醫(yī)學(xué)研究設(shè)計(jì)簡介課件

醫(yī)學(xué)研究設(shè)計(jì)簡介科學(xué)的思考方式演譯法歸納法猜測與否證-KarlPopper

我們雖然無法證明某個(gè)假說為真，卻可以經(jīng)由不斷地否證，除去錯(cuò)誤的假說，如此留下來的假說比較可能趨近為真。1/6/20242DepartmentofMedicalResearch,ChimeiMedicalCenter提出假說推論出小前提研究設(shè)計(jì)並進(jìn)行測量估計(jì)或檢定分析結(jié)果得到研究摘要演繹歸納流行病學(xué)方法論，p11。第三版。王榮德著1/6/20243DepartmentofMedicalResearch,ChimeiMedicalCenter吸煙導(dǎo)致肺癌臺(tái)灣的男子吸煙者比不吸煙者易得肺癌研究設(shè)計(jì)測量的臺(tái)灣男子中，吸煙者與不吸煙者的肺癌死亡率吸煙者之肺癌死亡率為不吸煙者之10倍演繹歸納流行病學(xué)方法論，p11。第三版。王榮德著1/6/20244DepartmentofMedicalResearch,ChimeiMedicalCenter因果關(guān)係的判定標(biāo)準(zhǔn)必要條件—時(shí)序性次要條件—一致性不能用機(jī)會(huì)解釋沒有其他干擾因素合理性參考條件—相關(guān)強(qiáng)度、相關(guān)的特定程度、劑量效應(yīng)關(guān)係、生物學(xué)上的贊同性1/6/20245DepartmentofMedicalResearch,ChimeiMedicalCenterConfoundingeffectexposureoutcomeConfoundingfactorassociationassociationMarkElwood.CriticalAppraisalofEpidemiologicalStudiesandClinicalTrails.2ndEdition.1/6/20246DepartmentofMedicalResearch,ChimeiMedicalCenterTheassociationbetweenO.C.useandMI-overestimationOralcontraceptiveuseMyocardialinfarctionsmoking+O.C.userssmokemoreheavilythannonusers+Smokers,irrespectiveofO.C.use,haveahigherinfarctionthannon-smokersMarkElwood.CriticalAppraisalofEpidemiologicalStudiesandClinicalTrails.2ndEdition.1/6/20247DepartmentofMedicalResearch,ChimeiMedicalCenterTheassociationbetweenO.C.useandMI-underestimationOralcontraceptiveuseMyocardialinfarctionobesity—O.C.usersarelessobesethannon-users+ObesityincreasestheriskofmyocardialinfarctionMarkElwood.CriticalAppraisalofEpidemiologicalStudiesandClinicalTrails.2ndEdition.1/6/20248DepartmentofMedicalResearch,ChimeiMedicalCenter常見的研究設(shè)計(jì)法觀察性研究法Caseseriesstudy世代研究法(cohortstudy)病例對(duì)照研究(casecontrolstudy)斷代研究法(cross-sectionalstudy)試驗(yàn)性研究法社區(qū)試驗(yàn)法臨床試驗(yàn)法1/6/20249DepartmentofMedicalResearch,ChimeiMedicalCenterCohortstudyRecruitedsamplesexposedgr.Non-exposedgr.diseasedgr.Non-diseasedgr.Now!(concurrentcohortstudy)Now!(retrospectivecohortstudy)1/6/202410DepartmentofMedicalResearch,ChimeiMedicalCenterCohortstudy-advantages適於研究rareexposure可直接觀察exposed和non-exposedgroup的疾病發(fā)生率可觀察多重健康事件事件發(fā)生的時(shí)序性、先後次序Eligibilitycriteria和outcomeassessments可以標(biāo)準(zhǔn)化比RCT的行政作業(yè)簡單、便宜1/6/202411DepartmentofMedicalResearch,ChimeiMedicalCenterCohortStudy-disadvantage前瞻性研究，費(fèi)時(shí)、昂貴不適於研究稀有或潛伏期夠長的疾病若是retrospectivecohortstudy，需要有過去exposurestatus的紀(jì)錄Lossoffollowup可能影響效度1/6/202412DepartmentofMedicalResearch,ChimeiMedicalCenterRR(RelativeRisk,RateRatio)MIMIO.C.6040100O.C.445610010496200RR=(60/100)/(44/100)=1.36MIMIO.C.aba+bO.C.cdc+da+cb+da+b+c+dRR=(a/(a+b))/(c/(c+d))1/6/202413DepartmentofMedicalResearch,ChimeiMedicalCenter0<=RR<∞0<=RR<1,protectionfactorRR=1,noeffectRR>1,riskfactorRR(RelativeRisk,RateRatio)1/6/202414DepartmentofMedicalResearch,ChimeiMedicalCenterCase-controlstudyRecruitedsamplesexposedgr.Non-exposedgr.diseasedgr.Non-diseasedgr.Now!1/6/202415DepartmentofMedicalResearch,ChimeiMedicalCenterSelectionofstudysubjectCases:一或多個(gè)醫(yī)院中就醫(yī)的所有病例整個(gè)社區(qū)或團(tuán)體族群的所有病例Control:整個(gè)社區(qū)醫(yī)院其他病人病人親屬病人的同學(xué)、同事、鄰居1/6/202416DepartmentofMedicalResearch,ChimeiMedicalCenterCase-ControlStudy-Advantage較適用於研究稀有或潛伏期長的疾病較快得到結(jié)果，較便宜保留探討多種致病因子的可能性1/6/202417DepartmentofMedicalResearch,ChimeiMedicalCenterMatching-控制干擾因子的一種方法Frequencymatching要有整個(gè)族群的資料Individualmatching找符合條件裡頭最接近病例的人Cases是陸續(xù)收案時(shí)先考慮age,gender,race，避免overmatching1/6/202418DepartmentofMedicalResearch,ChimeiMedicalCenterCase-ControlStudy-disadvantageExposurestatus需要靠過去的紀(jì)錄或回憶bias：recallbias,selectionbias不容易研究多重疾病事件，且疾病事件的時(shí)序性不易建立Controlgroup不好選Confoundingeffect1/6/202419DepartmentofMedicalResearch,ChimeiMedicalCenterOR(OddsRatio)MIMISmkaba+bsmkcdc+da+cb+da+b+c+dodds1=a/b

odds2=c/dOddsRatio=odds1/odds2=ad/bcMIMIsmk30200230smk1535036545550595OR=30×350/200×15=3.51/6/202420DepartmentofMedicalResearch,ChimeiMedicalCenterOR(OddsRatio)0<=OR<∞0<=OR<1,protectionfactorOR=1,noeffectOR>1,riskfactor1/6/202421DepartmentofMedicalResearch,ChimeiMedicalCenterCross-sectionalstudyexposedgr.Non-exposedgr.diseasedgr.Non-diseasedgr.1/6/202422DepartmentofMedicalResearch,ChimeiMedicalCenterCross-SectionalStudy-Advantage簡單便宜容易推廣到參考族群1/6/202423DepartmentofMedicalResearch,ChimeiMedicalCenterCross-SectionalStudy-disadvantage頂多建立相關(guān)(association),無法推論因果關(guān)係（causal-effectrelationship）注意selectivesurvivalbiasConfoundingeffect1/6/202424DepartmentofMedicalResearch,ChimeiMedicalCenterClinicalTrial

RCT(Randomizedcontrolledtrial)ParalleldesignCross-Overdesign1/6/202425DepartmentofMedicalResearch,ChimeiMedicalCenterRCT

ParallelDesignRecruitedsamplesTreatmentATreatmentBOutcome1Outcome2Randomization1/6/202426DepartmentofMedicalResearch,ChimeiMedicalCenterRCT

Cross-OverDesignRecruitedsamplesTreatmentATreatmentBOutcome1Outcome2TreatmentBTreatmentARandomization1/6/202427DepartmentofMedicalResearch,ChimeiMedicalCenterRandomizationControlledBlind-DesignIITorPP?1/6/202428DepartmentofMedicalResearch,ChimeiMedicalCenterBlindDesignStudysubjectInvestigatoranalyzerSingleBlindDoubleBlindTripleBlind○○○○○○1/6/202429DepartmentofMedicalResearch,ChimeiMedicalCenterIntentiontotreatrandomizationclofibraten=1065placebon=2095compliantn=708compliantn=1813Non-compliantn=882Non-compliantn=357TreatmentacceptedTreatmentallocatedPercentmortality15.024.615.128.2All18.2All19.41/6/202430DepartmentofMedicalResearch,ChimeiMedicalCenterAnalysisstrategy:IITorPP？IIT(intention-to-treat)依照隨機(jī)分派當(dāng)時(shí)的名單作資料分析，包括沒有完全服從治療策略以及沒有完成試驗(yàn)者。PP(perprotocol)只納入完全遵從治療分組及治療策略者，作資料分析。又稱作“Ontreatmentanalysis”1/6/202431DepartmentofMedicalResearch,ChimeiMedicalCenterRCT-advantageConfounders可能接近平衡分佈可以建立因果時(shí)序性Randomization的程序可能使統(tǒng)計(jì)分析較單純1/6/202432DepartmentofMedicalResearch,ChimeiMedicalCenterRCT-disadvantage昂貴、費(fèi)時(shí)注意volunteerbias注意醫(yī)療倫理的問題1/6/202433DepartmentofMedicalResearch,ChimeiMedicalCenterIEvidenceobtainedfromatleastoneproperlyrandomized,controlledtrialII-1Evidenceobtainedfromwell-designedcontrolledtrialswithoutrandomizationII-2Evidenceobtainedfromwell-designedcohortorcase-controlanalyticstudies,preferablyfrommoretheonecenterorresearchgroup.II-3Evidenceobtainedfrommultipletimeserieswithorwithouttheintervention.Dramaticresultsoftheuncontrolledexperimentscouldalsoberegardedasthistypeofevidence.IIIOpinionsofrespectedauthorities,basedonclinicalexperience;descriptivestudiesandcasereports;orreportsofexpertcommittees.GradesofEvidenceforthePurportedQualityofStudyDesign

(NEnglJMed2000;342:1887-92)1/6/202434DepartmentofMedicalResearch,ChimeiMedicalCenterMisclassification（錯(cuò)分）Cohortstudy診斷標(biāo)準(zhǔn)的效度不足疾病組與非疾病組的分類錯(cuò)誤Case-Controlstudy暴露(exposure)的分組的效度不足暴露組與非暴露組的分類錯(cuò)誤1/6/202435DepartmentofMedicalResearch,ChimeiMedicalCenterMisclassificationDifferentialmisclassification結(jié)論的偏差方向難以估計(jì)Non-Differentialmisclassification導(dǎo)致低估了「暴露」與「疾病」的相關(guān)性(towardthenull,H0)Ex:RR=2→RR=1.51/6/202436DepartmentofMedicalResearch,ChimeiMedicalCenterConfoundingeffectexposureoutcomeConfoundingfactorassociationassociationMarkElwood.CriticalAppraisalofEpidemiologicalStudiesandClinicalTrails.2ndEdition.1/6/202437DepartmentofMedicalResearch,ChimeiMedicalCenterTheassociationbetweenO.C.useandMI-overestimationOralcontraceptiveuseMyocardialinfarctionsmoking+O.C.userssmokemoreheavilythannonusers+Smokers,irrespectiveofO.C.use,haveahigherinfarctionthannon-smokersMarkElwood.CriticalAppraisalofEpidemiologicalStudiesandClinicalTrails.2ndEdition.1/6/202438DepartmentofMedicalResearch,ChimeiMedicalCenterTheassociationbetweenO.C.useandMI-underestimationOralcontraceptiveuseMyocardialinfarctionobesity—O.C.usersarelessobesethannon-users+ObesityincreasestheriskofmyocardialinfarctionMarkElwood.CriticalAppraisalofEpidemiologicalStudiesandClinicalTrails.2ndEdition.1/6/202439DepartmentofMedicalResearch,ChimeiMedicalCenterConfoundingeffectMIMIsmk202040Non-smk2810222850MIMIsmk104050Non-smk89210018132150MIMIO.C.222850O.C.1813215040160200RR=(22/50)/(18/150)=3.67O.C.組抽煙率80％Non-O.C.組抽煙率33.3％CrudeRR1/6/202440DepartmentofMedicalResearch,ChimeiMedicalCenterAdjustingconfoundingeffectsmkMIMIO.C202040O.C.104050306090Non-smkMIMIO.C.2810O.C.89210010100110RR=(20/40)/(10/50)=2.5RR=(20/40)/(8/92)=2.5Adjusted-RR1/6/202441DepartmentofMedicalResearch,ChimeiMedicalCenterConfoundingeffect1/6/202442DepartmentofMedicalResearch,ChimeiMedicalCenter控制干擾效應(yīng)的方法研究設(shè)計(jì)階段：Restriction限制條件Matching配對(duì)選樣Randomization隨機(jī)分派資料分析階段：Stratification分層分析Multivariatemodel(adjust)多元模型建構(gòu)1/6/202443DepartmentofMedicalResearch,ChimeiMedicalCenterInteraction(effectmodification)smkMIMIO.C301040O.C.54550355590Non-smkMIMIO.C.2810O.C.89210010100110RR=(30/40)/(5/50)=7.5RR=(2/10)/(8/100)=2.5CigarettesmokingisaneffectmodifierfortheassociationbetweenMIandO.C.use.1/6/202444DepartmentofMedicalResearch,ChimeiMedicalCenter交互作用1/6/202445DepartmentofMedicalResearch,ChimeiMedicalCenter交互作用(interaction)使用O.C.和MI發(fā)生率的相關(guān)性，隨著抽樣狀態(tài)不同而有不同。抽煙者，服用O.C.的風(fēng)險(xiǎn)是不服用者的7.5倍；不抽煙者，服用O.C.的風(fēng)險(xiǎn)是不服用者的2.5倍X和Y之間的相關(guān)性，受到Z的修飾作用；X和Z對(duì)於Y的相關(guān)性，有交互作用。X和Y是否相關(guān)？不能一言以蔽之，需依Z的不同狀態(tài)，分別討論之。1/6/202446DepartmentofMedicalResearch,ChimeiMedicalCenter下次議題：線性迴歸分析與相關(guān)分析1/6/202447DepartmentofMedicalResearch,ChimeiMedicalCenter1、病毒性肝炎：由病毒造成的肝炎按照其病毒系列不同分為甲、乙、丙、丁、戊和庚共六種類型病毒性肝炎。能引起肝臟細(xì)胞腫脹，是世界上流傳廣泛，危害很大的傳染病之一。

1908年，才發(fā)現(xiàn)病毒也是肝炎的致病因素之一。1947年，將原來的傳染性肝炎（infectioushepatitis）稱為甲型肝炎（HepatitisA,HA）；血清性肝炎（serumhepatitis）稱為乙型肝炎（HepatitisB,HB）。1965年人類首次檢測到乙型肝炎的表面抗原。我國經(jīng)濟(jì)和科學(xué)技術(shù)日益發(fā)展，學(xué)術(shù)文化領(lǐng)域百家爭鳴，（df高血壓958心臟病983u6糖尿病87fr）特別是思想家的革新精神，為中醫(yī)學(xué)理論的創(chuàng)新和突破性進(jìn)展，提供了有利的文化背景。宋代陳無擇著《三因極一病證方論》一書，（45傳染病q566丙肝964jo乙肝28jgsx甲肝gh）提出三因?qū)W說；并產(chǎn)生了最具盛名四大學(xué)派，劉完素倡導(dǎo)火熱論；張從正力倡“攻邪論”；李杲提出“內(nèi)傷脾胃，百病由生”的理論；朱震亨創(chuàng)造性地闡明了相火的演變規(guī)律。編輯本段明清時(shí)期（df肺25s血液f369血小板t5172紅血球gdf55m白血球fd2）是中醫(yī)學(xué)理論綜合匯編、深化發(fā)展，臨床各科辨證體系豐富、提高階段。如明代樓英的《醫(yī)學(xué)綱目》和王肯堂的《證治準(zhǔn)繩》，清代吳謙等編著的《醫(yī)宗金鑒》和陳夢(mèng)雷主編的《古今圖書集成·醫(yī)部全錄》等。王清任著《醫(yī)林改錯(cuò)》，注重實(shí)證研究，（df高血壓958心臟病983u6糖尿病87fr）糾正了古醫(yī)籍中關(guān)于解剖知識(shí)的某些錯(cuò)誤，肯定了“腦主思維”，發(fā)展了瘀血理論。溫病學(xué)說的形成和發(fā)展，標(biāo)志著中醫(yī)理論的創(chuàng)新與突破，吳有性著《溫疫論》，葉天士著《溫?zé)岵∑?，吳鞠通著《溫病條辨》等，在藥物學(xué)研究方面，（45傳染病q566丙肝964jo乙肝28jgsx甲肝gh）李時(shí)珍著的《本草綱目》，總結(jié)了16世紀(jì)以前我國藥物學(xué)研究的成就。醫(yī)的診察疾病能參考現(xiàn)代醫(yī)學(xué)的微觀分析，將辨證與辨病相結(jié)合，實(shí)現(xiàn)宏觀與微觀的統(tǒng)一，使中醫(yī)診斷客觀化，即把分析與綜合相結(jié)合的方法引入中醫(yī)理、法、方、藥的研究，使二者有機(jī)結(jié)合，互相借鑒、補(bǔ)充，避免各自的片面性、局限性，這將有利于中西醫(yī)學(xué)的優(yōu)勢(shì)互補(bǔ)