环境健康学-环境流行病学.ppt

**AlternativeexplanationsChanceBiasSelectionbiasRepresentativeLosstofollow-upInformationbiasExposureOutcomeConfoundingIntentiontotreatResidualconfounding*BiasSelectionbiasNotrepresentativeofthetargetpopulationLosstofollow-up*选择偏倚

(SelectionBias)不同类型(就研究的暴露、结局特征而言)的个体入选研究的概率不同。排除或过分代表某一类型个体。NotrepresentativeofthetargetpopulationLosstofollow-up*信息偏倚

(InformationBias)又称测量偏倚、观察偏倚。是在收集信息过程中由于测量暴露与结局的方法(工具)有缺陷，使采集到的信息不准确(即不完全真实)，从而引起偏倚。错分是测量不准确导致的最直接的结果。*Howtominimisebias?Identifyastudypopulation:RepresentativeSufficientlylargeSufficienteventsReducelossestofollow-up*DefinitionofConfoundingReferstoanobservedassociation(orlackofone)whichisinfactduetoamixingofeffectsbetweentheexposure,theoutcomeandathirdfactor.Theextraneous(ie,third)variableiscalledaconfoundingfactor(CF).ACFmustbeariskfactorfortheoutcome,associatedwiththeexposure,andcannotbeanintermediatestepinthecausalpath.*SourcesofconfoundingNorandomisationAllocateparticipantsinadequatelyNointentiontotreatMisclassifyexposure/outcomes*MEASUREMENTOFCONFOUNDERSMeasureconfoundersaccuratelyandpreciselyAlmostimpossibletocompletelyadjustforconfoundingunconsideredunmeasuredinadequateinformation*Howtominimiseconfounding?AppropriaterandomisationDoubleblindPlacebocontrolledIntentiontotreat*Howtominimiseconfounding?AppropriaterandomisationDoubleblindPlacebocontrolledIntentiontotreat*夜灯的故事我女儿Sonia，旁边是盏夜灯。夜里开着这盏夜灯很方便，我们可以随时从门口查看屋里的情况。*夜灯与近视(Quinnetal.NATURE|VOL399|13MAY1999|现患比=2岁以前周围环境的夜灯－儿童近视的发生**于是，我们关掉了Sonia房中的夜灯，开始每天在屋子里抹黑蹒跚*后来，还是在Nature上，又发表了2篇文章，说夜灯与近视的发生没有因果关联。为什么？Wefoundanassociationbetweenthenumberofmyopicparentsandnurserylightingbeforeagetwo(X2=435.02,P0.001),aswellasanassociationbetweenethnicityandroomlighting(X2=489.22,P0.001).Thissamplecar