GRADE系統(tǒng)簡介-Cochraneppt參考課件

TheGRADEApproach

toRatingtheQualityofEvidence

AVeryBasicIntroduction

CochraneHIV/AIDSGroupUniversityofCalifornia,SanFranciscoMarch20101Background:ToomanysystemsGuidelinedevelopersuseabewilderingvarietyofsystemstoratethequalityoftheevidenceunderlyingtheirrecommendations.Asaresult,guidelineusershavefacedchallenges

2Example Recommendationforuseoforalanticoagulationinpatientswithatrialfibrillationandrheumaticmitralvalvedisease:ORGANIZATIONEVIDENCERECOMMENDATIONAmericanHeartAssociation(AHA)BClassIAmericanCollegeofClinicalPharmacy(ACCP)A1ScottishIntercollegiateGuidelinesNetwork(SIGN)IVCThesameevidence–differentclassification3GradingofRecommendationsAssessment,DevelopmentandEvaluation(GRADE)AsystematicmethodofassessingthequalityofstudiesincludedinasystematicreviewanddevelopingrecommendationsorguidelinesbasedupontheevidenceGRADEWorkingGroupformedin2000Aim:todevelopacommon,transparentandsensiblesystemforgradingthequalityofevidenceandthestrengthofrecommendations4GRADEuptakeAgencyforHealthCareResearchandQuality(AHRQ)AllergicRhinitisinAsthmaGuidelines(ARIA)AmericanCollegeofChestPhysiciansAmericanCollegeofPhysiciansAmericanThoracicSocietyBritishMedicalJournalCanadianAgencyforDrugsandTechnologyinHealthClinicalEvidenceCochraneCollaborationEuropeanSocietyofThoracicSurgeonsInfectiousDiseasesSocietyofAmerica(IDSA)NationalInstituteClinicalExcellence(NICE)UpToDateWorldHealthOrganizationManyotherorganizations5Qualityofevidence“Theextenttowhichonecanbeconfidentthatanestimateofeffectorassociationiscorrect.”Authorsofsystematicreviewsgradequalityofabodyofevidenceseparatelyforeachpatient-importantoutcome.Authorsofsystematicreviewsdonotgradetheoverallqualityofevidenceacrossoutcomes:thatistheroleofguidelinedevelopers.6DeterminantsofqualityStudydesigniscriticaltojudgmentsaboutthequalityofevidence.Randomizedtrialsprovide,ingeneral,farstrongerevidencethanobservationalstudiesRigorousobservationalstudiesprovidestrongerevidencethanuncontrolledcaseseries.IntheGRADEapproachtoqualityofevidence:RandomizedtrialswithoutimportantlimitationsprovidehighqualityevidenceObservationalstudieswithoutspecialstrengthsprovidelowqualityevidenceLimitationsorspecialstrengthscan,however,modifythequalityoftheevidenceofbothrandomizedtrialsandobservationalstudies.78TheGRADEsystemclassifiesthequalityofevidenceinoneoffourgrades:GRADEDEFINITIONHigh⊕⊕⊕⊕Furtherresearchisveryunlikelytochangeourconfidenceintheestimateofeffect.Moderate⊕⊕⊕Furtherresearchislikelytohaveanimportantimpactonourconfidenceintheestimateofeffectandmaychangetheestimate.Low⊕⊕Furtherresearchisverylikelytohaveanimportantimpactonourconfidenceintheestimateofeffectandislikelytochangetheestimate.VeryLow⊕

Anyestimateofeffectisveryuncertain.9GRADEprofiler(GRADEpro)SoftwaredevelopedbytheGRADEWorkingGroupThreekindsofevidencetables:GRADEevidenceprofileCochraneSummaryofFindings(SoF)tableCochraneOverviewofReviewstableSoFTableisthemostusefulformostCochranereviews10EvidenceProfilevs.

SummaryofFindingsTableGRADEEvidenceprofileParticularlyusefulforguidelinedevelopersPresentsinformationaboutthebodyofevidence(e.g.numberofstudies),thejudgmentsabouttheunderlyingqualityofevidence,keystatisticalresults,andagradeforthequalityofevidenceforeachoutcomeSummaryofFindingsTableMostrelevantto(anddesignedforusein)CochranereviewsPresentthemainfindingsofasystematicreviewinatransparentandsimpletabularformatProvideskeyinformationconcerningthequalityofevidence,themagnitudeofeffectoftheinterventionsexamined,andthesumofavailabledataonmostimportantoutcomes.11GRADEEvidenceProfileInsomeways,moreinstructivethanSoFTablesNotyetfeasibletopresentinCochranereviews(technical/formattingissues)OrientedmoreforpresentationstoguidelinedevelopersMayeventuallybepossibleforuseinCochranereviews.Inthemeantime,useSoFtablesforCochranereviews.12GRADE

evidence

profile13SoFTableThereare3keycomponentsofaSoFtable:InformationaboutthereviewSummaryofthestatisticalresultsGradeofthequalityofevidence.AuthorsofsystematicreviewscanuseGRADEprotocreateaSoFtableintwoways:ImportmostofthedatafromaRevMan5fileEnterdatamanuallyRegardlessofhowyouenterthedata,youwillneedtoeditdataandmanuallyenteryourassessmentofthequalityofevidenceforeachoutcomeWhenSoFtableisfinished,youcanimportitintoRevMan.14SoFTableSoFtablesincludesevencolumns:Outcomes:listofimportantdesirableandundesirableoutcomes(upto7outcomes)Assumedrisk:ameasureofthetypicalburdenoftheseoutcomesCorrespondingrisk:ameasureoftheburdenoftheoutcomesaftertheinterventionisappliedRelativemagnitudeofeffect:fordichotomousoutcomesthetablewillusuallyprovideriskratio,oddsratio,orhazardratioNumberofparticipantsandstudiesaddressingtheseoutcomesRatingofthequalityofevidenceforeachoutcome(whichmayvarybyoutcome)Comments(ifneeded)Footnotes,fortransparencyaboutyourdecision-makingReviewswithmorethanonemaincomparisonrequireseparateSoFtablesforeachcomparison.15SoF

table16Howtomakea

SummaryofFindingsTable

(over-simplified)OpenGRADEproCreateprofilesImportdatafromRevMan5intoGRADEproCreateSoFtable.AuthormakesdecisionsaboutinformationtopresentandGRADEstheevidenceExporttablefromGRADEproandimportintoRevMan517Working

in

GRADEpro18GRADEprohasanexcellent“Help”file19GRADEResourcesTheCochraneHIV/AIDSGroupmaintainsapagewithmanyarticlesandotherresourcesaboutGRADE:/Cochrane/GRADE20Referencesforthispresentation:GuyattGH,OxmanAD,VistGE,KunzR,Falck-YtterY,Alonso-CoelloP,SchünemannHJ;GRADEWorkingGroup.GRADE:anemergingconsensusonratingqualityofevidenceandstrengthofrecommendations.BMJ.2008Apr26;336(7650):924-6.GuyattGH,OxmanAD,KunzR,Falck-Ytte