臨床試驗(yàn)的Meta分析幻燈片本課件PPT僅供大家學(xué)習(xí)使用學(xué)習(xí)完請(qǐng)自行刪除，謝謝！本課件PPT僅供大家學(xué)習(xí)使用學(xué)習(xí)完請(qǐng)自行刪除，謝謝！臨床試驗(yàn)的Meta分析幻燈片本課件PPT僅供大家學(xué)習(xí)1542-11-#2Meta-Analysis Alternatives? Occasionally Complementary? YesMeta-AnalysisCombinationofsimilarstudiesusingsimilarsubjectsandsimilartreatmentsandsimilaroutcomes542-11-#2Meta-Analysis Altern2542-11-#3Figure2

OddsRatiosand95%ConfidenceLimits

forVariousStudiesandaPooledEstimate542-11-#3Figure2

OddsRatios3542-11-#4NewMethodofAnalyzingHealthDataStirsDebate

byLawrenceK.AltmanIncreasinguseofacontroversialstatisticalmethodtoevaluatemedicaltherapiesandsurgicalproceduresisbeginningtoaffectprofoundlythecareofpregnantwomenandpatientswithcancer,heartdiseaseandmanyothercommonconditions.Themethod,knownasmeta-analysispromisestoplananincreasinglyimportantroleindetermininghealthrisks,environmentalhazardsandnationalpolicyonpaymentformedicalcare.Backerssaytechniquecandrawbig,reliableconclusionsfromsmall,inconsistentfindings.

Meta-analysisisatermderivedfromtheGreekmeaningananalysisthatismorecomprehensive.Thelargernumbersobtainedbycombiningstudiesprovideagreaterstatisticalpowerthananyoftheindividualstudies.Researchersareoftenabletodrawmorereliableinferencesornewconclusionsfromthecombinedresultsthanfromthesmallerstudiesthatmaybeinconclusiveindividually.Inearlierapplicationsofmeta-analysis,researchersevaluatedintelligencequotients,governmentsocialwelfareprogramsandmanyothertopics.Meta-analysishascometomedicinelate,but“itisnowundergoingaboominpopularity,”saidDr.ThomasC.Chalmers,adistinguishedphysicianoftheDepartmentofVeteransAffairsinBostonandapioneerinmethodology.Themethodinvolvesananalysisofpreviousanalyses.Itcombinestheresultsofawiderangeofexistingsmallerstudiesandthenappliesoneofseveralstatisticaltechniquestodiscovermorepreciselywhatisknownfrompreviousresearch.Itmayalsoproduceaunifiedresultfromdiverse,apparentlycontradictorystudies.Thetechniquehasalreadyshednewlightontheeffectivenessofmedicaltherapies.Althoughithasnot,initself,revolutionizedanymedicaltreatmentithashelpedclearawaytheconfusioncausedbystudieswithscatteredandapparentlyconflictingfindingsandhasstrengthenandconfirmedfindingsfromtraditionalclinicaltrials.NYTimes8/21/90542-11-#4NewMethodofAnalyzi4542-11-#5Reference:NIHProceedingsMethodologicIssuesinOverviewsofRandomizedClinicalTrialsNIHConferenceMay1986StatisticsinMedicineVol6,No.3,1987542-11-#5Reference:NIHProcee5542-11-#6WhatisthePurpose?a. Testingforatreatmenteffect(rejectingthenullhypothesis)b. Evaluatingasafetyissue(rareevents)c. Estimatingsizeoftreatmenteffectinsubgroupsd. Designofnewstudiese. Developpracticeguidelines542-11-#6WhatisthePurpose?a6542-11-#7IdealMetaAnalysisisRandomizedMulti-centerControlTrialSameprotocolSametreatmentSametypeofsubjectsSameoutcomemeasure542-11-#77542-11-#8IssuesinMetaAnalysisDifferencesAcrossStudiesin: a. Treatment b. ControlGroup/Population c. TimeSpan(Disease,BackgroundTherapy) d. OutcomeMeasuresPublicationBiasCompleteness/QualityofDataAccesstoData542-11-#8IssuesinMetaAnalys8542-11-#9WhatStudiesShouldBeIncluded?AllexistingstudiesAllpublishedstudies"Non-flawed"trialsOtherselectioncriteria542-11-#9WhatStudiesShouldB9542-11-#10Meta-Analysis:When?(1)RetrospectiveAnalysesTestTreatmentEffectWhen:DefinitiveanswernotyetavailableNomorestudieslikelyNeedtosalvageavailableresultsDevelopPracticeGuidelinesDesignNewStudies542-11-#10Meta-Analysis:When?10542-11-#11Meta-Analysis:When?(2)ProspectiveAnalysesNotrecommendedBettertodesigninadvancepropermulti-centertrial(s)542-11-#11Meta-Analysis:When?11542-11-#12Meta-AnalysisMethodologyNotNewCombiningp-values,Fisher(1948)AnalysisofVariance,Fisher(1938)Combining2x2TablesMantel-Haenszel(1959)Cochran(1954)542-11-#12Meta-AnalysisMethodo12542-11-#13OddsRatiomoreexplicitly

OR=ad/bcTCSaba+bFcdc+da+cb+d542-11-#13OddsRatiomoreexpli13542-11-#14MethodsofMeta-AnalysisCollapsingcanbemisleadingifthereisqualitativeinteraction.1.0CollapseDataRCT-1TCS155F8595OR=3.35RCT-2TCS515F9585OR=0.30CollapsedTC2020180180OR=1.0542-11-#14MethodsofMeta-Anal14542-11-#152. GraphicalSeeFigure95%CIforeachstudy(ad/bc)exp{±1.96(1/a+1/b+1/c+1/d)}MethodsofMetaAnalysis542-11-#152. GraphicalMethods15542-11-#16ApparenteffectsoffibrinolytictreatmentonmoralityintherandomisedtrialsofIVtreatmentofacutemyocardialinfarction.StatinMed7:890:1988.542-11-#16Apparenteffectsof16542-11-#17StatinMed6(3):320,1987.542-11-#17StatinMed6(3):3217542-11-#18Comparisonofmeta-analysisofmortalityin11RCTsandreinfarctionratesin10RCTsofi.v.streptokinasewithlargeco-operativestudy(GISSI).StatinMed6(3):320,1987.542-11-#18Comparisonofmeta-a18542-11-#19Comparisonofmeta-analysisof7smallRCTsofphenobarbitalinthetreatmentofneonatalintra-cranialhaemmorrhagewithonelargeco-operativestudy(3institutions).Endpointsaretotalinfantswithhaemmorrhageandtotalswithseverehaemorrhage(GradesIII-IV)only.StatinMed6(3):321,1987.542-11-#19Comparisonofmeta-a19542-11-#20OddsRatiosand95%ConfidenceLimits

forVariousStudiesandaPooledEstimate542-11-#20OddsRatiosand95%20542-11-#213. Blocking(Peto-MH)OverallEstimate LetO=ai E=Ei Ei=(ai+ci)(ai+bi) ni V=Vi Vi=(ai+ci))(bi+di)(ci+di)(ai+bi ni2(ni-1) Z=O-E

C PooledOR OR=exp{(O-E)/V} 95%CI=exp{(O-E)/V±1.96/}MethodsofMetaAnalysis542-11-#213. Blocking(Peto-MH21542-11-#224. AveragingP-values Fisher(1948) Pi=P-valueforithtrial Z=-2log(Pi)~2with2Ndf5. AveragingTestStatistics e.g.wi=niMethodsofMetaAnalysis542-11-#224. AveragingP-value22542-11-#23Meta-AnalysisExamplesCardiologyPostMITreatments (e.g.,beta-blockers,aspirin)ThrombolyticTherapy (e.g.,streptokinase)Anticoagulants542-11-#23Meta-AnalysisExampl23542-11-#24Registries/DatabasesByar(1980)BiometricsD'Ambrosia,Ellenberg(1980)BiometricsStarmeretal.(1980)BiometricsMantel(1983)StatisticsinMedicine542-11-#24Registries/Databases24542-11-#25Registries/DatabasesUseClinicalObservationalSeriesto:DescribeClinicalPracticeIdentifyRiskFactors"Evaluate"TreatmentHistoricalConcurrent542-11-#25Registries/Databases25542-11-#26DatabasesTreatmentEvaluationComparisonRequiresRiskFactorComparabilityMeasuredNotMeasuredorUnknownStatisticalModelsUsuallyNotAdequateAssociationvs.EstimationModelOnlyanApproximationSmallPortionofOutcomeExplained542-11-#26DatabasesTreatmentE26542-11-#27PotentialBiasesTimeTrends(DeclineinCHDDeath)AscertainmentChangesinDiagnosticCriteriaAvailabilityofTechnologySelectionBias542-11-#27PotentialBiasesTime27542-11-#28Compliance“Adjustment〞Compliance Clofibrate Placebo<80% 24.6% 28.2%>80% 15.0% 15.1%All 18.2% 19.4%CoronaryDrugProject(NEJM,1980)5YearMortality542-11-#28Compliance“Adjustme28542-11-#29RegistriesBiasinTreatmentEffect(Peto,Biomedicine,1978)TrialsofAnticoagulantTherapyDesign Studies Patients Effect Historical 18 900 50%Reduction Concurrent 8 3000 50%Reduction RCT 6 3000 20%Reduction 542-11-#29RegistriesBiasinTr29542-11-#30PTCAPTCARegistryTrackedandcomparedusageLeadtofurthertrialsNoPTCAvs.placeboTIMI-IIComparedimmediatevs.delayedPTCABARIComparesPTCAvs.CABG542-11-#30PTCAPTCARegistry30542-11-#31CABGCASSRCT(Circulation,1983)Comparisonofimmediatevs.delayedCABGCASSRegistry(JClinInv,1983)PrognosticvalueofAngiography542-11-#31CABGCASSRCT(Circul31542-11-#32Arboretum542-11-#32Arboretum32臨床試驗(yàn)的Meta分析幻燈片本課件PPT僅供大家學(xué)習(xí)使用學(xué)習(xí)完請(qǐng)自行刪除，謝謝！本課件PPT僅供大家學(xué)習(xí)使用學(xué)習(xí)完請(qǐng)自行刪除，謝謝！臨床試驗(yàn)的Meta分析幻燈片本課件PPT僅供大家學(xué)習(xí)33542-11-#34Meta-Analysis Alternatives? Occasionally Complementary? YesMeta-AnalysisCombinationofsimilarstudiesusingsimilarsubjectsandsimilartreatmentsandsimilaroutcomes542-11-#2Meta-Analysis Altern34542-11-#35Figure2

OddsRatiosand95%ConfidenceLimits

forVariousStudiesandaPooledEstimate542-11-#3Figure2

OddsRatios35542-11-#36NewMethodofAnalyzingHealthDataStirsDebate

byLawrenceK.AltmanIncreasinguseofacontroversialstatisticalmethodtoevaluatemedicaltherapiesandsurgicalproceduresisbeginningtoaffectprofoundlythecareofpregnantwomenandpatientswithcancer,heartdiseaseandmanyothercommonconditions.Themethod,knownasmeta-analysispromisestoplananincreasinglyimportantroleindetermininghealthrisks,environmentalhazardsandnationalpolicyonpaymentformedicalcare.Backerssaytechniquecandrawbig,reliableconclusionsfromsmall,inconsistentfindings.

Meta-analysisisatermderivedfromtheGreekmeaningananalysisthatismorecomprehensive.Thelargernumbersobtainedbycombiningstudiesprovideagreaterstatisticalpowerthananyoftheindividualstudies.Researchersareoftenabletodrawmorereliableinferencesornewconclusionsfromthecombinedresultsthanfromthesmallerstudiesthatmaybeinconclusiveindividually.Inearlierapplicationsofmeta-analysis,researchersevaluatedintelligencequotients,governmentsocialwelfareprogramsandmanyothertopics.Meta-analysishascometomedicinelate,but“itisnowundergoingaboominpopularity,”saidDr.ThomasC.Chalmers,adistinguishedphysicianoftheDepartmentofVeteransAffairsinBostonandapioneerinmethodology.Themethodinvolvesananalysisofpreviousanalyses.Itcombinestheresultsofawiderangeofexistingsmallerstudiesandthenappliesoneofseveralstatisticaltechniquestodiscovermorepreciselywhatisknownfrompreviousresearch.Itmayalsoproduceaunifiedresultfromdiverse,apparentlycontradictorystudies.Thetechniquehasalreadyshednewlightontheeffectivenessofmedicaltherapies.Althoughithasnot,initself,revolutionizedanymedicaltreatmentithashelpedclearawaytheconfusioncausedbystudieswithscatteredandapparentlyconflictingfindingsandhasstrengthenandconfirmedfindingsfromtraditionalclinicaltrials.NYTimes8/21/90542-11-#4NewMethodofAnalyzi36542-11-#37Reference:NIHProceedingsMethodologicIssuesinOverviewsofRandomizedClinicalTrialsNIHConferenceMay1986StatisticsinMedicineVol6,No.3,1987542-11-#5Reference:NIHProcee37542-11-#38WhatisthePurpose?a. Testingforatreatmenteffect(rejectingthenullhypothesis)b. Evaluatingasafetyissue(rareevents)c. Estimatingsizeoftreatmenteffectinsubgroupsd. Designofnewstudiese. Developpracticeguidelines542-11-#6WhatisthePurpose?a38542-11-#39IdealMetaAnalysisisRandomizedMulti-centerControlTrialSameprotocolSametreatmentSametypeofsubjectsSameoutcomemeasure542-11-#739542-11-#40IssuesinMetaAnalysisDifferencesAcrossStudiesin: a. Treatment b. ControlGroup/Population c. TimeSpan(Disease,BackgroundTherapy) d. OutcomeMeasuresPublicationBiasCompleteness/QualityofDataAccesstoData542-11-#8IssuesinMetaAnalys40542-11-#41WhatStudiesShouldBeIncluded?AllexistingstudiesAllpublishedstudies"Non-flawed"trialsOtherselectioncriteria542-11-#9WhatStudiesShouldB41542-11-#42Meta-Analysis:When?(1)RetrospectiveAnalysesTestTreatmentEffectWhen:DefinitiveanswernotyetavailableNomorestudieslikelyNeedtosalvageavailableresultsDevelopPracticeGuidelinesDesignNewStudies542-11-#10Meta-Analysis:When?42542-11-#43Meta-Analysis:When?(2)ProspectiveAnalysesNotrecommendedBettertodesigninadvancepropermulti-centertrial(s)542-11-#11Meta-Analysis:When?43542-11-#44Meta-AnalysisMethodologyNotNewCombiningp-values,Fisher(1948)AnalysisofVariance,Fisher(1938)Combining2x2TablesMantel-Haenszel(1959)Cochran(1954)542-11-#12Meta-AnalysisMethodo44542-11-#45OddsRatiomoreexplicitly

OR=ad/bcTCSaba+bFcdc+da+cb+d542-11-#13OddsRatiomoreexpli45542-11-#46MethodsofMeta-AnalysisCollapsingcanbemisleadingifthereisqualitativeinteraction.1.0CollapseDataRCT-1TCS155F8595OR=3.35RCT-2TCS515F9585OR=0.30CollapsedTC2020180180OR=1.0542-11-#14MethodsofMeta-Anal46542-11-#472. GraphicalSeeFigure95%CIforeachstudy(ad/bc)exp{±1.96(1/a+1/b+1/c+1/d)}MethodsofMetaAnalysis542-11-#152. GraphicalMethods47542-11-#48ApparenteffectsoffibrinolytictreatmentonmoralityintherandomisedtrialsofIVtreatmentofacutemyocardialinfarction.StatinMed7:890:1988.542-11-#16Apparenteffectsof48542-11-#49StatinMed6(3):320,1987.542-11-#17StatinMed6(3):3249542-11-#50Comparisonofmeta-analysisofmortalityin11RCTsandreinfarctionratesin10RCTsofi.v.streptokinasewithlargeco-operativestudy(GISSI).StatinMed6(3):320,1987.542-11-#18Comparisonofmeta-a50542-11-#51Comparisonofmeta-analysisof7smallRCTsofphenobarbitalinthetreatmentofneonatalintra-cranialhaemmorrhagewithonelargeco-operativestudy(3institutions).Endpointsaretotalinfantswithhaemmorrhageandtotalswithseverehaemorrhage(GradesIII-IV)only.StatinMed6(3):321,1987.542-11-#19Comparisonofmeta-a51542-11-#52OddsRatiosand95%ConfidenceLimits

forVariousStudiesandaPooledEstimate542-11-#20OddsRatiosand95%52542-11-#533. Blocking(Peto-MH)OverallEstimate LetO=ai E=Ei Ei=(ai+ci)(ai+bi) ni V=Vi Vi=(ai+ci))(bi+di)(ci+di)(ai+bi ni2(ni-1) Z=O-E

C PooledOR OR=exp{(O-E)/V} 95%CI=exp{(O-E)/V±1.96/}MethodsofMetaAnalysis542-11-#213. Blocking(Peto-MH53542-11-#544. AveragingP-values Fisher(1948) Pi=P-valueforithtrial Z=-2log(Pi)~2with2Ndf5. AveragingTestStatistics e.g.wi=niMethodsofMetaAnalysis542-11-#224. AveragingP-value54542-11-#55Meta-AnalysisExamplesCardiologyPostMITreatments (e.g.,beta-blockers,aspirin)ThrombolyticTherapy (e.g.,streptokinase)Anticoagulants542-11-#23Meta-AnalysisExampl55542-11-#56Registries/DatabasesByar(1980)BiometricsD'Ambrosia,Ellenberg(1980)BiometricsStarmeret