteregular(weekly)meetingtodiscusspatientsinpilot
o Self-reported/observedchangestonurses’duties
Monthlymonitoringvisits
● %ofpatientswithpost-acutecarefollowupcallorvisitwithORaveragetimebetweenacute-caredischargeandprimarycarefollowupcallorvisit
Dashboard/claimsdata
● Changesinreportedfrequencyofpost-hospitalcoordination
Providersurvey
● Changesinreportedfrequencyofcontactingpatientsbetweenvisits
Providersurvey
● Changesinreportedfrequencyofcoordinationcarewithsocialservicesorothercommunityproviders
Providersurvey
● Changesinreportedpreparednessofpracticestomanagepatientswithmultiplechronicconditions
Providersurvey
● Changesinreportedpreparednessofpracticestomanagepatientswithsubstance-userelatedissues
Providersurvey
● Changesinreportedpreparednessofpracticestomanagepatientsinneedofsocialservicesinthecommunity
Providersurvey
● Changesinreportedpracticeuseofpersonneltocoordinatecareforpatientswithchronicconditions
Providersurvey
● Changesinreportedfrequencyofpatientswithchronicconditionsbeinggivenwritteninstructionsformanagingcare
Providersurvey
● Changesinreportedfrequencyofpatientswithchronicconditionshavingself-managementgoalsrecorded
Providersurvey
● Changesinreportedfrequencyofpracticereceiptoftimelineinformationpost-specialistvisit
Providersurvey
5. Howeffectivewasthepilotatimprovingprocessofcaredelivery?
● Changesinreportedfrequencyofpracticereceiptofnotificationpatientshavebeenintheemergency
Providersurvey
40
department ● Changesinreportedfrequencyof
practicereceiptofnotificationofpatientdischargefromhospital
Providersurvey
● Changesinreportedtimeframeforpracticereceiptofinformationneededtomanagepatientspost-hospitaldischarge
Providersurvey
● Changesinreportedmethodofreceivingpost-hospitaldischargeinformation
Providersurvey
● Changesinreportedeaseofcoordinationpatientcarewithsocialservices
Providersurvey
OutcomesEvaluationQuestion Indicatorsforassessment DataSourcesHow effective was thepilot at improvingoutcomesofcaredelivery?
● Avoidablespecialistvisitrate Claimsdata
● %ofpatientsinneedofstatinswithprescription
Claimsdata
FacilitatorsandBarriersEvaluationQuestion Indicatorsforassessment DataSourcesHow can Estonia replicateandspread?
● Mainfacilitatorsofimplementation MonthlymonitoringvisitsEHIFKeyInformantInterviewsPatientfocusgroupdiscussions
● Mainbarrierstoimplementation MonthlymonitoringvisitsEHIFKeyInformantInterviewsPatientfocusgroupdiscussions
41
Annex2:FrameworksofKeyInformantandPatientInterviewsKeyInformantInterviewsInordertoassesstheresultsofthecaremanagementpilot,9keyinformantsfromtheEstonianhealthcaresystemwere interviewed. Interviewstookplacebeforethepilotstartandafterthepilotofficially
ended.Theobjectiveoftheinterviewswastoi)assesstheacceptabilityofenhancedcaremanagement,ii)understandthestrengthsandweaknessesofthepilot, iii)assessthefeasibilityanddiscoverbiggestconcernsanddrivers implementingenhancedcaremanagement, and iv) identifypotential constraints
andopportunitiesforscaling-up.
Keyinformantinterviewsincluded:
(i)RepresentativesfromtheEstonianHealthInsuranceFund,
(ii)RepresentativesfromtheMinistryofSocialAffairs,
(iii)Familyphysiciansparticipatinginthepilot,
(iv)Hospitalmanagers,and
(v)Socialworkers.
Pre-pilotinterviewsfocusedonthefollowingquestions:
1. Whatareyourgoalsandexpectationsforthiscarepilot?2. Howoptimisticareyouthatthesegoalswillbemet?3. Whatdoyouseeasthegreateststrengthsofthispilot?
4. Whatareyoumostconcernedaboutandwhy?5. WhatdoyouthinkisthemostimportantthingthatEHIFcanlearnfromthispilot?6. Whatdoyouthinkthispilotneedstodemonstrateinordertobetakentoscale?
Post-pilotinterviewsfocusedonthefollowingquestions:
1. How,ifatall,didyourgoalsandexpectationsforthispilotshiftovertime?Why?
2. Inwhatway(s)wereyourgoalsandexpectationsmetforthispilot?Inwhatway(s)weretheynot?
3. Whatdidyouseeasthegreateststrength/mostsuccessfulcomponentofthispilot?
4. Whatdoyouthinkdrovethesuccessesofthispilotandwhy?5. Whatdidyouseeasthebiggestchallengeofthispilot?6. Whatfactorsdoyouthinklimitedthesuccessofthispilotandwhy?
7. WhatisthemostimportantthingthatEHIFlearnedfromthispilot?Whyisthisimportant?8. Basedonthispilotexperience,wouldyourecommendtakingtheenhancedcaremanagement
programtoscaleinEstonia?Whyorwhynot?
42
9. Ifyes,whatfactordoyouthinkwouldbemostcriticalforscaling?10. What,ifanything,aboutthepilotexperiencedoyouthinkwouldneedtochangetomake
scalingeffective?PatientInterviewsArandomsampleof18patientsparticipatingintheECMpilotwasselectedaftertheterminationofthe
pilot for interviews to evaluate their experience with the pilot and to identify main obstacles ofimplementation as well as the potential constraints and opportunities for an extension of the ECMprogram. The final samplewas chosen froman intermediate sample containing 4 patients fromeach
practice. Another telephone survey was conducted among pilot patients from the Järveotsa primaryhealthcarecentre.
ListofQuestionsforthePatientInterviews:
1. Haveyoubeeninformedabouttheenhancedcaremanagementprogram(yourparticipationin
it)?2. Doyounoticechangesinthecarethatyoureceivecomparedtopreviousyears?3. DoyoufeelthatyourchroniccaremanagementhasimprovedsinceFebruary2017orthatithas
remainedthesame?4. Doyouknowwhetheryouhaveacareplan?5. Haveyouvisitedahospital,calledanambulanceorgonetotheemergencyroomsinceFebruary
2017?Ifso,wasthereanyfollowupfromyourfamilyphysicianand/ornurse,anddidthefollowupprocedurechangewithrespecttoprevioushospitalstays/emergencyroomvisits/usesofanambulance?
6. Dothefamilyphysicianornurseaskyouwhetheryouknowwhototurntoincaseyouareinneedforasocialservice?
7. Dothefamilyphysicianornurseaskyouwhetheryouhaveprescriptionsforallthenecessary
medicationsandwhetheryouhaveboughttheprescribedmedicines?8. Canyouaffordallthemedicinesyouneed?9. Whatdoyoulikemostabouttheenhancedcaremanagementprogram?
10. Wouldyourecommendextendingtheenhancedcaremanagementpilottootherfamilyphysicianpractices?
11. Doyouhaveanyrecommendationsforfutureimprovementsrelatedtochroniccare
managementprovidedbyfamilydoctorsandnurses?
43
Annex3:AdditionalTablesandFiguresTableA1:Listof50ChronicConditionsConsideredintheRisk-StratificationAlgorithm
No Chroniccondition ICDCodesMetabolicTriad
1 Hypertension I10-I152 Lipidmetabolismdisorders E783 Diabetesmellitus E10-E14
TargetConditions4 COPD J40-J44,J475 Asthma J45,J46 6 Chronicischemicheartdisease I20,I25,I217 Cerebralischemia/chronicstroke I60-I64,I69,G458 AtrialFibrillationandFlutter I489 Cardiacinsufficiency I50 10 MoodDisorders F30-F3911 Dementia F00-F03,F05.1,G30,G31,R5412 SubstanceAbuse F11-F19,F55,Z71.5,Z81.3,Z81.413 AlcoholAbuse F10,Z71.4,Z81.114 Frailty(falls) R41.81, R54, W00, W01, W04, W05, W06, W07, W08, W10,
W18,W19,Z91.8115 Severehearingloss/
SeverevisionreductionH90, H91.0, H91.1, H91.3, H91.8, H91.9,H17-H18, H25-H28,H31,H33,H34.1-H34.2,H34.8-H34.9,H35-H36,H40,H43,H47,H54
OtherChronicConditions
16 Anemia D50-D53, D55-D58, D59.0-D59.2, D59.4-D59.9, D60.0, D60.8,D60.9,D61,D63-D64
17 Anxiety F40-F4118 Atherosclerosis/PAOD I65-I66,I67.2,I70,I73.919 Cardiacarrhythmias* I44-I45,I47,I4920 Cardiacvalvedisorders I34-I3721 Chroniccholecystitis/Gallstones K80,K81.122 Chronicgastritis/GERD K21, K25.4-K25.9 K26.4-K26.9 K27.4-K27.9 K28.4-K28.9 K29.2-
K29.923 Dizziness H81-H82,R4224 EatingDisorders F50,R63.025 Epilepsy G4026 Hemorrhoids I8427 Hypotension I9528 Intestinaldiverticulosis K5729 Jointarthrosis M15-M19
44
30 Liverdisease K70,K71.3-K71.5,K71.7,K72.1,K72.7,K72.9,K73-K74,K7631 Lowerlimbvaricosis I83,I87.232 Migraine/chronicheadache G43,G4433 Neuropathies G50-G6434 Obesity E6635 Osteoporosis M80-M8236 Parkinson’sdisease G20-G2237 Prostatichyperplasia N4038 Psoriasis L4039 PsychologicalDisorders F840 Purine/pyrimidinemetabolism
disorders/goutE79,M10
41 Rheumatoidarthritis/Chronicpolyarthritis
M05-M06,M79.0
42 Somatoformdisorders F4543 Thyroiddiseases E01-E05,E06.1-E06.3,E06.5,E06.9,E0744 Urinaryincontinence N39.3-N39.4,R3245 Urinarytractcalculi N20
ExclusionsduetoHighSeverity46 Cancer(ifactive/acute) C00-C14, C15-C26, C30-C39, C40-C41, C43-C44, C45-C49, C50,
C51-C58, C60-C63, C64-C68, C69-C72, C73-C75, C81-C96, C76-C80,C97,D00-D09,D37-D48
47 CongenitalDisorders Q0-Q848 RareDiseases F01.1,D21.9,D47.4,D48.9,D56.0,D82.4,E70.3,E75.5,E80.0,
E85.0,G47.3,H16.3,H49.8,I78.8,K90.8,M60.9,N04.1,R23.849 RenalFailure(Advanced) N18-N19,Z4950 Schizophrenia F20
*WithoutAtrialFibrillation(seeseparatecondition)
AdaptedandModifiedfrom:VandenBusscheetal.-Patternsofambulatorymedicalcareutilizationinelderlypatientswithspecialreferencetochronicdiseasesandmultimorbidity-ResultsfromaclaimsdatabasedobservationalstudyinGermany(2011).
45
FigureA1:ConstructingtheConsolidatedPatientListsforEachPracticeList
TableA2:OverviewofIndicatorsandTracersUsedin“TheStateofHealthCareIntegrationinEstonia”
46
References1. SarnakDO,RyanJ.HowHigh-NeedPatientsExperiencetheHealthCareSysteminNineCountries.
2016.2. StarfieldB.Primarycare:concept,evaluation,andpolicy.OxfordUniversityPress;1992.3. HaasLR,TakahashiPY,ShahND,StroebelRJ,BernardME,FinnieDM,etal.Risk-stratification
methodsforidentifyingpatientsforcarecoordination.AmJManagCare.2013;19(9):725–32.4. BodenheimerT,Berry-MillettR.CareManagementofPatientswithComplexHealthCareNeeds.
TheSynthesisProject.Princeton,NJ;2009.5. HongCS,SiegelAL,FerrisTG.Caringforhigh-need,high-costpatients:whatmakesfora
successfulcaremanagementprogram?IssueBrief.2014.6. PartnersHealthCareandHallmarkHealth’sResponsetotheHealthPolicyCommission’s
PreliminaryCMIRReport.2014.7. LoveT.CaseStudy:PeopleCenteredHealthCareinCanterbury,NewZealand.2015.8. VaillancourtS,ShahinI,AggarwalP,PomedliS,HaydenL,PusL,etal.Usingarchetypestodesign
servicesforhighusersofhealthcare.HealthcPap.2014;14(2):37–41.9. PowersB,ChaguturuS,FerrisT.OptimizingHigh-RiskCareManagement.JAMA.
2015;313(8):795–6.10. HaimeV,HongC,MandelL,MohtaN,IezzoniLI,FerrisTG,etal.Clinicianconsiderationswhen
selectinghigh-riskpatientsforcaremanagement.AmJManagCare.2015;21(10):e576-82.11. ChangH,BoydCM,LeffB,LemkeKW,BodycombeDP,WeinerJP.IdentifyingConsistentHigh-
costUsersinaHealthPlanComparisonofAlternativePredictionModels.MedCare.2016;54(9):852–9.
12. ChakravartyS,CantorJ.InformingtheDesignandEvaluationofSuperuserAccountingforRegression-to-the-Mean.MedCare.2016;0(0):1–8.
13. GrantRW,AshburnerJM,HongCS,ChangY,BarryMJ,AtlasSJ.DefiningPatientComplexityFromthePrimaryCarePhysician’sPerspectiveACohortStudy.AnnInternMed.2011;155(12).
14. HayesBSL,MccarthyD,RadleyD,ExpenditureM,SurveyP.TheImpactofaBehavioralHealthConditiononHigh-NeedAdults.2016.
15. BriggsT,BurdM,FransooR.IdentifyingHighUsersofHealthcareinBritishColumbia,AlbertaandManitoba.HealthcPap.2014;14(2):31–6.
16. HibbardJH,GreeneJ,SacksR,OvertonV,ParrottaCD.Addingameasureofpatientself-managementcapabilitytoriskassessmentcanimprovepredictionofhighcosts.HealthAff.2016;35(3):489–94.
17. PolancoNT,ZabaleguiIB,IrazustaIP,SolinisRN,CamaraMDrio.Buildingintegratedcaresystems:acasestudyofBidasoaIntegratedHealthOrganisation.IntJIntegrCare.2015;15.
18. LoebDF,BinswangerIA,CandrianC,BaylissEA.Primarycarephysicianinsightsintoatypologyofthecomplexpatientinprimarycare.AnnFamMed.2015;13(5):451–5.
19. HostetterM,KleinS.InFocus:SegmentingpopulationstoTailorServices,ImproveCare.2015.20. MccarthyD,RyanJ,KleinS.ModelsofCareforHigh-Need,High-CostPatients :AnEvidence
Synthesis.2015.21. BakerA,LeakP,RitchieLD,LeeAJ,FieldingS.Anticipatorycareplanningandintegration:A
primarycarepilotstudyaimedatreducingunplannedhospitalisation.BrJGenPract.2012;62(595):e113-20.
22. NelsonL.LessonsfromMedicare’sDemonstrationProjectsonDiseaseManagementandCareCoordination.Washington,DC;2012.
23. GoodwinN,DixonA,AndersonG,WodchisW.Providingintegratedcareforolderpeoplewith
47
complexneeds:Lessonsfromseveninternationalcasestudies.TheKingsFund.London;2014.24. HostetterM,KleinS,MccarthyD,HayesSL.GuidedCare :AStructuredApproachtoProviding
ComprehensivePrimaryCareforComplexPatients.2016.25. FernandopulleR.Learningtofly:buildingdenovomedicalhomepracticestoimprove
experience,outcomes,andaffordability.JAmbulCareManage.2013;36(2):121–5.26. FernandopulleR.RestoringHumanitytoHealth.JAmbulCareManage.2014;37(2):189–91.27. SullivanE,HufstaderT,ArabadjisS.BehavioralHealthandFinancingforSouthcentral
Foundation’sNukaSystemofCare(B).2015.28. SinskyCA,Willard-GraceR,SchutzbankAM,SinskyTA,MargoliusD,BodenheimerT.Insearchof
joyinpractice:Areportof23high-functioningprimarycarepractices.AnnFamMed.2013;11(3):272–8.
29. PolancoNTetal.,ZabaleguiIB,IrazustaIP,SolinísRN,DelRíoCámaraM.Buildingintegratedcaresystems:acasestudyofBidasoaIntegratedHealthOrganisation.IntJIntegrCare.2015;15(June):e026.
30. GottliebK.TheNukaSystemofCare:Improvinghealththroughownershipandrelationships.IntJCircumpolarHealth.2013;72:1–6.
31. HealthLeads.2016.32. DenisJ-L,CambourieuC,RoyD.TakingChargeofHigh-RiskandHigh-CostPatientsinthePublic
HealthcareSystem.HealthcPap.2014;14(2):42–7.33. WagnerEH,AustinBT,KorffMVon.OrganizingCareforPatientswithChronicIllness.MilbankQ.
1996;74(4):511–44.34. BurtJ,RickJ,BlakemanT,ProtheroeJ,RolandM,BowerP.Careplansandcareplanninginlong
termconditions :aconceptualmodel.PrimHealthCareResDev.2014;15(4):342–54.35. RubakS,SandbaekA,LauritzenT,ChristensenB.Motivationalinterviewing:asystematicreview
andmeta-analysis.BrJGenPract.2005;55(513):305–12.36. CollinsB.Intentionalwholehealthsystemredesign:SouthcentralFoundation’s“Nuka”systemof
care.KingsFund.2015;37. DrewesH,StruijsJ,BaanC.HowtheNetherlandsIsIntegratingHealthandCommunityServices.
NewEnglJCatal.2016;1–14.38. ComprehensivePrimaryCare.Mid-Year2015Snapshot.2015.39. ComplexCareManagementToolkitIntroduction.2012.40. LevineS,AdamsJ,AttawayK,DorrDA,KeungM,PopescuB,etal.PredictingtheFinancialRisks
ofSeriouslyIllPatients.2011.41. GoodellS,BodenheimerT,Berry-MillettR.CareManagementofPatientswithComplexHealth
CareNeeds.2009.42. OruetaJF,Nuño-SolinisR,MateosM,VergaraI,GrandesG,EsnaolaS.Predictiveriskmodellingin
theSpanishpopulation:across-sectionalstudy.BMCHealthServRes.2013;13:269.43. EvansJ.LeadingtheImplementationofHealthLinksinOntario.HealthcPap.2014;14(2):21–5.44. OntarioMinistryofHealthandLong-TermCare.TransformingOntario’sHealthCareSystem.
2016.45. HostetterM,KleinS,MccarthyD.HennepinHealth:ACareDeliveryParadigmforNewMedicaid
Beneficiaries.2016.46. HennepinHealthpre-andpost-analysisofEHRdatafor123patientshoused,covering2012-mid-
2014,comparing12monthsofpre-andpost-experienceforallpatients.47. LaheyHealth[Internet].Wikipedia.2016[cited2016Nov7].Availablefrom:
https://en.wikipedia.org/wiki/Lahey_Health48. GerardAnderson&ClaudiaSalzberg(2016),IdentifyingHighNeedHighCostIndividuals,Johns
HopkinsUniversity.
48
49. ReddyA,SessumsL,GuptaR,JinJ,DayT,FinkeB,BittonA.-RiskStratificationMethodsandProvisionofCareManagementServicesinComprehensivePrimaryCareInitiativePractices,AnnFamMed.2017Sep;15(5):451-454.
50. DionneS.Kringos,WienkeBoerma,JoukevanderZeeandPeterGroenewegen-Europe'sStrongPrimaryCareSystemsAreLinkedToBetterPopulationHealthButAlsoToHigherHealthSpendingHealthAffairs,32,no.4(2013):686-694.
51.Vaillancourt,S.(2014),UsingArchetypestoDesignServicesforHighUsersofHealthcare, Healthcarepapers.52. EstoniaEnhancedCareManagementPilot:ProviderSurvey:Modifiedfromthe2015
CommonwealthFundInternationalSurveyofPrimaryCarePhysiciansin10Nationshttp://www.commonwealthfund.org/interactives-and-data/surveys/international-health-policy-surveys/2015/2015-international-survey
53. EuropeanHealthforAllDatabase,2014.54. HealthStatisticsandHealthResearchDatabase,2015.55. http://www.wpro.who.int/topics/primary_health_care/en/
49