臨床實(shí)踐證據(jù)

如何形成可回答的臨床問題(PICO)背景問題(Backgroundquestions):意指與疾病相關(guān)的知識(shí)性問題，問題的結(jié)構(gòu)通常包括有帶有動(dòng)詞的疑問句(who、what、where、when、how、why)與疾病。Who：病患特性、危險(xiǎn)因子等What：自然病程、癥狀學(xué)等Where:影響的器官與系統(tǒng)、疾病分布的區(qū)域等When:疾病好發(fā)的年齡、時(shí)間、季節(jié)等How：病理生理學(xué)Why：病因舉例：急性心肌梗塞Who：老年人、男性、家族史What：胸痛、心電圖的變化等等Where：冠狀動(dòng)脈阻塞的位置、程度等等When:氣溫低、心肌耗氧量高之活動(dòng)等How:動(dòng)脈硬化斑塊破裂、血栓形成Why：動(dòng)脈硬化危險(xiǎn)因子、糖尿病、高血壓、高血脂癥、抽菸，肥胖，少運(yùn)動(dòng)前景問題(Foregroundquestions):指與治療病患相關(guān)的特殊性問題，問題的架構(gòu)通常包括有4個(gè)項(xiàng)目：與病患相關(guān)的問題(Thepatientand/orproblemofinterest)主要的治療措施，包括診斷、治療、預(yù)后、及病患的觀點(diǎn)(Themainintervention(definedverybroadly,includinganexposure,adiagnostictest,aprognosticfactor,atreatment,apatientperception,andsoforth))比較相關(guān)的治療措施(Comparisonintervention(s),ifrelevant)相關(guān)的臨床成效(Theclinicaloutcome(s)ofinterest)PICO問題P：patientand/orproblem(病患)I:intervention(處理)C：comparisonofintervention(對(duì)照)O：clinicaloutcome(臨床結(jié)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實(shí)證醫(yī)學(xué)相關(guān)名詞診斷（Diagnosis）:假陽性：是指健康的人診斷試驗(yàn)結(jié)果為不正常，如同無辜的人假陰性：是指有病的人診斷試驗(yàn)結(jié)果為正常，如同消遙法外的歹徒將真陽性、假陽性、彳殳陰性、真陰性分別以a,b,c,d來表示Disease(+)生病Disease(-)健康TestResult(+)陽性a真陽性b假陽性a+bTestResult(-)陰性c假陰性d真陰性c+da+cb+d?Sensitivity（敏感度）：為有病者診斷結(jié)果為陽性的比率=真陽性率=真陽性/生病=a/a+c當(dāng)高靈敏診斷試驗(yàn)的結(jié)果為陰性，此為未罹患此疾病相當(dāng)可靠的指標(biāo)Specificity（特異度）：為沒病者診斷結(jié)果為陰性的比率=真陰性率=真陰性/健康=d/b+d在專一性高的診斷試驗(yàn)，結(jié)果陽性即表有病，因?yàn)楹币妭侮栃訮ositivePredictiveValue,PPV（陽性預(yù)測(cè)值）：診斷試驗(yàn)結(jié)果呈現(xiàn)陽性且確實(shí)有病者的比率=真陽性/陽性試驗(yàn)結(jié)果=a/a+bNegativePredictiveValue,NPV（陰性預(yù)測(cè)值）：診斷試驗(yàn)結(jié)果呈陰性且確實(shí)無患病者的比率=真陰性/陰性試驗(yàn)結(jié)果=d/c+dLikelihoodRatios(相似比)分子：疾病中診斷試驗(yàn)(陽性或陰性)比率分母：無疾病中診斷試驗(yàn)(陽性或陰性)比率LR(+)=Pr{T+/D+}/Pr{T+/D-}=真陽性率/假陽性率=Sensitivity/(1-Specificity)=(a/a+c)/(b/b+d)LR(-)=Pr{T-/D+}/Pr{T-/D-}=假陰性率/真陰性率=(1-Sensitivity)/Specificity=(c/a+c)/(d/b+d)LikelihoodRatios(相似比)數(shù)值所代表的臨床意義LikelihoodRatioInterpretation>10Strongevidencetoruleindisease5-10Moderateevidencetoruleindisease2-5Weakevidencetoruleindisease0.5-2.0Nosignificantchangeinthelikelihood0.2-0.5Weakevidencetoruleoutdiseas0.1-0.2Moderateevidencetoruleoutdisease<0.1Strongevidencetoruleoutdisease處置(Therapy):對(duì)照組事件發(fā)生率(CER,ControlEventRrate)實(shí)驗(yàn)組事件發(fā)生率(EER,ExperimentalEventRrate)相對(duì)風(fēng)險(xiǎn)比率差(RRR,RelativeRiskReduction)=|EER-CER|/CER接受治療組比未接受治療的對(duì)照組間，不良結(jié)果機(jī)率下降之比例，伴隨95%信賴區(qū)間(CI)Theproportionalreductioninratesofbadoutcomesbetweenexperimentalandcontrolparticipantsinatrial,andaccompaniedbya95%confidenceinterval(CI)絕對(duì)風(fēng)險(xiǎn)比率差(ARR,AbsoluteRiskReduction)=|EER-CER|治療組與對(duì)照組間不良結(jié)果機(jī)率差的絕對(duì)值，伴隨95%信賴區(qū)間(CI)AbsoluteRiskReduction(ARR)isthedifferenceintheeventratebetweencontrolgroup(CER)andtreatedgroup(EER).需要被治療的病人數(shù)目(NNT,NumberNeededtoTreat)=1/ARR為減少一個(gè)不良結(jié)果所需治療的病人數(shù)目，伴隨95%信賴區(qū)間(CI)Thenumberofpatientswhoneedtobetreatedtopreventonebadoutcome,andaccompaniedbya95%confidenceinterval(CI).需要被治療的病人數(shù)目(NNH,NumberNeededtoHarm)與對(duì)照組病患相比，接受實(shí)驗(yàn)性治療導(dǎo)致額一位病人被傷害的病人數(shù)目，伴隨95%信賴區(qū)間(CI),計(jì)算方式與NNT相同Thenumberofpatientswho,iftheyreceivedtheexperimentaltreatment,wouldresultinoneadditionalpatientbeingharmed,comparedwithpatientswhoreceivedthecontroltreatment,andaccompaniedbya95%CI.傷害(Harm):相對(duì)危險(xiǎn)性(RelativeRisk,RR)=EER/CER(在隨機(jī)試驗(yàn)與世代研究中)接受治療病人相對(duì)于未接受治療病人的不良事件風(fēng)險(xiǎn)RelativeRiskistheratioofriskinthetreatedgroup(EER)totheriskinthecontrolgroup(CER).RRisusedinrandomisedtrialsandcohortstudies.勝算(Odds)發(fā)生某事件的人數(shù)與未發(fā)生該事件人數(shù)的比值Odds：aratioofthenumberofpeopleincurringaneventtothenumberofpeoplewhodon'thaveanevent.勝算比(OddsRatio,OR)(在病例對(duì)照研究中)實(shí)驗(yàn)組中發(fā)生疾病的勝算與控制組中發(fā)生疾病的勝算比值，或罹患疾病的病患暴露于某變因的勝算除以控制組暴露的勝算OddsRatioistheratiooftheoddsofhavingthetargetdisorderintheexperimentalgrouprelativetotheoddsinfavourofhavingthetargetdisorderinthecontrolgrouportheoddsinfavourofbeingexposedinsubjectswiththetargetdisorderdividedbytheoddsinfavourofbeingexposedincontrolsubjects(withoutthetargetdisorder).信賴區(qū)間(ConfidenceInterval,CI)有95%的信心確定，群體的正確數(shù)值會(huì)落在這個(gè)數(shù)值范圍內(nèi)Quantifiestheuncertaintyinmeasurement.Itisusuallyreportedasa95%CIwhichistherangeofvalueswithinwhichwecanbe95%surethatthetruevalueforthewholepopulationlies.暴露與不良結(jié)果的相關(guān)性RR或OR=1表示無論有無暴露于假設(shè)因子中，發(fā)生不良結(jié)果的可能性一樣RR或OR>1表示暴露于假設(shè)因子中導(dǎo)致不良結(jié)果的風(fēng)險(xiǎn)增加RR或OR<1表示暴露于假設(shè)因子者比未暴露于假設(shè)因子者更不可能發(fā)生不良結(jié)果病例對(duì)照研究(Case-controlstudy)的偏差(bias)較多，因此當(dāng)OR>4較有意義.世代研究(Cohortstudy)較嚴(yán)謹(jǐn)，但仍會(huì)有偏差(bias)存在，RR>3時(shí)較有意義?除考慮RR與OR的數(shù)值大小，必須由信賴區(qū)間(Confidenceinterval，CI)來確認(rèn)其值地準(zhǔn)確度.當(dāng)信賴區(qū)間越窄，結(jié)果準(zhǔn)確度越高.示范:Adverse(Disease)eventoccursAdverseeventDisease)doe