DEVELOP AND DEMONSTRATE A METHOD FOR CLASSIFYING HOME HEALTH PATIENTS TO PREDICT RESOURCE REQUIREMENTS AND TO MEASURE OUTCOMES by Virginia K. Saba, RN, EdD, FAAN Principal Investigator/Project Director Federal Project Officer: Margaret Coopey Home Health Care Classification Project Georgetown University, School of Nursing HCFA Cooperative Agreement No. 17-C-98983/3-01 February 1991 The statements contained in this report are solely those of the authors and do not necessarily reflect the views or policies of the Health Care FinancingAdministration. The contractor assumes responsibility for the accuracy and completeness of the information contained in this report. ACKNOWLEDGEMENTS This report representsthe results ofHome Hearth Care Classification project DceomnodnuscttreadtiaotntGhreanGteofrungdeetdowansaUnciovoepresirtaytiSvcehaogorleeofmeNnutrswiintgh.theItHweaarsthaCRaerseeFairncahncainndg Administration.Thiseffortwhichtookapproximatelythreeyearsprovidesnewinformation forthehomehealthcareindustry. Thisinnovativeprojectcouldneverhavebeensuccessfullycompletedwithoutthe dedicationoftheprojectstafflistedbelow.AtItsinception,noonerealizedthecomplexity oftheresearch. However,thisfinishedreportgivesevidenceofastrongcollaborative effortandcommittmenttowardachievingtheprojectgoals. Aspecialtributeisextended toDr.AlanZuckermanwhoprovidedendlessanalyticalsupport. SpecialappreciationisextendedtothePanelofExpertswhoguidedtheproject throughout Its life cycle and contributed to its conceptual design as well as the participatinghomehearthagenciespersonnelwhocooperatedandgaveoftheirtime willinglytoprovidetheneededdata. ManythankstothemanymembersoftheSchoolofNursingstaffwhocontributed totheday-to-dayprojectoperationincludingtwograduatenursingstudents,IreneReyes, andSheilaNveva,whoassistedwiththecodingofthenarrativetextofnursingdiagnoses andinterventions. Finally, acknowledgements would not be complete without recognitionoftheNationalAssociationforHomeCare(NAHC)fortheirsupport. VirginiaK.Saba,EdD,RN,FAAN PrincipalInvestigator/ProjectDirector AlanE.Zuckerman,MD EugeneLevine,PhD AnalyticalDirector ResearchDirector DavidM.Oatway,RN,MPH PatriciaO'Hare,DrPH,RN ProcessingCoordinator clinicalResearchDirector WilliamScanlon,PhD JenniferBoondas,MPH,RN Economist ResearchAssociate PeterT.Glenshaw JuiieannDimmick AndrewF.McLaughlin AdministrativeAssistants 5 TABLEOFCONTENTS EXECUTIVESUMMARY 1. INTRODUCTION 1 2. BACKGROUND 3 HistoricalPerspective 3 ImpactofMedicareLegislation InfluenceofDiagnosisRelatedGroups StatementoftheProblem 5 3. REVIEWOFTHELITERATURE 7 AssessmentInstruments ClassificationInstruments ClassificationStudies InstrumentstoMeasureOutcomes HomeHealthDatabases 12 4. RESEARCHANDDEMONSTRATIONMETHODOLOGY 13 ConceptualFramework 13 PilotStudy 14 PilotStudyConclusions ConceptualIssues StructureandContentofAssessmentInstrument 1 ReconciliationofResourceUseandRequirements StructureandContentofClassificationMethod StructureandContentof OutcomeMeasurementInstrument EnhancingtheApplicabilityoftheClassificationMethod SampleDesign 18 TABLEOFCONTENTS EXECUTIVESUMMARY 1. INTRODUCTION 1 2. BACKGROUND 3 HistoricalPerspective 3 ImpactofMedicareLegislation InfluenceofDiagnosisRelatedGroups StatementoftheProblem 5 3. REVIEWOFTHEUTERATURE 7 AssessmentInstruments ClassificationInstruments ClassificationStudies InstrumentstoMeasureOutcomes HomeHealthDatabases 12 4. RESEARCHANDDEMONSTRATIONMETHODOLOGY 13 ConceptualFramework 13 PilotStudy 14 PilotStudyConclusions ConceptualIssues 15 StructureandContentofAssessmentInstrument ReconciliationofResourceUseandRequirements StructureandContentofClassificationMethod StructureandContentof OutcomeMeasurementInstrument EnhancingtheApplicabilityoftheClassificationMethod SampleDesign 18 ComputerDatasets 3g AbstractFormDataset NursingServicesDataset VisitDatesDatasets NursingDiagnosesDataset : NursingInterventionsDataset AgencyInformationFormDataset PatientRecordsofCasesDataset 7. STATISTICALMETHODS 43 DescriptiveAnalyses 43 ExcludedCases NationalSampleAgenciesandCases SummaryMeasures MeasuresofCaseIntensity PatternsofHomeHealthServices PatientAssessmentRelationships PredictiveAnalyses 44 ExcludedCases LinearRegression AnalysisofVariance(ANOVA) StepwiseRegression LogisticRegression EvaluatingthePerformanceofPredictionModels 47 StatisticalSignificanceofaModel VarianceExplainedbyModelR-square MagnitudeoftheRegressionCoefficients StatisticalSignificanceoftheRegressionCoefficients PartialCorrelationsintheStepwiseModels DevelopmentofRegressionandCategoricalModels 48 MeasuresofResourceUse(DependentVariables) 49 NursingVisitsforFirstThirtyDays(NVRN30) WeightedTotalVisitsforFirstThirtyDays(NVWT30) NursingVisitsforTotalEpisode(NVRNALL) WeightedTotalVisitsforTotalEpisode(NVWTALL) LengthofEpisode/CohortMembership DependentVariablesNotUsedinthisAnalysis 51 VisitsbyHomeHealthAides VisitsbyOtherProfessionals UnweightedVisitsbyAllProviders ActualTotalCostofEpisode ) Socio-demographics 63 AgeandMaritalStatus SexandRace LivingArrangementsandAvailableCaregiver HousingandPetsinResidence .* ComprehensionLevelandCommunicationLevel MentalStatusandPrognosis AdmissionStatusandReferralSource DischargeStatus,DischargeDisposition,andReasonforDischarge GoalsandMotivation FunctionalStatus 67 FrequenciesofLevelsofPerformance Functionalstatusscales FrequenciesofFunctionalStatusScales NursingDiagnoses 70 FrequenciesofNursingDiagnoses CountsofNursingDiagnoses NursingDiagnosesPatterns NursingDiagnosisHomeHealthCare(HHC)Components NursingInterventions 74 FrequenciesofNursingInterventions CountsofNursingInterventions NursingInterventionsHomeHearthCare(HHC)Components MedicalDiagnosesorSurgicalProcedures 77 HomeHealthVisits 78 Number.ofHomehealthVisits 79 ProviderVisits 79 SkilledNursingVisits HomeHearthAideVisits OtherProfessionalProvidersVisits MeanVisits 80 LengthofEpisodesofCare 80 Cohorts 81 9. PREDICTIVEFINDINGS 151 Overview 151 DemographicRegressionModels ? 152 FunctionalStatusCategoricalModels 153 FunctionalStatus: RUGsADL FunctionalStatus: GUADL NursingDiagnosisComponentsRegressionModels 155 NumberofNursingDiagnosisComponentsCategoricalModels DischargeStatusPatternCategoricalModels NursingDiagnosisComponentsbyDischargeStatus RegressionModels NursingNIunmtberevrenotfioNnurCsoimnpgoInnetnertvsenRteigonrsesCsoimopnoMnoednetlssCategoricalModels 157 TypeofNursingInterventionPatternsRegressionModels Nursing Intervention Components by Four Types of Intervention Regression Models MedicalDiagnosisCategoricalModels NursingDiagnosisComponentsandDemographicsRegressionModels NursingInterventionComponentsandDemographicsRegressionModels NursingDiagnosisandNursingInterventionComponentsRegressionModels NursingDiagnosisandInterventionComponents,andDemographicsRegression Model 10. COHORTFINDINGS 215 CohortCategoricalModels 215 DemographicCohortCategoricalModels NursingDiagnosisComponentsCohortCategoricalModels NursingInterventionComponentsCohortCategoricalModels MedicalDiagnosisorSurgicalProcedureGroupsCohortCategoricalModels CohortFrequencies 217 DemographicCohortFrequencies NursingDiagnosisandNursingInterventionComponentsCohortFrequencies MedicalDiagnosisorSurgicalProcedureGroupsCohortFrequencies V Recommendations 251 REFERENCES 254 LISTOFAPPENDICES.^ „_ - 260 APPENDICES- _ 262 LISTOFTABLES 273 LISTOFFIGURES : 281 EXECUTIVESUMMARY Overview ThemajorobjectiveoftheGeorgetownUniversitySchoolofNursingHomeHealth CareClassificationresearchprojectwastodevelopamethodtoassessandclassifythe homehealthMedicarepatientsinordertopredicttheirneedfornursingandotherhome careservicesaswellasmeasureoutcomesofcare. Toaccomplishthisgoal,dataon actualresourceusewhichcouldbeobjectivelymeasuredwereusedtopredictresource requirements. Theresearchdesignwasbasedonaconceptualframeworkwhichevolvedfrom thestatementoftheproblem,reviewoftheliterature,andresultsofanearlierpilotstudy. However,themethodologicalresearchwasbasedontheassumptionthatbycollecting alargevolumeofdataonMedicarepatientsandresourcesusedfortheirhomehealth care, aclassification method could be designed to predictcare requirements. The research focused on five major conceptual issues: (a) assessment instrument, (b) resourceuseandrequirements,(c)classificationmethod,(d)outcomemeasures,and(e) applicabilityoftheclassificationmethod. Anationalsampleofhomehealthagencies,randomlystratifiedbystaffsize,type ofownership,andgeographiclocationwasselected. Ofthe5,880MedicareHCFA1986 certifiedhomehealthagencies646participatedinthestudy. Theywerefromeverystate inthenationandfromPuertoRicoandtheDistrictofColumbia. Thesampleagencies eachabstracted5to50Medicarepatientrecordsprovidingatotalof8,961cases. The database assembled representsthe largestcompilation ofpertinentinformationever collectedonhomehealthagenciesandpatients. Retrospectivedatawerecollected,usingaspeciallydesignedAbstractForm,on eachoftheMedicarepatient'sentireepisodeofhomehealthcarefromadmissionto discharge. Dataonallrelevantvariablesconsideredtobepossiblepredictorsofhome healthcarewerecollected. Thedataincludedawiderangeofvariables: demographic, functional status, medical diagnoses, surgical procedures, nursing diagnoses/patient problems,nursinginterventions/services,admissionanddischargedata,anddataonthe lengthoftheepisodeofcareandhomevisitsmadebyvariousproviders. Datawere collectedfromallformsinthepatientrecordandbyquestioningtheprimarynursefor datanotfoundinthepatientrecord. Hi TheAbstractFormsweredistributedtoagencieswhowereresponsibleforthe selectionofthesamplepatientsandfortheabstractingofthepatient'sentirerecord.The Abstract Form consisted not only of single or multiple variable questions, but also narrativetext. Twonursingcareitems-nursingdiagnosesandpatientproblemsand nursinginterventions-wereabstractedastextualstatementsofdatarecordedinthe progressnotesinthepatient'srecord. Sincetherewasnoagreementwhenthestudybeganonwhichspecificvariables predictedhomehealthvisits,thedatafromthesampleof8,961 patientrecordswere thoroughlyexaminedtoanalyzethestatisticalsignificanceofalternativeclassification methods. Thedescriptive analysisfocused primarily onfrequencydistributions and means. Summarymeasures,measuresofintensity,patternsofhomehealthservicesand patientassessmentrelationshipswereconducted. Thepredictiveanalysisusedlinear regression and analysis of variance which were the most useful statistical tests. Predictionsofresourcerequirementsweremadeusingtwotypesofstatisticalmodels: regressionmodelsandcategoricalmodels. Thestudyproductsincludeddescriptivefindingsontheuniverseofhomehealth casesand predictivefindingsonthevariablesthatmeasured resourcerequirements. Several schemes were also developed for coding nursing diagnoses and nursing interventions. Finallythestudyproducedathreecohortmodelbasedonthelengthof theepisodeofcareasanapproachforapreliminaryhomehealthcareclassification method. Themethod,withfurtherrefinement, couldformthebasisforaprospective paymentsystemforhomehealthcare. Thestudyfindingscanbegroupedintofourmajorareas: • Descriptive findings on characteristics of home health agencies and patients; • Predictivefindingsontherelationshipbetweenmeasuresofresourceuse andpredictorsofresourcerequirements; • Cohortfindingsbasedonlengthoftheepisodeofcareandnumberand typeofhomevisits; • PreliminaryHomeHealthClassificationMethod. IV