使用GRADEpro评价证据质量.ppt

使用GRADEpro评价证据质量,武汉大学公共卫生学院流行病教研室2011年11月,GRADE系统简介,GRADE（GradingofRecommendationsAssessment,DevelopmentandEvaluation）是由2000年建立的GRADE工作组提出的一套评级系统。GRADE系统使用易于理解的方式评价证据质量和推荐等级，目前已被世界卫生组织（WHO）、Cochrane协作网等一批著名机构所采用。,GRADE系统较之其他系统的优势,由一个具有广泛代表性的国际指南制定小组制定明确界定了证据质量和推荐强度清楚评价了不同治疗方案的重要结局对不同级别证据的升级与降级有明确、综合的标准从证据到推荐全过程透明明确承认价值观和意愿就推荐意见的强弱，分别从临床医生、患者、政策制定者角度做了明确实用的诠释适用于制作系统评价、卫生技术评估及指南,证据质量（GRADEpro）,为达到透明和简化的目标，GRADE系统将证据质量分为高、中、低、极低4级。一些使用GRADE系统的组织甚至把低和极低归为一级。,推荐等级,使用GRADE系统时，指南小组用“强推荐”表示他们确信相关的干预措施利大于弊。用“弱推荐”表示干预措施有可能利大于弊，但他们把握不大。,GRADE中的证据质量和推荐等级,关键步骤一导入RevMan文件,关键步骤二判断结局的重要程度,GRADE系统建议采用9级分级判断结局的重要程度:13：不重要结局（NOTIMPORTANT）46：重要结局（IMPORTANT）79：关键结局（CRITICAL）,关键步骤三选择研究设计类型,按研究设计方法划分随机试验质量等级最高，为高；观察性研究（包括队列研究、病例-对照研究、病例-历史对照研究）质量等级低，为低；病例报告、病例分析等其他研究质量等级非常低，为极低。,关键步骤四评价证据质量,基于研究设计可因5条降级因素和4条升级因素上升或下降级别。,决定证据质量的因素,可能降低证据质量的因素研究的局限性结果不一致间接证据精确度不够发表偏倚可能增加证据质量的因素效应值很大可能的混杂因素会降低疗效剂量-效应关系,研究设计与证据质量,高：为不降级的随机试验和升2级的观察性研究；中：为降级的随机试验和升1级的观察性研究；低：降2级的随机试验和没有升降级的观察性研究；非常低：降3级的随机试验、降1级的观察性研究、病例分析/病例报告。,帮助教程要点,研究的局限性（Riskofbias）,这些局限性包括（RCTs）隐藏分组缺失盲法缺失（特别是结局指标为主观性指标且对其的评估极易受偏倚影响时）失访过多未进行意向性分析、观察到疗效就过早终止试验或未报道结果（通常是未观察到疗效的一些研究）,Toratestudylimitations:,IfyouthinkanylimitationswerenegligiblechoosenoIfyouthinktherewereseriouslimitationschooseseriousthiswilldowngradethequalityofevidenceforthisoutcomeby1levelIfyouthinktherewereveryseriouslimitationschooseveryseriousthiswilldowngradethequalityofevidenceforthisoutcomeby2levels,结果不一致（Inconsistency）,Inconsistencymayarisefromdifferencesin:populations(e.g.drugsmayhavelargerrelativeeffectsinsickerpopulations)interventions(e.g.largereffectswithhigherdrugdoses)outcomes(e.g.diminishingtreatmenteffectwithtime)差异可能源于人群（如药物对重症人群的疗效可能相对显著）、干预措施（如较高药物剂量会使疗效更显著）或结局指标（如随时间推移疗效降低）。,I2,OnesetofcriteriawouldsaythatanI2oflessthan40%islow3060%maybemoderate5090%maybesubstantial75100%isconsiderable,Torateinconsistency:,IfyouthinkanyinconsistencywasnegligiblechoosenoIfyouthinktherewasseriousinconsistencychooseseriousthiswilldowngradethequalityofevidenceforthisoutcomeby1levelIfyouthinktherewasveryseriousinconsistencychooseveryseriousthiswilldowngradethequalityofevidenceforthisoutcomeby2levels,间接证据（Indirectness),Therearetwotypesofindirectness:1.Indirectcomparison2.Indirectpopulationinterventioncomparatoroutcome,Torateindirectness:,IfyouthinktheevidenceisdirectchoosenoIfyouhaveseriousdoubtsaboutdirectnesschooseseriousthiswilldowngradetheevidenceforthisoutcomeby1levelIfyouhaveveryseriousdoubtsaboutdirectnesschooseveryseriousthiswilldowngradetheevidenceforthisoutcomeby2levels,精确度不够（Imprecision）,当研究纳入的患者和观察事件相对较少而致可信区间较宽时，指南小组将降低该研究的证据质量。Athresholdrule-of-thumbvalueTotalnumberofeventsislessthan300（dichotomous）Totalpopulationsizeislessthan400（continuous）,Torateimprecision:,IfyouthinktheresultswereprecisechoosenoIftherewasseriousimprecisionchooseseriousthiswilldowngradethequalityofevidenceforthisoutcomeby1levelIftherewasveryseriousimprecisionchooseveryseriousthiswilldowngradethequalityofevidenceforthisoutcomeby2levels,发表偏倚（Publicationbias）,Publicationbiasariseswheninvestigatorsfailtoreportstudiestheyhaveundertaken(typicallythosethatshownoeffect).Methodstodetectthepossibilityofpublicationbiasinsystematicreviewsexist,althoughauthorsofthereviewsandguidelinepanelsmustoftenguessaboutthelikelihoodofpublicationbias.Aprototypicalsituationthatshouldelicitsuspicionofpublicationbiasoccurswhenpublishedevidenceislimitedtoasmallnumberoftrials,allofwhichareshowingbenefitsofthestudiedintervention.若研究者未能发表研究（通常是阴性结果的研究）时，证据质量亦会减弱。典型情况是当公开的证据仅局限于少数试验而这些试验全部由企业赞助，此时不得不质疑存在发表偏倚。,FunnelPlot,Torateprobabilityofthepublicationbias:,IfyouthinkthereisnoevidenceofpublicationbiaschooseunlikelyIfthereishighprobabilityofpublicationbiaschooselikelythiswilldowngradethequalityofevidenceforthisoutcomeby1levelIfthereisveryhighprobabilityofpublicationbiaschooseverylikelythiswilldowngradethequalityofevidenceforthisoutcomeby2levels,效应值很大（Largeeffect）,疗效越显著，证据越有力。,Toratemagnitudeoftheeffect:,Iftheeffectwasnotlarge(RRbetween0.5and2.0)choosenoIftheeffectwaslarge(RReither2.0or2or5.0or5or0.2thiswillupgradethequalityofevidenceforthisoutcomeby2levels,可能的混杂因素会降低疗效（Plausibleconfoundingwouldchangetheeffect）,Onoccasion,allplausibleconfoundingfromobservationalstudiesorrandomisedtrialsmaybeworkingtoreducethedemonstratedeffectorincreasetheeffectifnoeffectwasobserved.Forexample,ifonlysickerpatientsreceiveanexperimentalinterventionorexposure,yettheystillfarebetter,itislikelythattheactualinterventionorexposureeffectislargerthanthedatasuggest.,Toratetheeffectofallplausibleresidualconfounding:,IfthereisnoevidencethattheinfluenceofallplausibleconfoundingwouldreduceademonstratedeffectorsuggestaspuriouseffectwhenresultsshownoeffectchoosenoIfthereisevidencethattheinfluenceofallplausibleconfoundingwouldreduceademonstratedeffectorsuggestaspuriouseffectwhenresultsshownoeffectchooseyesthiswillupgradethequalityofevidenceforthisoutcomeby1level,剂量-效应关系（Dose-responsegradi）,Thepresenceofadose-responsegradientmayincreaseourconfidenceinthefindingsofobservationalstudiesandtherebyincreasethequalityofevidence.,Toratethepresenceofdose-responsegradient:,Ifthereisnoevidenceofdose-responsegradientchoosenoIfthereisevidenceofdose-responsegradientchooseyesthiswillupgradethequalityofevidenceforthisoutcomeby1level,评级流程,关键步骤五完善数据,关键步骤六生成表格,EvidenceProfilevs.SummaryofFindingsTable,GRADEEvidenceprofileParticularlyusefulforguidelinedevelopersPresentsinformationaboutthebodyofevidence(e.g.numberofstudies),thejudgmentsabouttheunderlyingqualityofevidence,keystatisticalresults,andagradeforthequalityofevidenceforeachoutcomeSummaryofFindingsTableMostrelevantto(anddesignedforusein)CochranereviewsPresentthemainfindingsofasystematicreviewinatransparentandsimpletabularformatProvideskeyinformationconcerningthequalityofevidence,themagnitudeofeffectoftheinterventionsexamined,andthesumofavailabledataonmostimportantoutcomes.,SoFTable,Thereare3keycomponentsofaSoFtable:InformationaboutthereviewSummaryofthestatisticalresultsGradeofthequalityofevidence.,关键步骤七导出文件到RevMan,Meta分析（RevMan）,证据等级（GRADEpro）,推荐等级,系统综述流程图,