FrederickL.Brancati,MD,MHSProfessorofMedicine&EpidemiologyDirector,DivisionofGeneralInternalMedicineOslerJournalClub2006VisitHopkinsGIMat/gimProspectiveCohortStudies.FrederickL.Brancati,MD,MHS1..2BackgroundPhysicalactivitylowerCVDriskDHHSrecommendslife-longpursuitsSportsdifferinsustainabilityCVDbenefitsofindividualsportsuncertain.BackgroundPhysicalactivity3TheJohnsHopkinsPrecursorsStudyOver1300students(mainlywhitemen)fromtheJHUSOMClassesof1948-64.Baselinedatacollectedinpersoninmedicalschool.Follow-updatacollectedbyyearlymailedquestionnairesthereafter..TheJohnsHopkinsPrecu4CarolineThomas,MDTheJohnsHopkinsPrecursorsStudy.CarolineThomas,MD.5Hypothesis:TennisabilityinyouthpredictslowerCVDriskinmiddleageDesign:ProspectivecohortstudySetting:JohnsHopkinsPrecursorsStudyParticipants:1019malemedicalstudentsDataCollection:Extensiveinterviewandphysicalassessmentatbaseline(early20s);annualmailedfollow-upquestionnairesOutcome:IncidentCVD,includingMI,CHD,CABGorPTCA,hypertensiveheartdisease,heartfailure,&cerebrovasculardiseaseAnalysis:Kaplan-Meier,CoxmodelsOutline.Hypothesis:Tennisabilityin6AssessmentofSportsAbilityHowwouldyourateyouroverallabilityintennis(golf,football,baseball,basketball)duringandbeforemedicalschool?NoabilityPoororfairabilityGoodorexcellentabilityNodataonfrequency,intensity,orsubsequentparticipation.AssessmentofSportsAbilityHo7Results.Results.8..9..10..11..12..13Conclusions/ImplicationsSelf-describedtennisabilityinyoungadulthoodpredictslowerCVDriskinmiddleageAssociationoftennistolowerriskisGraded(i.e.dose-response)IndependentofmanypossibleconfoundersSpecifictotennis(ashypothesized)SuggestspromotionoftennisasameanstoreduceCVDrisk.Conclusions/ImplicationsSelf14StrengthsProspectivedesignLong-termfollow-upMultiavariateanalysisBlindedassessmentofCVD.StrengthsProspectivedesign.15WeaknessesObservationalstudiescan’tprovecausalityResidualconfoundingislikelyAssessmentofexposurewassuboptimalAbility,notactivitySinglepoint,notrepeatedmeasuresSelf-assessed,notobjectiveSamplelimitsgeneralizability.WeaknessesObservationalstudie16DiscussionPointsWhat’sspecialaboutacohortstudy?Whatarecommonobstacles?Canitbeusedforhousestaffresearch?Caniteverbesufficienttochangepractice?Howdocohortstudiesrelatetooutcomesresearch?.DiscussionPointsWhat’sspecia17TaxonomyofDesignsRandomizedControlledTrialProspectiveCohortStudyCase-ControlStudyCross-SectionalStudyOtherDesignsQuasi-ExperimentalEcologicCaseReport.TaxonomyofDesignsRandomized18Thebasicfightingunitwasacohort,composedofsixcenturies(480menplus6centurions).Thelegionitselfwascomposedoftencohorts,andthefirstcohorthadmanyextramen—theclerks,engineers,andotherspecialistswhodidnotusuallyfight—andtheseniorcenturionofthelegion,theprimipilus,or“numberonejavelin.”.Thebasicfightingunitwasa19pro·spec·tive

Pronunciation:pr&-'spek-tivalso'pr?-",prO-',pr?-'

Function:adjective

Date:circa1699

1

:relatingtooreffectiveinthefuture

2a

:likelytocomeabout:EXPECTED<theprospectivebenefitsofthislaw>b

:likelytobeorbecome<aprospectivemother>.pro·spec·tive

Pronunciation:20“Prospective”inEpidemiologyClearlydefinedcohort(group,sample)ofpersonsatriskfollowedthroughtimeDataregardingexposures(riskfactors,predictors)collectedpriortodataonoutcomes(endpoints)Research-gradedatacollectionmethodsusedforpurposeoftestinghypothesis(?).“Prospective”inEpidemiologyC21DiagramofHypothetical6-YearCohortStudytoIdentifyRiskFactorsforFacialAcneinTeenagers100012-year-oldswithoutacne50018-year-oldswithoutacne90015-year-oldswithoutacne50withAcne300withAcne5moved10noanswer35refused10moved40noanswer48refused2deaths350incidentcasesofacneover6years6-yrFollow-upRate=850/1000=85%IncidenceRateofAcne=350/5475PY=63.9per1000PY.DiagramofHypothetical6-Year22WhyDoACohortStudy?GetincidencedataStudyarangeofpossibleriskfactorsEstablishtemporalsequenceGetrepresentativedataPrepareforrandomizedcontrolledtrialEstablisharesearchempire.WhyDoACohortStudy?Getinci23TypesofCohortsOccupational(e.g.Asbestosworkers)Convenience(e.g.Precursors,Nurses)Geographic(e.g.Framingham,ARIC)DiseaseorProcedureNaturalHistory(e.g.Syncope,Lupus)OutcomesResearch(e.g.Dialysis,Cataracts).TypesofCohortsOccupational(24SourcesofCohortDataClinicVisitsLaboratoryAssaysInterviewPhysicalExaminationImagingPhysiologictestsHomevisitsMailedmaterialsTelephoneInterviewMedicalRecordsAdministrativeDataMedicareMedicaidManagedCareVeteransAdminBirthRecordsDeathCertificatesSpecimenBank.SourcesofCohortDataClinicV25WilliamCastelli,MDTheFraminghamHeartStudy.WilliamCastelli,MDTheFramin26RecentlyPublishedStudiesfromtheJohnsHopkinsPrecursorsStudyCoronaryDisease -Anger,Depression,Gout,-SportsAbilityType2Diabetes -Bloodpressure,AdiposityHypertension -CoffeeKneeOsteoarthritis -KneeinjuryDepression -InsomniaOutcomeExposure.RecentlyPublishedStudiesfro27WhatMightExplainObservedRelationshipofTennisAbilitytoHeartDiseaseRisk?TennisprotectsagainstheartdiseaseMenwholiketoplaytennisaredifferentThinnerHealthierLifestylesHigherSocioeconomicStatusMenwhoplaytenniswellaredifferentTaller,ThinnerGreaterCardiovascularFitnessChance(typeIerror)–Needsconfirmation.WhatMightExplainObservedRe28PlaysTennisPlaysTennisWellSustainedActivityThruMidlifeLoweradiposity,GreaterFitnessLowerBP,LowerLDL,HigherHDLLowerRiskofCHDHypotheticalCausalPathwayHealthierMenChooseTennisHealthierMenPlayTennisWellPotentialConfounders.PlaysTennisPlaysTennisWellS29GreyHairHigherRiskofCHDHypotheticalCausalPathwayOlderAgePotentialConfounders.GreyHairHigherRiskofCHDHyp30ChallengesinCohortStudiesPossiblylongdurationPossiblylargesamplesizeNeedtorecruitpeople“atrisk”Dropouts,Deaths,OtherlossesConcernaboutresidualconfoundingMultiplecomparisonsTypeIerror.ChallengesinCohortStudiesPo31HowtoExploitCohortDesignWhenTimeisShort&MoneyisScarceAnalyzeexistingdatafromanotherstudyPiggy-backontoon-goingstudyChoosehospital-basedcohortChooseshort-termoutcomeConsideradministrativedataConsiderpublic-usedataConsidernon-concurrentdesign.HowtoExploitCohortDesignW32..33..34FrederickL.Brancati,MD,MHSProfessorofMedicine&EpidemiologyDirector,DivisionofGeneralInternalMedicineOslerJournalClub2006VisitHopkinsGIMat/gimProspectiveCohortStudies.FrederickL.Brancati,MD,MHS35..36BackgroundPhysicalactivitylowerCVDriskDHHSrecommendslife-longpursuitsSportsdifferinsustainabilityCVDbenefitsofindividualsportsuncertain.BackgroundPhysicalactivity37TheJohnsHopkinsPrecursorsStudyOver1300students(mainlywhitemen)fromtheJHUSOMClassesof1948-64.Baselinedatacollectedinpersoninmedicalschool.Follow-updatacollectedbyyearlymailedquestionnairesthereafter..TheJohnsHopkinsPrecu38CarolineThomas,MDTheJohnsHopkinsPrecursorsStudy.CarolineThomas,MD.39Hypothesis:TennisabilityinyouthpredictslowerCVDriskinmiddleageDesign:ProspectivecohortstudySetting:JohnsHopkinsPrecursorsStudyParticipants:1019malemedicalstudentsDataCollection:Extensiveinterviewandphysicalassessmentatbaseline(early20s);annualmailedfollow-upquestionnairesOutcome:IncidentCVD,includingMI,CHD,CABGorPTCA,hypertensiveheartdisease,heartfailure,&cerebrovasculardiseaseAnalysis:Kaplan-Meier,CoxmodelsOutline.Hypothesis:Tennisabilityin40AssessmentofSportsAbilityHowwouldyourateyouroverallabilityintennis(golf,football,baseball,basketball)duringandbeforemedicalschool?NoabilityPoororfairabilityGoodorexcellentabilityNodataonfrequency,intensity,orsubsequentparticipation.AssessmentofSportsAbilityHo41Results.Results.42..43..44..45..46..47Conclusions/ImplicationsSelf-describedtennisabilityinyoungadulthoodpredictslowerCVDriskinmiddleageAssociationoftennistolowerriskisGraded(i.e.dose-response)IndependentofmanypossibleconfoundersSpecifictotennis(ashypothesized)SuggestspromotionoftennisasameanstoreduceCVDrisk.Conclusions/ImplicationsSelf48StrengthsProspectivedesignLong-termfollow-upMultiavariateanalysisBlindedassessmentofCVD.StrengthsProspectivedesign.49WeaknessesObservationalstudiescan’tprovecausalityResidualconfoundingislikelyAssessmentofexposurewassuboptimalAbility,notactivitySinglepoint,notrepeatedmeasuresSelf-assessed,notobjectiveSamplelimitsgeneralizability.WeaknessesObservationalstudie50DiscussionPointsWhat’sspecialaboutacohortstudy?Whatarecommonobstacles?Canitbeusedforhousestaffresearch?Caniteverbesufficienttochangepractice?Howdocohortstudiesrelatetooutcomesresearch?.DiscussionPointsWhat’sspecia51TaxonomyofDesignsRandomizedControlledTrialProspectiveCohortStudyCase-ControlStudyCross-SectionalStudyOtherDesignsQuasi-ExperimentalEcologicCaseReport.TaxonomyofDesignsRandomized52Thebasicfightingunitwasacohort,composedofsixcenturies(480menplus6centurions).Thelegionitselfwascomposedoftencohorts,andthefirstcohorthadmanyextramen—theclerks,engineers,andotherspecialistswhodidnotusuallyfight—andtheseniorcenturionofthelegion,theprimipilus,or“numberonejavelin.”.Thebasicfightingunitwasa53pro·spec·tive

Pronunciation:pr&-'spek-tivalso'pr?-",prO-',pr?-'

Function:adjective

Date:circa1699

1

:relatingtooreffectiveinthefuture

2a

:likelytocomeabout:EXPECTED<theprospectivebenefitsofthislaw>b

:likelytobeorbecome<aprospectivemother>.pro·spec·tive

Pronunciation:54“Prospective”inEpidemiologyClearlydefinedcohort(group,sample)ofpersonsatriskfollowedthroughtimeDataregardingexposures(riskfactors,predictors)collectedpriortodataonoutcomes(endpoints)Research-gradedatacollectionmethodsusedforpurposeoftestinghypothesis(?).“Prospective”inEpidemiologyC55DiagramofHypothetical6-YearCohortStudytoIdentifyRiskFactorsforFacialAcneinTeenagers100012-year-oldswithoutacne50018-year-oldswithoutacne90015-year-oldswithoutacne50withAcne300withAcne5moved10noanswer35refused10moved40noanswer48refused2deaths350incidentcasesofacneover6years6-yrFollow-upRate=850/1000=85%IncidenceRateofAcne=350/5475PY=63.9per1000PY.DiagramofHypothetical6-Year56WhyDoACohortStudy?GetincidencedataStudyarangeofpossibleriskfactorsEstablishtemporalsequenceGetrepresentativedataPrepareforrandomizedcontrolledtrialEstablisharesearchempire.WhyDoACohortStudy?Getinci57TypesofCohortsOccupational(e.g.Asbestosworkers)Convenience(e.g.Precursors,Nurses)Geographic(e.g.Framingham,ARIC)DiseaseorProcedureNaturalHistory(e.g.Syncope,Lupus)OutcomesResearch(e.g.Dialysis,Cataracts).TypesofCohortsOccupational(58SourcesofCohortDataClinicVisitsLaboratoryAssaysInterviewPhysicalExaminationImagingPhysiologictestsHomevisitsMailedmaterialsTelephoneInterviewMedicalRecordsAdministrativeDataMedicareMedicaidManagedCareVeteransAdminBirthRecordsDeathCertificatesSpecimenBank.SourcesofCohortDataClinicV59WilliamCastelli,MDTheFraminghamHeartStudy.WilliamCastelli,MDTheFramin60RecentlyPublishedStudiesfromtheJohnsHopkinsPrecursorsStudyCoronaryDisease -Anger,Depression,Gout,-SportsAbilityType2Diabetes -Bloodpressure,AdiposityHypertension -CoffeeKneeOsteoarthritis -KneeinjuryDepression -InsomniaOutcomeExposure.