病因和因果推斷課件

病因和病因推斷CauseofDiseaseandCausalInference流行病學(xué)教研室

Departmentofepidemiology，HarbinMedicalUniversity主要內(nèi)容(outline)

病因的概念

（conceptsofcausation）

因果關(guān)聯(lián)的推斷（inferenceofcausalassociation）

病因推斷的標(biāo)準(zhǔn)

(criteriaofcausalinference)第一節(jié)病因的概念

conceptsofcausation

鬼神、上帝、天意金木水火土人活的傳染物迷信階段樸素唯物主義生物學(xué)病因的萌芽病原物環(huán)境宿主三角模式生理社會物化機(jī)體輪狀模式病因網(wǎng)模型一、疾病病因概念的發(fā)展（Developmentofcausationinhistory

）

病因模型（causalmodel）三角模型（epdemiologicaltriad）疾病因素輪狀模型（wheelmodel）病因網(wǎng)絡(luò)模型(webofcausation)宿主環(huán)境病因健康（health）宿主環(huán)境病因疾病（disease）失衡三角模型（epidemiologicaltriad）環(huán)境病因宿主AgentEnvironmentalHost病因Agent環(huán)境Environmental宿主Host三角模型（epidemiologicaltriad）病因Agent環(huán)境Environmental宿主Host三角模型（epidemiologicaltriad）病因Agent環(huán)境Environmental宿主Host三角模型（epidemiologicaltriad）病因Agent環(huán)境Environmental宿主Host三角模型（epidemiologicaltriad）生物環(huán)境社會環(huán)境遺傳內(nèi)核理化環(huán)境宿主健康Health理化環(huán)境生物環(huán)境社會環(huán)境宿主遺傳內(nèi)核失衡疾病Disease輪狀模型（wheelmodel）生物環(huán)境社會環(huán)境理化環(huán)境先天性代謝異常Inbornerrorofmetabolism

宿主遺傳內(nèi)核健康Health病因網(wǎng)絡(luò)模型（webofcausation）MorbidityandMortality(Stroke,MI)BiologicalRiskFactors

(Hypertension,BloodLipids,Homocysteine)GeneticRiskFactors(FamilyHistory)BehavioralRiskFactors(Cigarette,Diet,Exercise)EnvironmentalFactors(SocioeconomicStatus,WorkEnvironment)病因網(wǎng)絡(luò)模型（webofcausation）BEINGSModel

PreventablecausesofdiseaseBiologicfactorsandBehavioralfactorsEnvironmentalfactorsImmunologicfactorsNutritionalfactorsGeneticfactorsServices,Socialfactors,andSpiritualfactorsAcausalrelationshipwouldberecognizedtoexistwheneverevidenceindicatesthatthefactorsformpartofthecomplexofcircumstancesthat

increasestheprobabilityoftheoccurrenceofdisease

andthatadiminutionofoneormoreofthesefactorsdecreasesthefrequencyofthatdisease.----Lilienfeld(1980)二、病因的定義（definitionofcausation）

LilienfeldAM.（1920-1984）約翰.霍普金斯大學(xué)流行病學(xué)教授

LilienfeldAM.（1920-1984）約翰.霍普金斯大學(xué)流行病學(xué)教授那些能使人群發(fā)病概率升高的因素，就可認(rèn)為是疾病的病因；當(dāng)其中某個或多個因素不存在時，人群的疾病頻率就會下降。二、病因的定義（definitionofcausation）第二節(jié)因果關(guān)聯(lián)的推斷

inferenceofcausalassociation一、因果推斷的一般過程

Commonprocessofcausalinference建立病因假設(shè)

hypothesisdevelopment

病因假設(shè)的檢驗和驗證

testandcertificationofhypothesis密爾氏準(zhǔn)則Mill’scanons建立病因假設(shè)（hypothesisdevelopment

）Ifeventsorriskfactorsarecommontoavarietyofdifferentcircumstancesandtheeventsorriskfactorshavebeenpositivelyassociatedwithadisease,thentheprobabilityofthatfactorbeingthecauseisextremelyhigh.求同法（agreementcanon）鉤端螺旋體病流行，共同點---下水勞動Thefrequencyofadiseaseoccurrenceisextremelydifferentunderdifferentsituationsorconditions.Ifariskfactororeventcanbeidentifiedinoneconditionandnotinasecond,itmaybethatfactor,ortheabsenceofit,thatmaybethecauseofthedisease.差異法（differentcanon）察布查爾病Thefrequencyorstrengthofaneventorriskfactorvarieswiththefrequencyofthediseaseorcondition.Increasednumbersofchildrennotimmunizedagainstmeaslescausestheincidencerateformeaslestogoup.共變法（concomitantvariationscanon）吸煙與肺癌；氟齲齒、斑釉齒Subductfromanyphenomenonsuchpartasisknownbypreviousinductionstobetheeffectofcertainantecedents,andtheresidueofthephenomenonistheremainingantecedents.剩余法（residuescanon）乙肝病毒＋黃曲霉毒素－肝癌（藻類毒素）

病例對照研究（case-controlstudy）隊列研究（cohortstudy）病因假設(shè)的檢驗和驗證（testandcertificationofhypothesis）二、病因推斷中暴露因素與疾病關(guān)聯(lián)的解釋Associationbetweenexposureanddiseaseincausalinference統(tǒng)計學(xué)關(guān)聯(lián)（statisticsassociation）虛假關(guān)聯(lián)（spuriousassociation）

間接關(guān)聯(lián)（indirectassociation）

InterpretingAssociations-CausalandNon-CausalCausalNon-Causal(duetoconfounding)CharacteristicUnderStudyCharacteristicUnderStudyDiseaseFactorXDiseaseInterpretingAssociations-CausalandNon-CausalCausalNon-Causal(duetoconfounding)CoffeeConsumptionCoffeeConsumptionPancreaticCancerSmokingPancreaticCancerSpuriousAssociationRealAssociationRealAssociation暴露與疾病exposureanddisease有無統(tǒng)計學(xué)關(guān)聯(lián)?statisticalassociation有無偏倚?bias有時間先后否?temporalsequence統(tǒng)計關(guān)聯(lián)到因果關(guān)聯(lián)

FromEpidemiologicalAssociationtoCausation

關(guān)聯(lián)的強(qiáng)度

關(guān)聯(lián)的重復(fù)性

關(guān)聯(lián)的時間順序病因與疾病分布一致終止效應(yīng)劑量反映關(guān)系關(guān)聯(lián)的特異性關(guān)聯(lián)的生物學(xué)合理性第三節(jié)病因推斷標(biāo)準(zhǔn)criteriaofcausalinferencestrengthconsistencytemporalitycoherence

reversibilitydose-responserelationshipspecificitybiologicplausibility因果推斷標(biāo)準(zhǔn)(1)--關(guān)聯(lián)的強(qiáng)度關(guān)聯(lián)強(qiáng)度越大，為因果的可能性越大

Astrongassociationbetweenpossiblecauseandeffect,ismorelikelytobecausalthanaweakassociation.

關(guān)聯(lián)強(qiáng)度的測定

measurementofstrengthRR,ORcriteriaofcausalinference1--strength因果推斷標(biāo)準(zhǔn)(1)--關(guān)聯(lián)的強(qiáng)度criteriaofcausalinference1--strength不同人群、不同地區(qū)、不同時間、不同類型的研究少數(shù)或個別研究不同甚或相反的結(jié)果并不能簡單地反駁因果假設(shè)多數(shù)得到一致結(jié)果的研究可能遭遇共同的偏倚研究間不同的結(jié)果可能是因為某些因素的修正作用因果推斷標(biāo)準(zhǔn)(2)－關(guān)聯(lián)的重復(fù)性criteriaofcausalinference2--consistency因果推斷標(biāo)準(zhǔn)(3)－時間順序前因后果：唯一必須滿足的標(biāo)準(zhǔn)實驗性研究隊列研究采用新病例的病例對照研究生態(tài)趨勢研究橫斷面研究>>時間暴露結(jié)局criteriaofcausalinference3--temporaliryThecausemustprecedetheeffect.因果推斷標(biāo)準(zhǔn)(4)－病因與疾病分布一致criteriaofcausalinference4--coherence飲食脂肪攝入與乳腺癌發(fā)病Correlationbetweendietaryfatintakeandbreastcancerbycountry因果推斷標(biāo)準(zhǔn)(5)－終止效應(yīng)這是強(qiáng)有力的反證病因關(guān)系的一種方法。例如戒煙與肺癌的關(guān)系研究中，我們發(fā)現(xiàn)戒煙后，人群肺癌死亡率下降。criteriaofcausalinference5--reversibilityWhenremovalofapossiblecauseresultsinreductionofdiseaserisk,thelikelihoodoftheassociationbeingcausalisstrengthenedForexample,smokerswhogiveupsmokinghavealowerriskoflungcancerthanthosewhocontinuetosmoke.因果推斷標(biāo)準(zhǔn)(6)－劑量反映關(guān)系A(chǔ)dose-responserelationshipoccurswhenchangesinthelevelofapossiblecauseareassociatedwithchangesinthemagnitudeoftheeffect.criteriaofcausalinference6—dose-responserelationship如果觀察到隨著某因素暴露劑量的增加．人群發(fā)生某病