GRADE系统简介-Cochrane

TheGRADEApproachtoRatingtheQualityofEvidenceAVeryBasicIntroductionCochraneHIV/AIDSGroupUniversityofCalifornia,SanFranciscoMarch2010Background:ToomanysystemsGuidelinedevelopersuseabewilderingvarietyofsystemstoratethequalityoftheevidenceunderlyingtheirrecommendations.Asaresult,guidelineusershavefacedchallengesExampleRecommendationforuseoforalanticoagulationinpatientswithatrialfibrillationandrheumaticmitralvalvedisease:ORGANIZATIONEVIDENCERECOMMENDATIONAmericanHeartAssociation(AHA)BClassIAmericanCollegeofClinicalPharmacy(ACCP)A1ScottishIntercollegiateGuidelinesNetwork(SIGN)IVCThesameevidence–differentclassificationGradingofRecommendationsAssessment,DevelopmentandEvaluation(GRADE)AsystematicmethodofassessingthequalityofstudiesincludedinasystematicreviewanddevelopingrecommendationsorguidelinesbasedupontheevidenceGRADEWorkingGroupformedin2000Aim:todevelopacommon,transparentandsensiblesystemforgradingthequalityofevidenceandthestrengthofrecommendationsGRADEuptakeAgencyforHealthCareResearchandQuality(AHRQ)AllergicRhinitisinAsthmaGuidelines(ARIA)AmericanCollegeofChestPhysiciansAmericanCollegeofPhysiciansAmericanThoracicSocietyBritishMedicalJournalCanadianAgencyforDrugsandTechnologyinHealthClinicalEvidenceCochraneCollaborationEuropeanSocietyofThoracicSurgeonsInfectiousDiseasesSocietyofAmerica(IDSA)NationalInstituteClinicalExcellence(NICE)UpToDateWorldHealthOrganizationManyotherorganizationsQualityofevidence“Theextenttowhichonecanbeconfidentthatanestimateofeffectorassociationiscorrect.”Authorsofsystematicreviewsgradequalityofabodyofevidenceseparatelyforeachpatient-importantoutcome.Authorsofsystematicreviewsdonotgradetheoverallqualityofevidenceacrossoutcomes:thatistheroleofguidelinedevelopers.DeterminantsofqualityStudydesigniscriticaltojudgmentsaboutthequalityofevidence.Randomizedtrialsprovide,ingeneral,farstrongerevidencethanobservationalstudiesRigorousobservationalstudiesprovidestrongerevidencethanuncontrolledcaseseries.IntheGRADEapproachtoqualityofevidence:RandomizedtrialswithoutimportantlimitationsprovidehighqualityevidenceObservationalstudieswithoutspecialstrengthsprovidelowqualityevidenceLimitationsorspecialstrengthscan,however,modifythequalityoftheevidenceofbothrandomizedtrialsandobservationalstudies.TheGRADEsystemclassifiesthequalityofevidenceinoneoffourgrades:GRADEDEFINITIONHigh⊕⊕⊕⊕Furtherresearchisveryunlikelytochangeourconfidenceintheestimateofeffect.Moderate⊕⊕⊕Furtherresearchislikelytohaveanimportantimpactonourconfidenceintheestimateofeffectandmaychangetheestimate.Low⊕⊕Furtherresearchisverylikelytohaveanimportantimpactonourconfidenceintheestimateofeffectandislikelytochangetheestimate.VeryLow⊕Anyestimateofeffectisveryuncertain.GRADEprofiler(GRADEpro)SoftwaredevelopedbytheGRADEWorkingGroupThreekindsofevidencetables:—GRADEevidenceprofile—CochraneSummaryofFindings(SoF)table—CochraneOverviewofReviewstableSoFTableisthemostusefulformostCochranereviewsEvidenceProfilevs.SummaryofFindingsTableGRADEEvidenceprofile—Particularlyusefulforguidelinedevelopers—Presentsinformationaboutthebodyofevidence(e.g.numberofstudies),thejudgmentsabouttheunderlyingqualityofevidence,keystatisticalresults,andagradeforthequalityofevidenceforeachoutcomeSummaryofFindingsTable—Mostrelevantto(anddesignedforusein)Cochranereviews—Presentthemainfindingsofasystematicreviewinatransparentandsimpletabularformat—Provideskeyinformationconcerningthequalityofevidence,themagnitudeofeffectoftheinterventionsexamined,andthesumofavailabledataonmostimportantoutcomes.GRADEEvidenceProfileInsomeways,moreinstructivethanSoFTablesNotyetfeasibletopresentinCochranereviews(technical/formattingissues)OrientedmoreforpresentationstoguidelinedevelopersMayeventuallybepossibleforuseinCochranereviews.Inthemeantime,useSoFtablesforCochranereviews.GRADEevidenceprofileSoFTableThereare3keycomponentsofaSoFtable:—Informationaboutthereview—Summaryofthestatisticalresults—Gradeofthequalityofevidence.AuthorsofsystematicreviewscanuseGRADEprotocreateaSoFtableintwoways:—ImportmostofthedatafromaRevMan5file—EnterdatamanuallyRegardlessofhowyouenterthedata,youwillneedtoeditdataandmanuallyenteryourassessmentofthequalityofevidenceforeachoutcomeWhenSoFtableisfinished,youcanimportitintoRevMan.SoFTableSoFtablesincludesevencolumns:—Outcomes:listofimportantdesirableandundesirableoutcomes(upto7outcomes)—Assumedrisk:ameasureofthetypicalburdenoftheseoutcomes—Correspondingrisk:ameasureoftheburdenoftheoutcomesaftertheinterventionisapplied—Relativemagnitudeofeffect:fordichotomousoutcomesthetablewillusuallyprovideriskratio,oddsratio,orhazardratio—Numberofparticipantsandstudiesaddressingtheseoutcomes—Ratingofthequalityofevidenceforeachoutcome(whichmayvarybyoutcome)—Comments(ifneeded)Footnotes,fortransparencyaboutyourdecision-makingReviewswithmorethanonemaincomparisonrequireseparateSoFtablesforeachcomparison.SoFtableHowtomakeaSummaryofFindingsTable(over-simplified)OpenGRADEproCreateprofilesImportdatafromRevMan5intoGRADEproCreateSoFtable.Authormakesdecisio