TableTheCochraneCollaboration’stoolforassessingriskofbias 偏倚类型 选择偏倚 随机序列的产生 判断指标 评价员的判断 分配隐藏 足够详细的描述用于生成分配序列的方法,以评估产生的分组是否具有可比性。 足够详细的描述隐藏分配序列的方法,以决定干预的分配在纳入之前或纳入过程中是否可见 生成随机序列不充分,发生选择偏倚 分配前分配隐藏不充分发生选择偏倚 实施偏倚 实施者和参与者双盲应对每个主要结局进行评估(或分类结局)? 测量偏倚 结局评估中的盲法每如果有,描述对参与者和实施者行参与者和实施者了解干盲法,避免其了解干预信息的所有预的相关信息导致实施措施。提供任何与所实施的盲法是偏倚 否有效地相关信息。 个主要结局均应评估(或分类结局) 失访偏倚 不全结局数据每个主如果有,描述对结局者行盲法,避结局评估者了解分配的免其了解自己所接受的干预信息干预措施将导致测量偏的所有措施。提供任何与所实施的倚 盲法是否有效地相关信息。 描述每个主要结局数据的完整性,不全结局数据的数量,要结局均应评估(或分包括分析中的自然缺失和排除。这性质,处理方式导致失类结局) 些缺失数据是否报告,在各个干预访偏倚 组的数目(并与总样本量比较),数据缺失以及重新纳入分析的原因 发表偏倚 Selectivereporting. 说明如何审查选择性报道结局的选择性报道结局导致发可能性,以及审查结果 表偏倚 其它偏倚 其它偏倚来源 说明不包括在上述偏倚中的其它不包括在上述各项中的重要偏倚 偏倚 如果特定的问题或条目事先在计划书中指出,应对每一项说明 TableCriteriaforjudgingriskofbiasinthe‘Riskofbias’assessmenttool
随机序列的产生 随机序列产生不充分导致选择偏倚 判断为低风险的标准 研究者描述随机序列产生过程譬如: 参考随机数字表 使用计算机随机数字生成器 扔硬币 洗牌的卡片和信封 掷骰子 抽签 最小化 ?*最小化,可实现无随机元素,被认为相当于是随机的。 判断为高风险的标准 研究者描述序列的产生使用的是非随机的方法。通常是系统的非随机方法,例如: 通过奇偶或出生日期产生序列 通过入院日期产生序列 通过类似住院号或门诊号产生序列? 相对于上面提到的系统方法,其它非随机的方法少见的多,也更明显。通常包括对参与者进行判断或非随机的方法,例如: 临床医生判断如何分配 参与者判断如何分配 基于实验室检查或系列测试的结果分配 基于干预的可获取性进行分配 偏倚风险不清楚的判断标准 没有足够的信息判断随机序列的产生存在高风险或低风险 分配隐藏 分配前不充足的分配隐藏导致选择偏倚 低风险判断标准 参与者以及纳入参与者的研究者因以下掩盖分配的方法或相当的方法,事先不了解分配情况 中心分配(包括电话,网络,药房控制随机) 相同外形的顺序编号的药物容器; 顺序编号、不透明、密封的信封 高风险判断标准 参与者以及纳入参与者的研究者可能事先知道分配,因而引入选择偏倚,譬如基于如下方法的分配: 使用摊开的随机分配表(如随机序列清单) 分发信封但没有合适的安全保障(如透明、非密封、非顺序编号) 交替或循环 出生日期 病历号 其它明确的非隐藏过程 风险未知 没有足够信息判断为低风险或高风险。通常因分配隐藏的方法未描述或描述不充分。例如描述为使用信封分配,但为描述信封是否透明?密封?顺序编号? 对参与者和实施者的盲法 因参与者和实施者了解干预情况而导致实施偏倚 偏倚低风险标准 任何如下标准: 无盲法或盲法不充分,但系统评价员判断结局不太可能受到缺乏盲法的影响 参与者和主要实施者均实施可靠的盲法,且盲法不太可能被打破 偏倚高风险标准 任何如下标准: 无盲法或盲法不充分,但系统评价员判断结局很可能受到缺乏盲法的影响 尝试对关键的参与者和实施者行盲法,但盲法很可能被打破,结局很可能受到缺乏盲法的影响 风险未知 任何如下标准: 没有足够信息判断为低风险或高风险 研究未描述此情况 对结局评价实施盲法 结局评价者了解干预分配信息将导致测量偏倚 偏倚低风险标准 任何如下标准: 无盲法或盲法不充分,但系统评价员判断结局不太可能受到缺乏盲法的影响 参与者和主要实施者均实施可靠的盲法,且盲法不太可能被打破 高风险判断标准 任何如下标准: 无盲法或盲法不充分,但系统评价员判断结局很可能受到缺乏盲法的影响 尝试对关键的参与者和实施者行盲法,但盲法很可能被打破,结局很可能受到缺乏盲法的影响 风险未知 任何如下标准: 没有足够信息判断为低风险或高风险 研究未描述此情况 结局数据不完整 不全结局数据的数量,性质,处理方式导致失访偏倚 偏倚低风险标准 任何如下标准: 无缺失数据 缺失数据的产生不大可能与真实结局相关(对于生存数据,删失不大可能引入偏倚) 缺失数据的数目在各干预组相当,且各组缺失原因类似 对二分类变量,与观察事件的发生风险相比,缺失比例不足以影响预估的干预效应 对连续性结局数据,缺失数据的合理效应规模(均数差或标准均数差)不会大到影响观察的效应规模; 缺失的数据用合适的方法进行估算 高风险判断标准 任何如下标准: 缺失数据的产生很大可能与真实结局相关,缺失数据的数目及缺失原因在各干预组相差较大 对二分类变量,与观察事件的发生风险相比,缺失比例足以影响预估的干预效应 对连续性结局数据,缺失数据的合理效应规模(均数差或标准均数差)足以影响观察的效应规模; 意向治疗分析中存在实际干预措施与随机分配的干预相违背的情况 对缺失数据进行简单的不合适的估算 风险未知 任何如下标准: 没有报道缺失或排除的情况,无法判断高风险或低风险(如未说明随机的数量,未提供数据缺失的原因) 研究未描述此情况 选择性发表 选择性发表导致发表偏倚 偏倚低风险标准 任何如下标准: 实验的计划书可获取,系统评价感兴趣的所有首要或次要结局均按计划书预先说明的方式报道 实验计划书不可得,但很明显发表的报告包括所有的结局,包括预先说明的结局(这种性质的有说服力的文字可能少见) 高风险判断标准 任何如下标准: 不是所有的预先说明的首要结局均被报道 一个或多个首要结局为采用预先说明的测量方法、分析方法或数据子集来报道 系统评价感兴趣的一个或多个首要结局报道不全,以至于不能纳入meta分析 研究未报道此研究应当包含的主要关键结局 风险未知 没有足够信息判断高风险或低风险,貌似大部分研究会被分为此类 OTHERBIAS? 不包括在以上五种的其它偏倚 偏倚低风险标准 研究应未引入其它来源的偏倚 高风险判断标准 至少有一种重要的偏倚风险,例如: 具有与特殊试验设计相关的潜在偏倚来源 或被指欺诈 或其它问题 风险未知 可能存在偏倚风险,但存在以下两种中的一种 没有足够信息评估是否存在其它重要的偏倚风险 没有足够的证据认为发现的问题会引入偏倚 TablePossibleapproachforsummaryassessmentsoftheriskofbiasforeachimportantoutcome(acrossdomains)withinandacrossstudies Riskofbias Lowriskofbias. 解释 合理的偏倚不太可能严重改变结果 Unclearriskofbias. 合理的偏倚会对结果产生一定的怀疑 Highriskofbias. 偏倚严重削弱结果的可信度 对单个研究 每一类偏倚均为低风险 一类或多类偏倚风险未知 一类或多类偏倚为高风险 对多个研究整体 绝大多数信息均来自偏倚低风险的研究 绝大多数信息均来自偏倚低风险或风险未知的研究 来自高偏倚风险研究的信息比例足以影响结果的解释 英文:
TableTheCochraneCollaboration’stoolforassessingriskofbias
Domain Supportforjudgement Reviewauthors’judgement Selectionbias. RandomsequenceDescribethemethodusedtogeneratetheallocationsequenceinsSelectionbias(biasgeneration. ufficientdetailtoallowanassessmentofwhetheritshouldproduceedallocationtointercomparablegroups. ventions)duetoinadequategenerationofarandomisedsequence. AllocationconcealDescribethemethodusedtoconcealtheallocationsequenceinsufSelectionbias(biasment. ficientdetailtodeterminewhetherinterventionallocationscouldhaedallocationtointervebeenforeseeninadvanceof,orduring,enrolment. ventions)duetoinadequateconcealmentofallocationspriortoassignment. Performancebias. BlindingofparticipDescribeallmeasuresused,ifany,toblindstudyparticipantsandpPerformancebiasdantsandpersonnelersonnelfromknowledgeofwhichinterventionaparticipantreceivuetoknowledgeofthAssessmentsshouled.Provideanyinformationrelatingtowhethertheintendedblindineallocatedinterventdbemadeforeachmgwaseffective. ainoutcome(orclassofoutcomes).? Detectionbias. ionsbyparticipantsandpersonnelduringthestudy. BlindingofoutcomDescribeallmeasuresused,ifany,toblindoutcomeassessorsfroDetectionbiasdueteassessmentAssemknowledgeofwhichinterventionaparticipantreceived.Provideoknowledgeofthealssmentsshouldbemanyinformationrelatingtowhethertheintendedblindingwaseffectlocatedinterventionadeforeachmainoutive. come(orclassofoutcomes). Attritionbias. sbyoutcomeassessors. IncompleteoutcoDescribethecompletenessofoutcomedataforeachmainoutcomAttritionbiasduetoamedataAssessmene,includingattritionandexclusionsfromtheanalysis.Statewhethemount,natureorhantsshouldbemadeforrattritionandexclusionswerereported,thenumbersineachintervdlingofincompleteoeachmainoutcome(entiongroup(comparedwithtotalrandomizedparticipants),reasoutcomedata. orclassofoutcomes)nsforattrition/exclusionswherereported,andanyre-inclusionsin.? Reportingbias. analysesperformedbythereviewauthors. SelectivereportingStatehowthepossibilityofselectiveoutcomereportingwasexamiReportingbiasduet. nedbythereviewauthors,andwhatwasfound. oselectiveoutcomereporting. Otherbias. OthersourcesofbiStateanyimportantconcernsaboutbiasnotaddressedintheotherBiasduetoproblemas. domainsinthetool. snotcoveredelsewIfparticularquestions/entrieswerepre-specifiedinthereview’sprhereinthetable. otocol,responsesshouldbeprovidedforeachquestion/entry.
TableCriteriaforjudgingriskofbiasinthe‘Riskofbias’assessmenttool
RANDOMSEQUENCEGENERATION Selectionbias(biasedallocationtointerventions)duetoinadequategenerationofarandomisedsequence. CriteriaforajTheinvestigatorsdescribearandomcomponentinthesequencegenerationprocesssuchas: udgementof‘Lowrisk’ofbias. Referringtoarandomnumbertable; Usingacomputerrandomnumbergenerator; Cointossing; Shufflingcardsorenvelopes; Throwingdice; Drawingoflots; Minimization*. ?*Minimizationmaybeimplementedwithoutarandomelement,andthisisconsideredtobeequivalenttobeingrandom. CriteriaforthTheinvestigatorsdescribeanon-randomcomponentinthesequencegenerationprocess.Usuejudgementally,thedescriptionwouldinvolvesomesystematic,non-randomapproach,forexample: of‘Highrisk’ofbias. Sequencegeneratedbyoddorevendateofbirth; Sequencegeneratedbysomerulebasedondate(orday)ofadmission; Sequencegeneratedbysomerulebasedonhospitalorclinicrecordnumber. Othernon-randomapproacheshappenmuchlessfrequentlythanthesystematicapproachesmentionedaboveandtendtobeobvious.??Theyusuallyinvolvejudgementorsomemethodofnon-randomcategorizationofparticipants,forexample: Allocationbyjudgementoftheclinician; Allocationbypreferenceoftheparticipant; Allocationbasedontheresultsofalaboratorytestoraseriesoftests; Allocationbyavailabilityoftheintervention. CriteriaforthInsufficientinformationaboutthesequencegenerationprocesstopermitjudgementof‘Lowrisk’ejudgementor‘Highrisk’. of?‘Unclearrisk’ofbias. ALLOCATIONCONCEALMENT? Selectionbias(biasedallocationtointerventions)duetoinadequateconcealmentofallocationspriortoassignment. CriteriaforajParticipantsandinvestigatorsenrollingparticipantscouldnotforeseeassignmentbecauseoneudgementof‘ofthefollowing,oranequivalentmethod,wasusedtoconcealallocation: Lowrisk’ofbias. Centralallocation(includingtelephone,web-basedandpharmacy-controlledrandomization); Sequentiallynumbereddrugcontainersofidenticalappearance; Sequentiallynumbered,opaque,sealedenvelopes. CriteriaforthParticipantsorinvestigatorsenrollingparticipantscouldpossiblyforeseeassignmentsandthusejudgementintroduceselectionbias,suchasallocationbasedon: of‘Highrisk’ofbias. Usinganopenrandomallocationschedule(e.g.alistofrandomnumbers); Assignmentenvelopeswereusedwithoutappropriatesafeguards(e.g.ifenvelopeswereunsealedornonopaqueornotsequentiallynumbered); Alternationorrotation; Dateofbirth; Caserecordnumber; Anyotherexplicitlyunconcealedprocedure. CriteriaforthInsufficientinformationtopermitjudgementof‘Lowrisk’or‘Highrisk’.Thisisusuallythecaseiftheejudgementmethodofconcealmentisnotdescribedornotdescribedinsufficientdetailtoallowadefinitejudgof?‘Unclearrement–forexampleiftheuseofassignmentenvelopesisdescribed,butitremainsunclearwhethisk’ofbias. erenvelopesweresequentiallynumbered,opaqueandsealed. BLINDINGOFPARTICIPANTSANDPERSONNEL Performancebiasduetoknowledgeoftheallocatedinterventionsbyparticipantsandpersonnelduringthestudy. CriteriaforajAnyoneofthefollowing: udgementof‘Lowrisk’ofbias. Noblindingorincompleteblinding,butthereviewauthorsjudgethattheoutcomeisnotlikelytobeinfluencedbylackofblinding; Blindingofparticipantsandkeystudypersonnelensured,andunlikelythattheblindingcouldhavebeenbroken. CriteriaforthAnyoneofthefollowing: ejudgementof‘Highrisk’ofbias. Noblindingorincompleteblinding,andtheoutcomeislikelytobeinfluencedbylackofblinding; Blindingofkeystudyparticipantsandpersonnelattempted,butlikelythattheblindingcouldhavebeenbroken,andtheoutcomeislikelytobeinfluencedbylackofblinding. CriteriaforthAnyoneofthefollowing: ejudgementof?‘Unclearrisk’ofbias. Insufficientinformationtopermitjudgementof‘Lowrisk’or‘Highrisk’; Thestudydidnotaddressthisoutcome. BLINDINGOFOUTCOMEASSESSMENT Detectionbiasduetoknowledgeoftheallocatedinterventionsbyoutcomeassessors. CriteriaforajAnyoneofthefollowing: udgementof‘Lowrisk’ofbias. Noblindingofoutcomeassessment,butthereviewauthorsjudgethattheoutcomemeasurementisnotlikelytobeinfluencedbylackofblinding; Blindingofoutcomeassessmentensured,andunlikelythattheblindingcouldhavebeenbroken. CriteriaforthAnyoneofthefollowing: ejudgementof‘Highrisk’ofbias. Noblindingofoutcomeassessment,andtheoutcomemeasurementislikelytobeinfluencedbylackofblinding; Blindingofoutcomeassessment,butlikelythattheblindingcouldhavebeenbroken,andtheoutcomemeasurementislikelytobeinfluencedbylackofblinding. CriteriaforthAnyoneofthefollowing: ejudgementof?‘Unclearrisk’ofbias. Insufficientinformationtopermitjudgementof‘Lowrisk’or‘Highrisk’; Thestudydidnotaddressthisoutcome. INCOMPLETEOUTCOMEDATA? Attritionbiasduetoamount,natureorhandlingofincompleteoutcomedata. CriteriaforajAnyoneofthefollowing: udgementof‘Lowrisk’ofbias. Nomissingoutcomedata; Reasonsformissingoutcomedataunlikelytoberelatedtotrueoutcome(forsurvivaldata,censoringunlikelytobeintroducingbias); Missingoutcomedatabalancedinnumbersacrossinterventiongroups,withsimilarreasonsformissingdataacrossgroups; Fordichotomousoutcomedata,theproportionofmissingoutcomescomparedwithobservedeventrisknotenoughtohaveaclinicallyrelevantimpactontheinterventioneffectestimate; Forcontinuousoutcomedata,plausibleeffectsize(differenceinmeansorstandardizeddifferenceinmeans)amongmissingoutcomesnotenoughtohaveaclinicallyrelevantimpactonobservedeffectsize; Missingdatahavebeenimputedusingappropriatemethods. CriteriaforthAnyoneofthefollowing: ejudgementof‘Highrisk’ofbias. Reasonformissingoutcomedatalikelytoberelatedtotrueoutcome,witheitherimbalanceinnumbersorreasonsformissingdataacrossinterventiongroups; Fordichotomousoutcomedata,theproportionofmissingoutcomescomparedwithobservedeventriskenoughtoinduceclinicallyrelevantbiasininterventioneffectestimate; Forcontinuousoutcomedata,plausibleeffectsize(differenceinmeansorstandardizeddifferenceinmeans)amongmissingoutcomesenoughtoinduceclinicallyrelevantbiasinobservedeffectsize; ‘As-treated’analysisdonewithsubstantialdepartureoftheinterventionreceivedfromthatassignedatrandomization; Potentiallyinappropriateapplicationofsimpleimputation. CriteriaforthAnyoneofthefollowing: ejudgementof?‘Unclearrisk’ofbias. Insufficientreportingofattrition/exclusionstopermitjudgementof‘Lowrisk’or‘Highrisk’(e.g.numberrandomizednotstated,noreasonsformissingdataprovided); Thestudydidnotaddressthisoutcome. SELECTIVEREPORTING? Reportingbiasduetoselectiveoutcomereporting. CriteriaforajAnyofthefollowing: udgementof‘Lowrisk’ofbias. Thestudyprotocolisavailableandallofthestudy’spre-specified(primaryandsecondary)outcomesthatareofinterestinthereviewhavebeenreportedinthepre-specifiedway; Thestudyprotocolisnotavailablebutitisclearthatthepublishedreportsincludeallexpectedoutcomes,includingthosethatwerepre-specified(convincingtextofthisnaturemaybeuncommon). CriteriaforthAnyoneofthefollowing: ejudgementof‘Highrisk’ofbias. Notallofthestudy’spre-specifiedprimaryoutcomeshavebeenreported; Oneormoreprimaryoutcomesisreportedusingmeasurements,analysismethodsorsubsetsofthedata(e.g.subscales)thatwerenotpre-specified; Oneormorereportedprimaryoutcomeswerenotpre-specified(unlessclearjustificationfortheirreportingisprovided,suchasanunexpectedadverseeffect); Oneormoreoutcomesofinterestinthereviewarereportedincompletelysothattheycannotbeenteredinameta-analysis; Thestudyreportfailstoincluderesultsforakeyoutcomethatwouldbeexpectedtohavebeenreportedforsuchastudy. CriteriaforthInsufficientinformationtopermitjudgementof‘Lowrisk’or‘Highrisk’.Itislikelythatthemajorityofejudgementstudieswillfallintothiscategory. of?‘Unclearrisk’ofbias. OTHERBIAS? Biasduetoproblemsnotcoveredelsewhereinthetable. CriteriaforajThestudyappearstobefreeofothersourcesofbias. udgementof‘Lowrisk’ofbias. CriteriaforthThereisatleastoneimportantriskofbias.Forexample,thestudy: ejudgementof‘Highrisk’ofbias. Hadapotentialsourceofbiasrelatedtothespecificstudydesignused;or Hasbeenclaimedtohavebeenfraudulent;or Hadsomeotherproblem. CriteriaforthTheremaybeariskofbias,butthereiseither: ejudgementof?‘Unclearrisk’ofbias. Insufficientinformationtoassesswhetheranimportantriskofbiasexists;or Insufficientrationaleorevidencethatanidentifiedproblemwillintroducebias. TablePossibleapproachforsummaryassessmentsoftheriskofbiasforeachimportantoutcome(acrossdomains)withinandacrossstudies
RiskofbInterpretation ias LowriskPlausiblebiasunlikelytoseriLowriskofbiasforallkMostinformationisfromstudiesatlowriskofbias. ofbias. ouslyaltertheresults. eydomains. Withinastudy Acrossstudies UnclearPlausiblebiasthatraisessoUnclearriskofbiasforMostinformationisfromstudiesatloworunclearrriskofbioneormorekeydomaas. medoubtabouttheresults. ins. iskofbias. HighriskPlausiblebiasthatseriouslyHighriskofbiasforonTheproportionofinformationfromstudiesathighofbias. weakensconfidenceinthereeormorekeydomainriskofbiasissufficienttoaffecttheinterpretationosults.
s. fresults.
因篇幅问题不能全部显示,请点此查看更多更全内容
Copyright © 2019- efsc.cn 版权所有 赣ICP备2024042792号-1
违法及侵权请联系:TEL:199 1889 7713 E-MAIL:2724546146@qq.com
本站由北京市万商天勤律师事务所王兴未律师提供法律服务