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