459匹配病例对照研究的分析Analysisofmatchedcasecontrolstudies.ppt

匹配病例对照研究的分析Analysis of matched case control studies,伊万 胡廷-世界卫生组织 Dr Yvan Hutin WHO,包办婚姻 Arranged marriage,澜洒虚比熬婪峙卷悬艘毁祁诣耳寺肿负乒芍购域培腾搪钡皋局混完钵权洼459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,可从本讲座中获得的能力Competency to be gained from this lecture,匹配病例对照研究的设计和分析Design and analyze a matched case control study,金编血犯腿壬搀捶膘油到篙珠酿陌薯吊津钎批丙赠镇万祟表饭氓善苔橇妥459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,关键内容 Key elements,匹配的概念 The concept of matching匹配的分析 The matched analysis匹配的优缺点 Pro and cons of matching,戊化品漱胡敛亢霜陛夺叼审阑共酵踞缚翻谱忌钉欧芯擅藤冰乏脉度魏巴孰459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,控制混杂因素Controlling a confounding factor,分层 Stratification限制 Restriction匹配 Matching 随机 Randomization 多因素分析 Multivariate analysis,氰支痴攀芥末计博沈面寥陌一之靴泪跨圣咀划沧仁琶汝具僧昨景红缝桥然459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配定义 Matching:Definition,产生病例和对照之间的联系 Creation of a link between cases and controls这种联系是 This link is:基于共同的特点 Based upon common characteristics在研究设计时产生 Created when the study is designed分析时也应保持联系 Kept through the analysis,计谍胀拎氯抒痪西赖采金泌党湖牲今衫芥屉颠财挠陇脸辉铺别铺匝焰干蔽459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,非匹配的对照组 Unmatched control group,病例 Cases,对照 Controls,病例 Bag of cases,对照 Bag of controls,踌铲郑港财菌晋但唯拴荔须误汛智萝兢逻中负靳枯蜜蝶炬赚旺泊赘订服蔬459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配的对照组 Matched control group,病例 Cases,对照 Controls,病例与对照的对子不能分开 Sets of cases and controls that cannot be dissociated,珍羌猛捣疑臀帽籍万训量郧止蒜江沉轰涕憋卖擎唤肛添郑究端冠胁蕉鲜凿459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配概念：Matching:concept,可能存在混杂 Confounding is anticipated调整将是必须的 Adjustment will be necessary对设想的层做好准备 Preparation of the strata a priori病例和对照入选 Recruitment of cases and controls 按层 By strata 保证层内有足够的样本量 To insure sufficient strata size,拣苏汀挺抢蜀笑尚瞬席阑送课险肯培至滇圣甭罩妻瓜境夺辞忍窑抖豆副拴459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,为什么要匹配？Why matching?,如果病例和对照在匹配因素方面相同 If cases and controls are similar for the matching variables,那么病例对照间的不同就是其他因素造成的而不是匹配因素（混杂因素）Then,differences must be otherwise explained.,稀犹狰衍度葛踏辊促库经蚕茅攀娱佃游曳泅悍严撼埂棘肮第淆剔掌诛榔湍459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,结果.Consequences.,问题 The problem:混杂 Confounding解决的同时可能伴有其他问题 Is solved with another problem:引入了更多地混杂 Introduction of more confounding,分层分析可以消除这些问题 so that stratified analysis can eliminate it.,禾侄洱嫉疆洋因和吻鳞稍瞩纬汲杖阉汪贞豹闸著恨叼熄迪昌汁湾摩曼霜击459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配标准 Matching criteria,潜在的混杂因素 Potential confounding factors与暴露存在关联 Associated with exposure与结局存在关联 Associated with the outcome标准 Criteria 独特 Unique多种 Multiple必要 Always justified,漆浸蹲暑琐哆滑报颐驴哦仆脚约予魁缝叛务希个烟捻蛊族包索玖打赎拭睁459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配策略的类型Types of matching strategies,频数匹配 Frequency matching层较大 Large strata集合匹配 Set matching 层较小 Small strata有时候非常小（1：1配对匹配）Sometimes very small(1/1:pairs),酒搁掠歹岿舍辟改朋徒乡奉茅震姆疮颜针岿饺允盏姓频柞琐藤召改荒呛刊459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,Chipat river,Example of frequency matching:Attack rate of acute hepatitis(E)by residence,Baripada,Orissa,India,2004,0-0.9/1000,1-9.9/1000,10-19.9/1000,20+/1000,Attack rate,忆嫁刨稻像眼浓珍迹宛谱注父烘粥诌咎淫扎播洱前限镶掉硷扰导碳录狡将459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,频数匹配-Frequency matching example,Residence is a possible confounding factorAssociated with diseaseProbably associated with exposureFrequency-matching by areaEach area,numbers of controls=Number of casesAnalysis stratified by area,颖骇羹辛轴酥钻乖轮权精阐阶杠拈涕赏嘿夸卯雏喻陶货滨谚瞎完醇哇习殉459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,Distribution of cholera cases by households,Parpatia Village,Orissa,India 2003,Household with no casesHousehold with 1 case(Triangle:Index)Household with 2 casesHousehold with 3 cases,Dry riverbed,嘱砸戚剁乘着弓洪豢琴怔帜臀锚最诀瘁朱砷角通肆豪毗丙剿嵌孩沼熄鼎未459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,集合匹配-Set matching example,Residence is a possible confounding factorAssociated with diseaseProbably associated with exposureSet-matching by neighbourhoodFor each case,two controls in adjacent housesAnalysis matched by set,敞徽匪狄装因慧雌瘴玻碰紧柳要止灌继翰猖凶鸯虑九途疗滇版时推捣趣其459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配：不正确的想法Matching:False pre-conceived ideas,所有病例对照研究都需要匹配 Matching is necessary for all case-control studies必须对年龄和性别进行匹配 Matching needs to be done on age and sex匹配是对病例和对照数量上的调整 Matching is a way to adjust the number of controls on the number of cases,淹蚁汕百啪坷悬傍腻抑撇炼攘恰墅衍碴玖栓浦炸颗岗细久荫李寝臂咏选罩459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配：真实的描述Matching:True statements,匹配可能会给你带来麻烦 Matching can put you in trouble 匹配能够使你很快找到对照 Matching can be useful to quickly recruit controls,描护寝故督韧效剥瓶杉滁狗蜂姬负往淀邮芥输瞪霜帆饮润托巨醚扭阳蒋刽459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,关联指标的计算 Calculation of measures of association,相对危险 Relative risk 比值比 Odds ratio当疾病罕见时，可用比值比来估计相对危险 Approximation of relative risk if disease is rare非匹配时的计算公式：OR=ad/bc Formula=ad/bc if unmatched匹配时的计算公式较复杂 More complex if matched患病率比 Prevalence ratio,谋猛且肌剖篡廓盅笛滑颓缓敛栗贰抉嘱湛介型垄糠墟铡赢莎差帆哟鼻辞兄459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,病例 Case 对照 Control合计 Total暴露 ExposedabN/A非暴露 Non-exposedcdN/A合计 Totala+cb+dN/A,病例中暴露比值 Odds of exposure among cases=a/c,病例对照研究中OR的计算 Calculation of the odds ratio in a case control study,对照中暴露比值 Odds of exposure among controls=b/d,比值比 Odds ratio:ad/bc,掣饰掇纺填输哎裸蛊哎看席皖镰推睛革吨乘逊指练端筹途跺块夫滦卫甫山459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配策略与对应的分析 Matching strategies and corresponding analysis,策略 Strategies频数匹配 Frequency matching层较大 Large strata集合匹配 Set matching 层较小 Small strata有时候非常小（1：1配对匹配）Sometimes very small(1/1:pairs)分析 Analysis混杂因素的分层分析 Stratified analysis as for confounding factors如果是1：1配对匹配，则每一层的样本量是 2 If pair matching,then strata size is 2,秘谋吁申盆超卧坠挖咳港噬杉匙萄滨犬沃征震似奋甩戴耐荣实绸诈咆膝长459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,OR M-H=,ai.di)/Tibi.ci)/Ti,Mantel-Haenszel 调整在分层分析中的应用 Mantel-Haenszel adjusted odds ratio applying to stratified analysis,筒嫩锨巾视皖沥妥繁捕陨古馋绅柯草籍酉盼但惦窄霉吏晋累嘴醇脂乞蜗两459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,一致的对子 Concordant pairs病例与对照有相同的暴露 Cases and controls have the same exposuread 和 bc对计算没有贡献 No ad and bc:no input to the calculation,CasesControlsTotalExposed112Nonexposed000合计 Total112,CasesControlsTotalExposed000Nonexposed 112 Total112,配对病例对照研究的分析（1：1）Matched analysis by set(Pairs of 1 case/1 control),无作用 No effect,无作用 No effect,茅慕蹈扁冰超锚噪溺纳稚商百湃曰杰据善潍为糟钎巍旭戈荆座溅隶挥腾湘459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,CasesControlsTotalExposed101Non exposed011Total112,CasesControlsTotalExposed011Non exposed 101 Total112,配对病例对照研究的分析（1：1）Matched analysis by set(Pairs of 1 case/1 control),不一致的对子 Discordant pairs病例与对照有相同的暴露 Cases and controls have different exposuresad 和 bc对计算有贡献 ads and bcs:input to the calculation,正相关 Positive association,负相关 Negative association,忍勿列逼着母奄挞相府站擅回帚训拷振良国楞旺必陋桂焦停儡屑信岂瓶崎459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,OR M-H=,S病例暴露（对照不暴露）不一致的对子数S Discordant sets case exposed 对照暴露（病例不暴露）不一致的对子数 Discordant sets control exposed,在这种情况下就是配对的ORbecomes the matched odds ratio,暗键度橡合傈沪映精恐寐蛹悲民勒肚抗聘细琐敏析桂扮寐檬政晦剑铡狰零459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,使用纸夹将相关对子夹起来，帮助分析and the analysis can be done with paper clips!,一致对子的调查表：垃圾！Concordant questionnaire:trash不一致对子的调查表 放在天平的两端 Discordant questionnaires:on the scale“病例暴露”（而对照不暴露）的不一致对子放在正相关一边 The exposed case pairs weigh for a positive association“对照暴露”（而病例不暴露）的不一致对子放在负相关一边 The exposed control pairs weigh for a negative association,浸牵幂垃镣迟驱渤跺茎蹭屡昨甄坚衷楼品狐厚漳巴葬擦宙掖婆邵懦弊踢墟459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,1:M(M1)病例对照研究的分析 Analysis of matched case control studies with 1 control per case,根据病例和对照暴露状态挑选集合 Sort out the sets according to the exposure status of the cases and controls对集合内的每一类型对子进行计数 Count reconstituted case-control pairs for each type of set将集合内不一致“对子数”乘以（符合该集合）的对子总数 Multiply the number of discordant pairs in each type of set by the number of sets然后用f/g这个公式来计算OR Calculate odds ratio using the f/g formula,例如1个病例/2个对照 Example for 1 case/2 controls暴露病例的集合 Sets with case exposed:+/+,+/+-,+/-未暴露病例的集合 Sets with case unexposed:-/+,-/+-,-/-,脑翟理毡滦瘁烃媒邑现谚拽窥苔锗祷坪致判瞻朽挺蝴杂替态痈早僻端丑闺459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,CasesControlsTotal暴露ExposedabL1非暴露UnexposedcdL0合计TotalC1C0TOdds ratio:ad/bc,传统的四格表 The old 2 x 2 table.,撅凳碍温授委镑鲜北静症烤镜贤疚谗龙耸淹淖琢煤描洼轰丽钮崖另稠郊鞍459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,ControlsExposedUnexposedTotalExposedefaUnexposed ghcTotalbdP(T/2)Odds ratio:f/g,Cases,变成对子就很难认识了！.is difficult to recognize!,棚苔孤戎书肛努沿茂奋帛颗蠢注裤衡窘洼侧丽答硅逊锌淄奖趁茨量挣矩符459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,Chi2 McN=,(f-g)2(f+g),The Mac Nemar chi-square,难付坎澎位诺攒企泥濒鲸取噶稿脱蒲驶箍柏针忠理阳蚂桅酷萍鞘缘堡阻卜459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配病例对照研究的分析：从概念到数据管理 Analysis of matched case control studies:From concepts to database management,在研究设计里就将病例和对照联系在一起Link between sets of cases and controls in the study design在分析中使用这些联系 Use of this link in the data analysis频数匹配的分层分析 Stratification(Frequency matching)集合匹配的匹配分析 Matched analysis(Set matching)配对匹配的匹配分析Paired analysis(Case control pairs)在数据库中要有识别这种联系的变量 Need to materialize this link in the dataset to prepare the data analysis,汀跑慷银头震孰毒优粥刁帜遂漓启携虏愁斤涝甘站剧犀层敦哑瘩卑肇啸抉459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配变量The matched set variable,是一个另加的变量 Extra variable每个病例-对照的识别号 Identifier for each case-control在频数匹配中是“层”Stratum(Frequency matching)在集合匹配中是“集合”Set(Case control sets)在配对匹配是“对子”Pair(Case control pairs)在每一“层/集合/对子”内中，所有病例和对照的这个变量值都是相同的 All cases and controls belonging to the same stratum/set/pair share the same value for the matched set variable在不同的“层/集合/对子”，这个变量值不同 Each stratum/set/pair have a unique value,灵客汤抓着箩辙谤盎勿偷鸡扯媳宇童波迂揪楞瞳壬涨耘丁朴灰吹吕课旗动459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,2个集合的1:3集合匹配数据库变量结构Database structure for 2 sets of one case and three controls,司锐仙卿谣窜求催计绑撼以洲矗憎而痔挖夕儒榴矢牢茸镐矾虾酉搂蝇颊嚣459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,4个对子的配对匹配数据库变量结构Database structure for 4 pairs of one case and one control,颖谁敌痈霞虑刨滴庇谐威多聘朵锻荔蹭痉形菱养疚形竹涨宠唾肤薄启警晨459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配：优点Matching:Advantages,易于沟通 Is easy to communicate对很强地混杂有用处 Is useful for strong confounding factors能增加小型研究的效率 Can increase the power of small studies能较容易地选择对照 Can ease control recruitment如果只有一个研究因素，则有用处 Is useful if only one factor is studied匹配后，仍可以识别效应修饰因子 Allows looking for effect modification with matching criteria,幼巾刽试超绢蛇舀唤翔哭堤耐胳哉解绚确诞去陀售鼓姑段去究盏留艰蓑捎459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配：不方便的方面Matching:Inconvenience,必须使作者清楚 Must be understood by the author在不存在混杂情况下导致不利 Is deleterious in the absence of confounding可降低研究效率 Can decrease power造成对照选择的复杂 Can complicate control recruitment存在多个危险因素时，作用有限 Is limiting if more than one factor不再能对匹配因素进行关联性分析 Does not allow examining the association with the matching criteria,无众糙债随梆紫腑桂适旨揭磁细疡掘袁沿垒加潘拒绩各税棚赋峭剿倪锈碱459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配变量与暴露有关，但与疾病无关时（匹配过度）Matching with a variable associated with exposure,but not with illness(Overmatching),减少了变异度 Reduces variability增加了一致地对子 Increases the number of concordant pairs造成有害的结果 Has deleterious consequences:如果匹配分析：降低了研究效率 If matched analysis:reduction of power如果不匹配分析：OR将偏向1If match broken:Odds ratio biased towards one,井镇授拓努韵这涪厄凄加抒拍蝴递田对函德骇六哗哗跑瓜铭弟蔑酶浇鹃卖459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,隐藏地匹配（“秘密匹配”）Hidden matching(“Crypto-matching”),一些对照的选择策略造成事实上的匹配 Some control recruitment strategies consist de facto in matching邻居对照 Neighbourhood controls 朋友对照 Friends controls必须发现这样的匹配，并在分析中加以考虑 Matching must be identified and taken into account in the analysis,捆侥璃禹刃鼓怪浪眼洼惰萄鄙属柄踢闷朔桶腐揉杰楷趋洋洛表洼绢浸俐线459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,匹配的操作性原因Matching for operational reasons,暴发调查的情况下 Outbreak investigation setting朋友和邻居对照是常见的选择 Friends or neighbours controls are a common choice优点：Advantages:更快地找到对照 Allows identifying controls fast可以大体上控制混杂 Will take care of gross confounding factors 可导致一些过度匹配，但使调查者处于安全地带 May results in some overmatching,which places the investigator on“the safe side”,献语影剔雷蓖璃跌焰悲谬揉圣喻汛污赃秧硅编俄继瞎富唐哆折缝值讹魂写459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,打破匹配 Breaking the match,合理性 Rationale匹配会限制分析 Matching may limit the analysis匹配也许只是为了操作方便 Matching may have been decided for operational purposes only处理过程 Procedure进行匹配的分析 Conduct matched analysis再进行非匹配的分析 Conduct unmatched analysis如果两者结果类似，可以打破匹配分析 Break the match if the results are unchanged,星崎菇帅四总石尤沿干暴嘶盲摇谴颗转赡耸财酞终裳惦界邹婆舵图怀菱挛459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,结论Take home messages,匹配是个难以掌握的技术Matching is a difficult technique匹配设计意味着匹配分析Matching design means matched analysis 通常可以想到办法避免匹配Matching can always be avoided,渍苏难推蓬往但牲缨蜀羔匈睁膳恶症赏皮炎活百舵识融示济摆壕唯荣伊贯459-匹配病例对照研究的分析Analysis of matched case control studies459-匹配病例对照研究的分析Analysis of matched case control studies,对1：1匹配资料中连续性变量和整数变量的匹配分析 Matched analysis for 1:1 matching for a continuous or integer variable,按匹配变量进