GRADEpro軟件教程.ppt

使用GRADEpro評價證據(jù)質(zhì)量,武漢大學(xué)公共衛(wèi)生學(xué)院流行病教研室2011年11月,GRADE系統(tǒng)簡介,GRADE（GradingofRecommendationsAssessment,DevelopmentandEvaluation）是由2000年建立的GRADE工作組提出的一套評級系統(tǒng)。GRADE系統(tǒng)使用易于理解的方式評價證據(jù)質(zhì)量和推薦等級，目前已被世界衛(wèi)生組織（WHO）、Cochrane協(xié)作網(wǎng)等一批著名機(jī)構(gòu)所采用。,GRADE系統(tǒng)較之其他系統(tǒng)的優(yōu)勢,由一個具有廣泛代表性的國際指南制定小組制定明確界定了證據(jù)質(zhì)量和推薦強(qiáng)度清楚評價了不同治療方案的重要結(jié)局對不同級別證據(jù)的升級與降級有明確、綜合的標(biāo)準(zhǔn)從證據(jù)到推薦全過程透明明確承認(rèn)價值觀和意愿就推薦意見的強(qiáng)弱，分別從臨床醫(yī)生、患者、政策制定者角度做了明確實(shí)用的詮釋適用于制作系統(tǒng)評價、衛(wèi)生技術(shù)評估及指南,證據(jù)質(zhì)量（GRADEpro）,為達(dá)到透明和簡化的目標(biāo)，GRADE系統(tǒng)將證據(jù)質(zhì)量分為高、中、低、極低4級。一些使用GRADE系統(tǒng)的組織甚至把低和極低歸為一級。,推薦等級,使用GRADE系統(tǒng)時，指南小組用“強(qiáng)推薦”表示他們確信相關(guān)的干預(yù)措施利大于弊。用“弱推薦”表示干預(yù)措施有可能利大于弊，但他們把握不大。,GRADE中的證據(jù)質(zhì)量和推薦等級,關(guān)鍵步驟一導(dǎo)入RevMan文件,關(guān)鍵步驟二判斷結(jié)局的重要程度,GRADE系統(tǒng)建議采用9級分級判斷結(jié)局的重要程度:13：不重要結(jié)局（NOTIMPORTANT）46：重要結(jié)局（IMPORTANT）79：關(guān)鍵結(jié)局（CRITICAL）,關(guān)鍵步驟三選擇研究設(shè)計(jì)類型,按研究設(shè)計(jì)方法劃分隨機(jī)試驗(yàn)質(zhì)量等級最高，為高；觀察性研究（包括隊(duì)列研究、病例-對照研究、病例-歷史對照研究）質(zhì)量等級低，為低；病例報告、病例分析等其他研究質(zhì)量等級非常低，為極低。,關(guān)鍵步驟四評價證據(jù)質(zhì)量,基于研究設(shè)計(jì)可因5條降級因素和4條升級因素上升或下降級別。,決定證據(jù)質(zhì)量的因素,可能降低證據(jù)質(zhì)量的因素研究的局限性結(jié)果不一致間接證據(jù)精確度不夠發(fā)表偏倚可能增加證據(jù)質(zhì)量的因素效應(yīng)值很大可能的混雜因素會降低療效劑量-效應(yīng)關(guān)系,研究設(shè)計(jì)與證據(jù)質(zhì)量,高：為不降級的隨機(jī)試驗(yàn)和升2級的觀察性研究；中：為降級的隨機(jī)試驗(yàn)和升1級的觀察性研究；低：降2級的隨機(jī)試驗(yàn)和沒有升降級的觀察性研究；非常低：降3級的隨機(jī)試驗(yàn)、降1級的觀察性研究、病例分析/病例報告。,幫助教程要點(diǎn),研究的局限性（Riskofbias）,這些局限性包括（RCTs）隱藏分組缺失盲法缺失（特別是結(jié)局指標(biāo)為主觀性指標(biāo)且對其的評估極易受偏倚影響時）失訪過多未進(jìn)行意向性分析、觀察到療效就過早終止試驗(yàn)或未報道結(jié)果（通常是未觀察到療效的一些研究）,Toratestudylimitations:,IfyouthinkanylimitationswerenegligiblechoosenoIfyouthinktherewereseriouslimitationschooseseriousthiswilldowngradethequalityofevidenceforthisoutcomeby1levelIfyouthinktherewereveryseriouslimitationschooseveryseriousthiswilldowngradethequalityofevidenceforthisoutcomeby2levels,結(jié)果不一致（Inconsistency）,Inconsistencymayarisefromdifferencesin:populations(e.g.drugsmayhavelargerrelativeeffectsinsickerpopulations)interventions(e.g.largereffectswithhigherdrugdoses)outcomes(e.g.diminishingtreatmenteffectwithtime)差異可能源于人群（如藥物對重癥人群的療效可能相對顯著）、干預(yù)措施（如較高藥物劑量會使療效更顯著）或結(jié)局指標(biāo)（如隨時間推移療效降低）。,I2,OnesetofcriteriawouldsaythatanI2oflessthan40%islow3060%maybemoderate5090%maybesubstantial75100%isconsiderable,Torateinconsistency:,IfyouthinkanyinconsistencywasnegligiblechoosenoIfyouthinktherewasseriousinconsistencychooseseriousthiswilldowngradethequalityofevidenceforthisoutcomeby1levelIfyouthinktherewasveryseriousinconsistencychooseveryseriousthiswilldowngradethequalityofevidenceforthisoutcomeby2levels,間接證據(jù)（Indirectness),Therearetwotypesofindirectness:1.Indirectcomparison2.Indirectpopulationinterventioncomparatoroutcome,Torateindirectness:,IfyouthinktheevidenceisdirectchoosenoIfyouhaveseriousdoubtsaboutdirectnesschooseseriousthiswilldowngradetheevidenceforthisoutcomeby1levelIfyouhaveveryseriousdoubtsaboutdirectnesschooseveryseriousthiswilldowngradetheevidenceforthisoutcomeby2levels,精確度不夠（Imprecision）,當(dāng)研究納入的患者和觀察事件相對較少而致可信區(qū)間較寬時，指南小組將降低該研究的證據(jù)質(zhì)量。Athresholdrule-of-thumbvalueTotalnumberofeventsislessthan300（dichotomous）Totalpopulationsizeislessthan400（continuous）,Torateimprecision:,IfyouthinktheresultswereprecisechoosenoIftherewasseriousimprecisionchooseseriousthiswilldowngradethequalityofevidenceforthisoutcomeby1levelIftherewasveryseriousimprecisionchooseveryseriousthiswilldowngradethequalityofevidenceforthisoutcomeby2levels,發(fā)表偏倚（Publicationbias）,Publicationbiasariseswheninvestigatorsfailtoreportstudiestheyhaveundertaken(typicallythosethatshownoeffect).Methodstodetectthepossibilityofpublicationbiasinsystematicreviewsexist,althoughauthorsofthereviewsandguidelinepanelsmustoftenguessaboutthelikelihoodofpublicationbias.Aprototypicalsituationthatshouldelicitsuspicionofpublicationbiasoccurswhenpublishedevidenceislimitedtoasmallnumberoftrials,allofwhichareshowingbenefitsofthestudiedintervention.若研究者未能發(fā)表研究（通常是陰性結(jié)果的研究）時，證據(jù)質(zhì)量亦會減弱。典型情況是當(dāng)公開的證據(jù)僅局限于少數(shù)試驗(yàn)而這些試驗(yàn)全部由企業(yè)贊助，此時不得不質(zhì)疑存在發(fā)表偏倚。,FunnelPlot,Torateprobabilityofthepublicationbias:,IfyouthinkthereisnoevidenceofpublicationbiaschooseunlikelyIfthereishighprobabilityofpublicationbiaschooselikelythiswilldowngradethequalityofevidenceforthisoutcomeby1levelIfthereisveryhighprobabilityofpublicationbiaschooseverylikelythiswilldowngradethequalityofevidenceforthisoutcomeby2levels,效應(yīng)值很大（Largeeffect）,療效越顯著，證據(jù)越有力。,Toratemagnitudeoftheeffect:,Iftheeffectwasnotlarge(RRbetween0.5and2.0)choosenoIftheeffectwaslarge(RReither2.0or2or5.0or5or0.2thiswillupgradethequalityofevidenceforthisoutcomeby2levels,可能的混雜因素會降低療效（Plausibleconfoundingwouldchangetheeffect）,Onoccasion,allplausibleconfoundingfromobservationalstudiesorrandomisedtrialsmaybeworkingtoreducethedemonstratedeffectorincreasetheeffectifnoeffectwasobserved.Forexample,ifonlysickerpatientsreceiveanexperimentalinterventionorexposure,yettheystillfarebetter,itislikelythattheactualinterventionorexposureeffectislargerthanthedatasuggest.,Toratetheeffectofallplausibleresidualconfounding:,IfthereisnoevidencethattheinfluenceofallplausibleconfoundingwouldreduceademonstratedeffectorsuggestaspuriouseffectwhenresultsshownoeffectchoosenoIfthereisevidencethattheinfluenceofallplausibleconfoundingwouldreduceademonstratedeffectorsuggestaspuriouseffectwhenresultsshownoeffectchooseyesthiswillupgradethequalityofevidenceforthisoutcomeby1level,劑量-效應(yīng)關(guān)系（Dose-responsegradi）,Thepresenceofadose-responsegradientmayincreaseourconfidenceinthefindingsofobservationalstudiesandtherebyincreasethequalityofevidence.,Toratethepresenceofdose-responsegradient:,Ifthereisnoevidenceofdose-responsegradientchoosenoIfthereisevidenceofdose-responsegradientchooseyesthiswillupgradethequalityofevidenceforthisoutcomeby1level,評級流程,關(guān)鍵步驟五完善數(shù)據(jù),關(guān)鍵步驟六生成表格,EvidenceProfilevs.SummaryofFindingsTable,GRADEEvidenceprofileParticularlyusefulforguidelinedevelopersPresentsinformationaboutthebodyofevidence(e.g.numberofstudies),