臨床研究方法病因?qū)W方

1、病因?qū)W研究北京協(xié)和醫(yī)院普通內(nèi)科方病因?qū)W研究前言基本概念研究方法測量指標(biāo)評價原則前言流行病學(xué) 實(shí)驗(yàn)性研究（experimental research） 隨機(jī) 非隨機(jī) 觀察性研究（observational research） 描述性研究(descriptive epidemiology) 分析性研究(analytic epidemiology)前言病因?qū)W研究的重要性 治療 預(yù)防基本概念(cause)a cause is that which produces an effect的分類必要條件(necessary cause)充分條件(sufficient cause)基本概念病因(cause)

2、病因是導(dǎo)致疾病發(fā)生的因素病因的分類充分，且必要21三體唐氏綜合癥充分，但不必要放射線白血病必要，但不充分宮頸癌HPV既不充分，也不必要高血壓冠心病充分，且必要疾病X因素A充分，但不必要或或因素C疾病X因素A因素B必要，但不充分疾病X 因素B + 因素A + 因素C 既不充分，也不必要 因素E +或 因素D +或 因素F +因素C疾病X因素A因素B病因的分類直接病因（近因） 引起疾病的直接因素間接病因（遠(yuǎn)因） 與發(fā)病有間接因素直接病因與間接病因 因素A 疾病X 因素C因素B疾病X因素A(risk factor)流行病學(xué)層次的“病因”使疾病發(fā)生概率(風(fēng)險)增高的因素 因素(factor)人群疾病發(fā)

3、生的概率(risk) 因素(factor)消失人群疾病發(fā)生的概率(risk)Abraham Liliendeld病因模型學(xué)模型疾病因素模型病因模型充分病因集合模型（Rothmans pie）影響健康的因素遺傳政治生物物理醫(yī)療政策行為化學(xué)環(huán)境種族學(xué)模型生物環(huán)境環(huán)境宿主宿主遺傳內(nèi)核致病因子環(huán)境理化環(huán)境輪狀模型流行病學(xué)三角疾病因素模型因素生物學(xué)因素環(huán)境因素醫(yī)學(xué)生物學(xué)因素疾病心理行為因素衛(wèi)生因素遠(yuǎn)因近因病因模型Rothmans pieBerkman. Soc Sci Med. 2000 Sep;51(6):843-57Glass & McAtee. Soc Sci Med 2006;62:16

4、50Healthy People 2010Association is NOT Cause機(jī)遇Chance (random error)偏倚Bias (systematic error)Confounding因果推斷的邏輯方法：Mill準(zhǔn)則求異法（Method of difference）求同法（Method of agreement）共變法（Method of concomitant variation ） 類推法（Method of analogy）排除法（Method of exclusion）因果推斷標(biāo)準(zhǔn)單病因致病假說(Koch, 1882)患者體內(nèi)均能檢出引起該病的病原體(必要病因)

5、其他疾病患者中不能檢出該病原體(效應(yīng)特異性)該病原體能夠從患者體內(nèi)分離培養(yǎng)，接種易感動物能病(充分病因)該能從患病動物體內(nèi)分離到相同病原體Robert Koch多病因致病推斷標(biāo)準(zhǔn)(Bradford-Hill Criteria)時間順序(temporality) Did the exposure precede the disease?關(guān)聯(lián)強(qiáng)度(strength of association) How strongly associated is the putative risk with the outcome of interest?一致性(consistency across inves

6、tigations) Have the results been replicated by different studies, in different settings, by different investigators, and under different conditions?推斷標(biāo)準(zhǔn)(Bradford-Hill Criteria)劑量-效應(yīng)關(guān)系(biological gradient) Are increasing exposures (i.e., dose and duration) associated with increasing risks of disease?

7、試驗(yàn)證據(jù)(experimental evidence) Is there evidence from true experiments in humans?生物可能性(biological plausibility) Is there a credible biological or physical mechanism that can explain the association?推斷標(biāo)準(zhǔn)(Bradford-Hill Criteria)符合性(coherence) Is the association consistent with the natural history and epi

8、demiology of the disease?特異性(specificity) Is the exposure associated with a very specific disease rather than a wide range of diseases?類比性(Analogy) The effect of similar factors may be considered推斷標(biāo)準(zhǔn)(Bradford-Hill Criteria)"None of my nine viewpoints can bringindisputable evidence for or agains

9、t the cause- and-effect hypothesis and none can berequired sine qua non”-HillHill, Austin Bradford (1965). “The environment and disease: association or causation?” Proceedings of the Royal Society of Medicine 58: 295300.推斷標(biāo)準(zhǔn)(Bradford-Hill Criteria)常用Strength Consistency TemporalityBiological gradien

10、t不常用Coherence Experiment Analogy SpecificityPlausibility誤差(error)研究結(jié)果與客觀實(shí)際不符合的地方誤差的分類 隨機(jī)誤差(random error)：機(jī)遇(chance) 隨機(jī)抽樣/測量時隨機(jī)出現(xiàn)的變異 沒有固定的大小和方向，一般呈正態(tài)分布 系統(tǒng)誤差(systematic error)：偏倚(bias) 對象選取/測量/分析時的方法學(xué)缺陷 有固定的大小和方向偏倚選擇偏倚(selection bias) 在選取研究對象的過程中，由于選取方式不當(dāng)導(dǎo)致入選對象與未入選對象之間系統(tǒng)差異 如：志愿者，方便樣本，研究對象不同意/失訪信息偏倚(in

11、formation bias) 由于測量、觀察或資料收集過程中方法問題，使獲得的資料 如：回憶偏倚，系統(tǒng)誤差者偏倚混雜(Confounding)因素與疾病發(fā)生的關(guān)聯(lián)程度受其他因素的干擾ExposureOutcomeConfoundingExposureOutcomeFACTORWhat is this?Directed Acyclic Graphs (DAGs)“DAGs are a set of arrows drawn along atimeline, characterizing causal and temporal relationships between variables. T

12、here can never be a cycle (hence acyclic) because we can never go back in time.”CommuDent Oral Epidemiol 2002;30:399-404.DAGsDirected 沿時間軸走行的一組箭頭 箭頭代表“可能導(dǎo)致” 無箭頭代表相互之間“沒有因果關(guān)系”Acyclic 某因素不能導(dǎo)致自身的出現(xiàn)Graphs 圖示各種因間的關(guān)系DAGs的作用與結(jié)局之間的因果關(guān)系與結(jié)局之間的關(guān)系間的因果關(guān)系用圖代表 有效組織 找出因 找出混雜因素 找出中介因素分析時可避免過度調(diào)整Causation of Bias: The

13、 EpiscopeFIGURE 1 . “The Episcope.” A user of epidemiologic evidence sees a “known” RD (kRD) that results from informationtransmitted, like light rays through a telescope, from a causal (“aetiologic”) RD (aRD) in a target population, through layers of “l(fā)enses” and “filters.” Each layer is a distinct

14、 domain where certain types of biases operate, potentially adding additional distortions to the association of interest. Domains a through k are illustrated in Figures 2-10.Epidemiology. 2001,12(1):114-122.A. The original cause of the image transmitted through the episcope is theassociation (if it e

15、xists) between the causal agent and morbidity.B. Other background factors cloud the association randomly.C. In the source population, the association of interest can be further distorted by correlated causes.None of the above phenomena is observable without diagnoses and exposure measurements, which

16、 are produced as follows:D. Morbidity is a contributing cause of diagnoses, which are recorded in medicalcharts or deaths, as well as self-reported on questionnaires. Thesediagnoses and recordings have varying sensitivities and specificities.E. The occurrence of the causal agent is a contributing ca

17、use of the occurrence of evidence of exposure, which is recorded via interviews, self-administered questionnaires, administrative forms, and other instruments with varying sensitivities and specificities.F. Data on diagnoses and exposures are collated by various means into files anddatabases, usuall

18、y for administrative purposes and sometimes specifically for epidemiologic studies.G. In a descriptive study, the selection usually involves grouping subjects into various exposure levels and examining disease rates. For a cohort study, the database may be used for choosing specific exposure groups

19、to follow.H. Then comes the harvesting of new cases in a cohort study or, analogously, the selection of cases and controls for a case-control study.I. The investigators do many data analyses but submit only their best for publication.J. The journal judges the submitted paper and may decide not to pu

20、blish it.K. Knowledge brokers, such as Cochrane Collaboration meta-analysts, guideline committees, or local experts, help decision makers use the published papers.病因?qū)W研究前言基本概念研究方法（設(shè)計方案）測量指標(biāo)評價原則研究方法描述性研究(descriptive study) 個案報告(case report) 病例系列(case series)發(fā)現(xiàn)線索提出病因假設(shè)分析性研究(analytic study)初步驗(yàn)證病因假設(shè)證實(shí)病因病

21、例對照研究(case-control study)橫斷面研究(cross-sectional study)隊列研究(cohort study)干預(yù)性研究(interventional study)進(jìn)一步驗(yàn)證病因隨機(jī)(randomization)盲法 (blinded)對照(controlled)綜合確定因果關(guān)系研究方法causeSilicone Breast ImplantsRheumatologic DiseasesRCTRexposed= a/(a + b)RR (Relative Risk) =Runexposedc/(c + d )RCT優(yōu)點(diǎn) Randomized Strongest

22、design establishing cause and effect relationship （interventional study）缺點(diǎn) ethical Time consuming Expensive隊列研究Cohort StudyTimeonset of studyDiseaseNo DiseaseDiseaseEligible SubjectsNo DiseaseDirection of InquiryUnexposedExposed隊列研究Cohort StudyRexposed= a/(a + b)RR (Relative Risk) =Runexposedc/(c +

23、d )隊列研究Cohort Study優(yōu)點(diǎn) Direct calculation risk ratio (relative risk) Clear temporal relationship between exposure and disease Minimize bias Strongest observational design establishing cause and effect relationship隊列研究Cohort Study缺點(diǎn) Time consuming Often requires a large sample size Expensive Not effic

24、ient for study of rare diseases Losses to follow-up may diminish validity Changes over time in diagnostic methods may lead to biased results病例對照研究Case-Control StudyTimeonset of studyExposedCasesUnexposedExposedControlsUnexposedDirection of Inquiry病例對照研究Case-Control StudyOddscasea/cOR (Odds Ratio) =O

25、ddscontrolb/d病例對照研究Case-Control Study優(yōu)點(diǎn) Efficient for study of rare diseases Efficient for study of chronic diseases Tend to require smaller sample size Less expensive May be completed more rapidly病例對照研究Case-Control Study缺點(diǎn) Risk of disease cannot be estimated directly Not efficient for study of rare

26、 exposures More susceptible to selection bias Information on exposure may be less accurate病因?qū)W研究前言基本概念研究方法（設(shè)計方案）測量指標(biāo)評價原則Incidence RateIncidence RateCumulative Incidence (Risk)Prevalence (Rate?)RatioIRR = Incidence Rate RatioRR = Relative Risk= Risk Ratio= Cumulative Incidence Ratio OR = Odds RatioPR

27、= Prevalence RatioPOR = Prevalence Odds RatioDifferenceIRD = Incidence Rate DifferenceRD = Cumulative Incidence Difference (Risk Difference)PD = Prevalence Difference隊列研究Cohort StudyTimeonset of studyDiseaseNo DiseaseDiseaseEligible SubjectsNo DiseaseDirection of InquiryUnexposedExposed測量指標(biāo)隊列研究(Coho

28、rt Study): RRs with outcomeRisk =度AllsOutcomeExposedUnexposedTotalDiseaseABNo DiseaseCDA+BC+DTotalA+CB+DA+B+C+D測量指標(biāo)隊列研究(Cohort Study)Riskexposed = Exposeds with outcome =AA + CAll exposedsRiskunexposed= Unexposeds with outcome =BB+ DAll unexposedsRexposed= A/(A + C)RR(Relative Risk) =RunexposedB/(B

29、+ D)相對度測量指標(biāo)隊列研究(Cohort Study)Attributable Risk = Rexposed- Runexposed= A/(A + C) - B/(B + D)歸因度Attributable Risk Percent = Rexposed- Runexposed = A/(A + C) - B/(B + D)A/(A + C)Rexposed歸因比測量指標(biāo)隊列研究(Cohort Study)Outcome吸煙不吸煙Total肺癌24915無肺癌531393194526485084Total533883196085348測量指標(biāo)隊列研究(Cohort Study)吸煙發(fā)生

30、肺癌的為0.47%A249Risk吸煙組 = 0.47%A + C249 + 53139不吸煙發(fā)生肺癌的為0.05%B15Risk不吸煙組 = 0.05%B+ D15 + 31945吸煙發(fā)生肺癌的是不吸煙的9.94倍Rexposed= 0.47% = 9.94度(RR) =相對Runexposed0.05%測量指標(biāo)隊列研究(Cohort Study)度(AR) = Rexposed- Runexposed= 0.47% - 0.05%0.42%歸因?yàn)?.42%吸煙導(dǎo)致肺癌發(fā)生的絕對比(ARP) = Rexposed- Runexposed = 0.47% - 0.05%0.89歸因Rexpos

31、ed0.4789%的肺癌歸因于吸煙測量指標(biāo)節(jié)育器和輸卵管炎？2000女性隊列Rexpose？Runexpose？RR？Outcome用IUD不用IUDTotalyes4515no955985601940Total100010002000的女性 1200有一個RR？的女性 800有超過一個RR？Outcome用IUD不用IUDTotalyes426no6589448752Total700100800Outcome用IUD不用IUDTotalyes39no297891121188Total3009001200設(shè)計方案病例對照研究(Case-Control Study)Timeonset of studyExposedCasesUnexposedExposedControlsUnexposedDirection of Inquiry測量指標(biāo)病例對照研究