临床实践证据

如何形成可回答的临床问题(PICO)背景问题(Backgroundquestions):意指与疾病相关的知识性问题，问题的结构通常包括有带有动词的疑问句(who、what、where、when、how、why)与疾病。Who：病患特性、危险因子等What：自然病程、症状学等Where:影响的器官与系统、疾病分布的区域等When:疾病好发的年龄、时间、季节等How：病理生理学Why：病因举例：急性心肌梗塞Who：老年人、男性、家族史What：胸痛、心电图的变化等等Where：冠状动脉阻塞的位置、程度等等When:气温低、心肌耗氧量高之活动等How:动脉硬化斑块破裂、血栓形成Why：动脉硬化危险因子、糖尿病、高血压、高血脂症、抽菸，肥胖，少运动前景问题(Foregroundquestions):指与治疗病患相关的特殊性问题，问题的架构通常包括有4个项目：与病患相关的问题(Thepatientand/orproblemofinterest)主要的治疗措施，包括诊断、治疗、预后、及病患的观点(Themainintervention(definedverybroadly,includinganexposure,adiagnostictest,aprognosticfactor,atreatment,apatientperception,andsoforth))比较相关的治疗措施(Comparisonintervention(s),ifrelevant)相关的临床成效(Theclinicaloutcome(s)ofinterest)PICO问题P：patientand/orproblem(病患)I:intervention(处理)C：comparisonofintervention(对照)O：clinicaloutcome(临床结果)形成臨床問題一PICO|ComponentsofClinicalQuestionsfPatient\PopulationJIntervention/ExposureComparisonInpatientswithd口日呂early七止皿-comparedtoa^cuteMlmeritwfthastatinplaceboInwomenwkthv/hatfsthecomparedtosuspectedaccuracyorexercisecoronarydiseaseexerciseECHOECGInpost-doeshormonecompainedtonomenopausalreplacementHRTwomentherapyfordiagnosingsignificantCAD7increasetheriskofbreastcancer?decreasecardiovascularmortality?OutcomeLevelsofEvidenceLevelTherapy/Prevention,Aetiology/HarmDiagnosis1aSystematicReview(SR)ofRCTsSRofLevel1diagnosticstudies;ClinicalDecisionRule(CDR)with1bstudiesfromdifferentclinicalcentres1bIndividualRandomizedcontrolledtrial(RCT)Validatingcohortstudywithgoodreferencestandards;orCDRtestedwithinoneclinicalcentre1cAll-or-noneAbsoluteSpPinsandSnNouts2aSystematicreviewofcohortstudiesSRofLevel>2diagnosticstudies2bIndividualcohortstudyExploratorycohortstudywithgoodreferencestandards;CDRafterderivation,orvalidatedonlyonsplit—sampleordatabases2c"Outcomes"Research;Ecologicalstudies3aSystematicreviewofcase-controlstudiesSRof3bandbetterstudies3bIndividualCase-ControlStudyNon-consecutivestudy;orwithoutconsistentlyappliedreferencestandards4Case-seriesCase-controlstudy,poorornon-independentreferencestandard5Expertopinionwithoutexplicitcriticalappraisal,orbasedonphysiology,benchresearchor"firstprinciples"Expertopinionwithoutexplicitcriticalappraisal,orbasedonphysiology,benchresearchor"firstprinciples"LevelsofEvidence(March2009)GradesofRecommendationsAConsistentlevel1studiesBConsistentlevel2or3studiesorextrapolationsfromlevel1studiesClevel4studiesorextrapolationsfromlevel2or3studiesDlevel5evidenceortroublinglyinconsistentorinconclusivestudiesofanylevel实证医学相关名词诊断（Diagnosis）:假阳性：是指健康的人诊断试验结果为不正常，如同无辜的人假阴性：是指有病的人诊断试验结果为正常，如同消遥法外的歹徒将真阳性、假阳性、彳殳阴性、真阴性分别以a,b,c,d来表示Disease(+)生病Disease(-)健康TestResult(+)阳性a真阳性b假阳性a+bTestResult(-)阴性c假阴性d真阴性c+da+cb+d•Sensitivity（敏感度）：为有病者诊断结果为阳性的比率=真阳性率=真阳性/生病=a/a+c当高灵敏诊断试验的结果为阴性，此为未罹患此疾病相当可靠的指标Specificity（特异度）：为没病者诊断结果为阴性的比率=真阴性率=真阴性/健康=d/b+d在专一性高的诊断试验，结果阳性即表有病，因为罕见伪阳性PositivePredictiveValue,PPV（阳性预测值）：诊断试验结果呈现阳性且确实有病者的比率=真阳性/阳性试验结果=a/a+bNegativePredictiveValue,NPV（阴性预测值）：诊断试验结果呈阴性且确实无患病者的比率=真阴性/阴性试验结果=d/c+dLikelihoodRatios(相似比)分子：疾病中诊断试验(阳性或阴性)比率分母：无疾病中诊断试验(阳性或阴性)比率LR(+)=Pr{T+/D+}/Pr{T+/D-}=真阳性率/假阳性率=Sensitivity/(1-Specificity)=(a/a+c)/(b/b+d)LR(-)=Pr{T-/D+}/Pr{T-/D-}=假阴性率/真阴性率=(1-Sensitivity)/Specificity=(c/a+c)/(d/b+d)LikelihoodRatios(相似比)数值所代表的临床意义LikelihoodRatioInterpretation>10Strongevidencetoruleindisease5-10Moderateevidencetoruleindisease2-5Weakevidencetoruleindisease0.5-2.0Nosignificantchangeinthelikelihood0.2-0.5Weakevidencetoruleoutdiseas0.1-0.2Moderateevidencetoruleoutdisease<0.1Strongevidencetoruleoutdisease处置(Therapy):对照组事件发生率(CER,ControlEventRrate)实验组事件发生率(EER,ExperimentalEventRrate)相对风险比率差(RRR,RelativeRiskReduction)=|EER-CER|/CER接受治疗组比未接受治疗的对照组间，不良结果机率下降之比例，伴随95%信赖区间(CI)Theproportionalreductioninratesofbadoutcomesbetweenexperimentalandcontrolparticipantsinatrial,andaccompaniedbya95%confidenceinterval(CI)绝对风险比率差(ARR,AbsoluteRiskReduction)=|EER-CER|治疗组与对照组间不良结果机率差的绝对值，伴随95%信赖区间(CI)AbsoluteRiskReduction(ARR)isthedifferenceintheeventratebetweencontrolgroup(CER)andtreatedgroup(EER).需要被治疗的病人数目(NNT,NumberNeededtoTreat)=1/ARR为减少一个不良结果所需治疗的病人数目，伴随95%信赖区间(CI)Thenumberofpatientswhoneedtobetreatedtopreventonebadoutcome,andaccompaniedbya95%confidenceinterval(CI).需要被治疗的病人数目(NNH,NumberNeededtoHarm)与对照组病患相比，接受实验性治疗导致额一位病人被伤害的病人数目，伴随95%信赖区间(CI),计算方式与NNT相同Thenumberofpatientswho,iftheyreceivedtheexperimentaltreatment,wouldresultinoneadditionalpatientbeingharmed,comparedwithpatientswhoreceivedthecontroltreatment,andaccompaniedbya95%CI.伤害(Harm):相对危险性(RelativeRisk,RR)=EER/CER(在随机试验与世代研究中)接受治疗病人相对于未接受治疗病人的不良事件风险RelativeRiskistheratioofriskinthetreatedgroup(EER)totheriskinthecontrolgroup(CER).RRisusedinrandomisedtrialsandcohortstudies.胜算(Odds)发生某事件的人数与未发生该事件人数的比值Odds：aratioofthenumberofpeopleincurringaneventtothenumberofpeoplewhodon'thaveanevent.胜算比(OddsRatio,OR)(在病例对照研究中)实验组中发生疾病的胜算与控制组中发生疾病的胜算比值，或罹患疾病的病患暴露于某变因的胜算除以控制组暴露的胜算OddsRatioistheratiooftheoddsofhavingthetargetdisorderintheexperimentalgrouprelativetotheoddsinfavourofhavingthetargetdisorderinthecontrolgrouportheoddsinfavourofbeingexposedinsubjectswiththetargetdisorderdividedbytheoddsinfavourofbeingexposedincontrolsubjects(withoutthetargetdisorder).信赖区间(ConfidenceInterval,CI)有95%的信心确定，群体的正确数值会落在这个数值范围内Quantifiestheuncertaintyinmeasurement.Itisusuallyreportedasa95%CIwhichistherangeofvalueswithinwhichwecanbe95%surethatthetruevalueforthewholepopulationlies.暴露与不良结果的相关性RR或OR=1表示无论有无暴露于假设因子中，发生不良结果的可能性一样RR或OR>1表示暴露于假设因子中导致不良结果的风险增加RR或OR<1表示暴露于假设因子者比未暴露于假设因子者更不可能发生不良结果病例对照研究(Case-controlstudy)的偏差(bias)较多，因此当OR>4较有意义.世代研究(Cohortstudy)较严谨，但仍会有偏差(bias)存在，RR>3时较有意义•除考虑RR与OR的数值大小，必须由信赖区间(Confidenceinterval，CI)来确认其值地准确度.当信赖区间越窄，结果准确度越高.示范:Adverse(Disease)eventoccursAdverseeventDisease)doe