应用实例分析---临床流行病学研究-胎源性疾病-DOHAD

应用实例分析-临床流行病学研究胎源性疾病（DOHAD),骆忠诚上海交通大学附属新华医院,学习目标,通过应用实例分析，加深对流行病学重要基本概念的理解通过应用实例分析，提高对流行病学研究重要方法的应用能力,内容,实例分析一个经典的有关胎源性疾病的临床流行病学研究报告实例分析-设计一个有关胎源性疾病的临床流行病学研究项目,胎源性疾病/发育源性疾病(DOHAD),Thesuscepatibleitytomanychronicdiseasesinadulthoodcanbetracedbacktoexposuresduringearlylife(duringfetalandearlypostnatallife)成年期的许多慢性疾病的易感性可以追溯到生命早期的暴露因素（胎儿期和出生后早期）无致病基因突变,流行病学-重要作用之一发现新联系（DOHAD）,*Oddsratiofortwohourglucoseconcentrationof7.8mmol/ladjustedforcurrentbodymassindex.(X2fortrend=15.4;p7.8mmol/ladjustedforcurrentbodymassindex.,AnswerstoQuestions（1）,Whatisthestudydesign研究设计是？Retrospectivecohortstudy回顾性队列研究WhyOR?whynotRR?Eithergoodforreasoning,thelatterismoreaccurateindefiningtherelativeriskdisparity.为什么用比值比（OR）？而不是相对风险度（RR？推理任一均可，RR在定义相对风险差距更准确。CanwehavecrudeORandcrudeRR?Yes,incohortstudiesorRCT,youcancalculateRR.我们可以计算粗OR（未调整比值比）,粗RR（未调整相对风险度）吗？是的，在队列研究或试验，可以计算出RR。WhyadjustedOR?WhynotadjustedRR?Eitherisgoodforreasoning,itiseasiertocalculatetheadjustedOR.为什么是调整OR？而不是调整RR？推理任一均可，调整OR更容易计算.,Forcohortstudydata,youcanuselogbinomialmodel,toobtaintheadjustedRR.队列研究的数据，您可以使用log二项式模型，得到调整后的RR。,HowtocalculateadjustedRR,inthisstudy?在本研究中，如何计算调整后的RR？,AnswerstoQuestions（2）,DoweneedadjustmentsinORorRR?Mosttimes,yes.我们需要调整吗？大多数时候，是的。Isadjustmentsalwaysnecessary?No,sometimesunnecessary.调整是必要的吗？不，有时不必要。WhatthedifferenceBetweenORandRR?ORdoesnotalwaysrepresentRR.ORcanbecalculatedinanystudydesigns,RRscannotbecalculateddirectlyincasecontrolstudies.OR和RR有什么区别？OR有时不能代表RR。可以在任何临床流行病学研究设计中计算OR。在病例对照研究不能直接计算RR。WhatsthedifferencebetweenCrudeORvs.AdjustedOR?TheadjustedORmoreoften(butnotalways)representsthetrueassociation。未调整OR或与调整OR区别是什么？调整OR更经常（但并不总是）代表真正的联系。WhatsthedifferencebetweenCrudeRRvs.AdjustedRR?调整RR更经常（但并不总是）代表真正的联系。,真理？假象？,CausalInferenceConsiderations因果推理思考Informationbias?信息偏倚lesslikelyConfoundingfactors？混杂因素possiblyConsistencyofassociation?联系的一致性yesStrengthofassociation?关联强度OKDose-responserelationship?剂量-反应关系yesTemporallyorderconsistent?时间一致性yesDeterministic/probabilistic?决定性noNecessary?必要性noSufficient?充分noSpecificity?特异性noBiologicalplausibility?生物合理性yesSurrogateriskfactor?替代风险因素maybeAnimalmodelexperiment?实验动物模型yes,ORsforimpairedglucosetoleranceAdjustedORsCrudeORsIsCurrentBMIaconfounder?,*adjustedforcurrentbodymassindex.,ConfounderorEffectMediator?混杂因素,或影响介质?,Glucosetolerance糖耐量Bloodpressure血压,CurrentBMI现体重指数,Birthweight出生体重,Confounders混杂因素(e.g.ethnicity如种族),Whenyouinappropriateadjustforafactorinthecausalpathway,youcouldproduceafalseassociation,orexaggeratedassociation当你不适当的调整一个在因果通路途径上的因素，你可能会产生一个虚假的关联，或夸张的关联。ItmaybeinappropriatetoadjustforcurrentBMIinestimatingtheeffectofbirthweightoncurrentglucosetoleranceorbloodpressure.调整现在的体重指数以估计出生体重对目前的糖耐量或血压的影响可能是不合适的。,Reversalparadox逆转谜题,Whyevidenceforthefetaloriginsofadultdiseasemightbeastatisticalartifact:thereversalparadoxfortherelationbetweenbirthweightandbloodpressureinlaterlife.TuYK,WestR,EllisonGT,GilthorpeMS.AmJEpidemiol;161(1):27-32.,Someresearchershaverecentlyquestionedthevalidityofassociationsbetweenbirthweightandhealthinlaterlife.Theyarguethattheseassociationsmightbedueinparttoinappropriatestatisticaladjustmentforvariablesonthecausalpathway(suchascurrentbodysize),whichcreatesanartifactualstatisticaleffectknownasthereversalparadox.Computersimulationswereconductedforthreehypotheticalrelationsbetweenbirthweightandadultbloodpressure.Theauthorsexaminedtheeffectofstatisticallyadjustingfordifferentcorrelationsbetweencurrentweightandbirthweightandbetweencurrentweightandadultbloodpressuretoassesstheirimpactonassociationsbetweenbirthweightandbloodpressure.Whentherewasnogenuinerelationbetweenbirthweightandbloodpressure,adjustmentforcurrentweightcreatedaninverseassociationwhosesizedependedonthemagnitudeofthepositivecorrelationsbetweencurrentweightandbirthweightandbetweencurrentweightandbloodpressure.Whentherewasagenuineinverserelationbetweenbirthweightandbloodpressure,theassociationwasexaggeratedfollowingadjustmentforcurrentweight,whereasapositiverelationbetweenbirthweightandbloodpressurecouldbereversedafteradjustingforcurrentweight.Thus,researchersmustconsiderthereversalparadoxwhenadjustingforvariablesthatliewithincausalpathways.,Surrogateriskfactors替代风险因素?,Glucosetolerance糖耐量,Birthweight出生体重,Surrogateriskfactors替代风险因素,SharedgeneticvariantscausebothLBWandimpairedglucosetolerance?,Glucosetolerance糖耐量,Birthweight出生体重,Geneticvariants遗传变异,CausalMechanisms/pathways?epieneticchanges，etc.,Glucosetolerance糖耐量,Birthweight出生体重,Epigeneticchanges,etc.表观遗传改变，等.,IntrauterineenvironmentGlucocorticoids,hormones，etc糖皮质激素,激素等,Unknownconfounders未知的混杂因素,HalesCN,BarkerDJ.Thethriftyphenotypehypothesis.BrMedBull.2001;60:5-20,Thethriftyphenotypehypothesisproposesthattheepidemiologicalassociationsbetweenpoorfetalandinfantgrowthandthesubsequentdevelopmentoftype2diabetesandthemetabolicsyndromeresultfromtheeffectsofpoornutritioninearlylife,whichproducespermanentchangesinglucose-insulinmetabolism.Thesechangesincludereducedcapacityforinsulinsecretionandinsulinresistancewhich,combinedwitheffectsofobesity,ageingandphysicalinactivity,arethemostimportantfactorsindeterminingtype2diabetes.Sincethehypothesiswasproposed,manystudiesworld-widehaveconfirmedtheinitialepidemiologicalevidence,althoughthestrengthoftherelationshipshasvariedfromonestudytoanother.Therelationshipwithinsulinresistanceisclearatallagesstudied.Lessclearistherelationshipwithinsulinsecretion.Therelativecontributionofgenesandenvironmenttotheserelationshipsremainsamatterofdebate.Thecontributionsofmaternalhyperglycaemiaandthetrajectoryofpostnatalgrowthneedtobeclarified.,Project-Maternalglucosetolerance,oxidativestress,andprogrammingoftheMetabolic,CIHRFunded2006-2010,Luoetal.DiabetesCare2010,MaternalGlucoseToleranceinPregnancyAffectsFetalInsulinSensitivity,Luoetal.DiabetesCare2010,MaternalBMIwasalsoinverselycorrelatedFetalInsulinSensitivity,butlessstronglyso,IGF-1(butnotIGF-2)levelsinmaternalandfetalcirculationswereelevatedingestationaldiabetes,LuoZC,etal.JClinicalEndocrinologyMetabolism2012,HighermaternalIGF-1(notIGF-2)levelspredictincreasedriskofLGA/macrosomiaLuoZC,etal.JCinicalEndocrinologyandMetabolism2012,Leptinandadiponectinlevelswerepositivelycorrelatedinmaternalversusfetalcirculations,LuoZC,etal.Obesity2013,Fetalinsulinsensitivitywasnegativelyassociatedwithcordbloodleptin(p0.4)levels.,LuoZC,etal.Obesity2013,Summary总结(1),ORcanbecalculatedinanyclinicalepidemiologicstudies(cross-sectional,casecontrol,cohort),butORmay“overstate”theeffectsizewhentheoutc