循證醫(yī)學(xué)EBM的過(guò)去現(xiàn)在與未來(lái)課件

實(shí)證醫(yī)學(xué)

EBM

過(guò)去、現(xiàn)在與未來(lái)林口長(zhǎng)庚醫(yī)院

實(shí)證醫(yī)學(xué)中心余光輝醫(yī)師MedicalEducationintheNewCentury醫(yī)學(xué)人文教育BioinformaticsPatient-centeredcareProblem-basedlearningEvidence-basedmedicine實(shí)證醫(yī)學(xué)

(EBM,Evidence-basedMedicine)以流行病學(xué)和統(tǒng)計(jì)學(xué)的方法，從龐大的醫(yī)學(xué)資料庫(kù)中嚴(yán)格評(píng)讀、綜合分析並找出值得信賴(lài)的部分，並將所能獲得的最佳文獻(xiàn)證據(jù)，應(yīng)用於臨床工作中，使病人得到最佳的照顧。

Dr.SydneyBurwell,DeanofHarvardMedicalSchool

Halfofwhatyouaretaughtasmedicalstudentswillintenyearshavebeenshowntobewrong.Andthetroubleis,noneofyourteachersknowswhichhalf.

※Useofcurrentbestevidenceinmakingdecisionsaboutthecareofindividualpatients※Notonlyaskillbutalsoanattitudechange※Conscientious,explicit,andjudicioususeofcurrentbestevidenceinmakingdecisionsaboutindividualpatients.

~ArchieCochrane1972EBM實(shí)證醫(yī)學(xué)的實(shí)施醫(yī)師在處置病人所面臨的問(wèn)題時(shí)有許多不確定性，這種不確定性常以機(jī)率來(lái)表示。機(jī)率的估計(jì)，可以來(lái)自個(gè)人的經(jīng)驗(yàn)，但是免不了有某種程度的偏差，因此我們必須借重客觀可信的臨床研究結(jié)果，來(lái)作為照顧病人的主要依據(jù)。實(shí)證醫(yī)學(xué)是兼具科學(xué)方法，及人文特質(zhì)的診療模式，且以病人為中心，以問(wèn)題為導(dǎo)向的精神，正確的應(yīng)用可以提昇整體的醫(yī)療品質(zhì)。隨著資訊科技的發(fā)展，醫(yī)師與病人可以很容易的從公正、可靠的資訊中獲得以實(shí)證為基礎(chǔ)的相關(guān)訊息。為了要跟得上醫(yī)學(xué)最新的腳步及病人被告知的權(quán)利，提供高品質(zhì)的醫(yī)療照護(hù)，實(shí)證醫(yī)學(xué)應(yīng)為醫(yī)學(xué)教育的核心原則。Step1.Convertingtheneedforinformation(aboutprevention,diagnosis,prognosis,therapy,causation,etc.)intoananswerablequestion.Step2.Searchingthebestevidencewithwhichtoanswerthatquestion.Step3.

Criticallyappraisingtheevidenceforitsvalidity(closenesstothetruth),impact(sizeoftheeffect),andapplicability(usefulnessinourclinicalpractice).Step4.

Integrating

theevidencewithourclinicalexpertiseandpatients’uniquebiology,valuesandcircumstances.Step5.Evaluatingoureffectivenessandefficiencyinexecutingsteps1-4andseekingwaystoimprovethembothfornexttime.實(shí)證醫(yī)學(xué)的五大進(jìn)行步驟FiveStepstoPracticeEBM

TheEvidencePyramidAnimalresearchInvitro(testtube)researchCaseseries/ReportsIdeas,Editorials,OpinionsCaseControlStudiesCohortstudiesRandomizedControlledStudiesRandomizedControlledDoubleBlindStudies

Meta-analysis實(shí)證醫(yī)學(xué)的五個(gè)進(jìn)行步驟由個(gè)案的臨床資料尋找能解答、且具體的臨床問(wèn)題。﹝askingananswerablequestion﹞尋找相關(guān)的醫(yī)學(xué)證據(jù)﹝包括各種文獻(xiàn)及醫(yī)學(xué)資料庫(kù)，發(fā)表及未發(fā)表的研究成果﹞。﹝searchingthebestevidence﹞評(píng)估各種醫(yī)學(xué)報(bào)告的可信度及臨床相關(guān)性，並作成結(jié)論。﹝criticalappraisaltheevidence﹞將結(jié)論應(yīng)用於實(shí)際患者的治療。﹝臨床應(yīng)用﹞評(píng)估治療的效果。﹝evaluation﹞Patient~Whoisthepatientorwhatistheproblembeingaddressed?Intervention~Whatistheintervention?Comparison~Whatarethealternatives?Outcome~Whataretheoutcomes?Therearefourelementsofawell-formulatedquestionSearchingTheBestEvidence

尋找最佳實(shí)證資料一是研究論文資料庫(kù)(primaryjournalsordatabases)~如Medline,NEJM,Lancet…或是直接使用實(shí)證醫(yī)學(xué)資料庫(kù)(secondaryjournalsordatabases)~ACPjournalclub,Cochrane.盡量搜尋與病人問(wèn)題相同且證據(jù)等級(jí)(levelofevidence)較高之文獻(xiàn)，再謹(jǐn)慎的評(píng)讀與評(píng)估其在此人問(wèn)題的適用性。實(shí)證醫(yī)學(xué)資料庫(kù)查詢(xún)的優(yōu)點(diǎn)文獻(xiàn)資料可經(jīng)由網(wǎng)路快速查詢(xún)，最新的醫(yī)學(xué)進(jìn)展可以立即見(jiàn)於網(wǎng)路，比傳統(tǒng)教科書(shū)更update。以實(shí)證醫(yī)學(xué)特有的文獻(xiàn)格式，表達(dá)資料取得方法及重要結(jié)果，醫(yī)師不須花很多時(shí)間便能了解其要旨，與長(zhǎng)篇大論的傳統(tǒng)文獻(xiàn)不同。利用這些資源，醫(yī)師可以節(jié)省閱讀大量文獻(xiàn)的時(shí)間。實(shí)證醫(yī)學(xué)＝臨床流行病學(xué)＋醫(yī)學(xué)資訊學(xué)＋團(tuán)隊(duì)合作本院EBM資料庫(kù)比較ACPJournalClubDARECDSRCCTR出版者美國(guó)內(nèi)科醫(yī)師學(xué)會(huì)出版NHSCRDCochraneCochrane資料來(lái)源每月至少過(guò)濾50種以上之核心期刊，例：NEJM、JAMA、Lancet、Circulation….等經(jīng)過(guò)評(píng)估、挑選有學(xué)術(shù)價(jià)值的醫(yī)學(xué)期刊、Medline、CINAL、Biosis、ERIC、PsycINFO、上百種期刊文獻(xiàn)

Medline、EMBASE、其他電子資料庫(kù)資料庫(kù)類(lèi)型全文型全文型全文型書(shū)目型實(shí)證醫(yī)學(xué)的參考書(shū)籍TheEvidencePyramidAnimalresearchInvitro(testtube)researchCaseseries/ReportsIdeas,Editorials,OpinionsCaseControlStudiesCohortstudiesRandomizedControlledStudiesRandomizedControlledDoubleBlindStudies

Meta-analysisGradeofRecommendationLevelofEvidenceTherapy[A]1aSystemicreviewofRCTs1bSingleRCT1c‘All-or-none’[B]2aSystemicreviewofcohortstudies2bCohortstudyorpoorRCT2c‘Outcomes’research3aSystemicreviewofcase-controlstudies3bCase-controlstudy[C]4Caseseries[D]5Expertopinion,physiology,benchresearch統(tǒng)計(jì)數(shù)字會(huì)說(shuō)話

評(píng)估時(shí)以具體的數(shù)字呈現(xiàn)結(jié)果敏感度（sensitivity）、特異度（specificity）、概似比（likelihoodratio）、檢測(cè)前機(jī)率（pre-testprobability）、檢測(cè)後機(jī)率（post-testprobability）ARR(Absoluteriskreduction)=EER(ExperimentalEventRate)-CER(ControlEventRate)、Numberneededtotreat,NNT=1/ARR

（增加一位病患得到某種處置好處所需的治療病人數(shù)）、相對(duì)危險(xiǎn)度減少百分比（relativeriskreduction，RRR）絕對(duì)危險(xiǎn)度增加百分比（absoluteriskincrease，ARI）=EER(ExperimentalEventRate)-CER(ControlEventRate)、

Numberneededtoharm,NNH=1/ARI（增加一位受試者罹患某種醫(yī)源性傷害的治療病人數(shù)）相對(duì)危險(xiǎn)(Relativerisk)、勝算(Odds)、勝算比(Oddsratio)、信賴(lài)區(qū)間(confidenceinterval)CalculationofOR/RRTreatmentEventPositiveNegativeExposed(experimental)A=1B=29Notexposed(control)C=9D=21EER=a/a+b=0.033CER=c/c+d=0.30Relativerisk=EER/CER=(a/a+b)/(c/c+d)=0.11ExperimentaleventOdds=a/b=0.034ControleventOdds=c/d=0.43RelativeOdds=Oddsratio=(a/b)/(c/d)=ad/bc=0.08Odd:aratioofnon-eventstoeventsAskingAnswerableClinicalQuestionPatient/ProblemInsulin-dependentdiabeticsInterventionIntensiveinsulinregimenComparisonRegularinsulinregimenOutcomesRetinopathySymptomatichypoglycemiaTreatmentofClassIII-IVLupusNephritisMeta-AnalysisTreatmentofdiffuseproliferativelupusnephritis:Ameta-analysisofrandomizedcontrolledtrialsAmJKidneyDis.2004Feb;43(2):197-208.

CochraneRenalGroupCochraneDatabaseofSystematic Reviews.1,2004.ValuesofRRlessthan1indicateareductioninriskfortheoutcomewiththeexperimentaltreatment.Conversely,valuesofRRmorethan1indicateanincreaseinrisk.The95%CIsareameasureofvariabilityintheprecisionoftheRRestimateanditsstatisticalsignificance.

Heterogeneityoftreatmenteffectsbetweenstudieswasinvestigatedbyvisualexaminationofgraphicmeta-analysisplotsandfromtheCochranQ(heterogeneitychi-square)andI2statistic.Thisisaforestplot,withaverticallineat1.0representingequivalenceinriskforanoutcomewithexperimentalandcontroltreatment(nullhypothesis).TheRRforeachoutcomeandits95%CIareindicatedbyasolidsquareandaline.Thesizeofthesolidsquarerepresentsthecontribution(weight)ofthetrialtotheanalysis.Diamond-shapedsymbolsrepresentthesummaryestimatorofoveralleffectpoolingtheweightedeffectofindividualRCTs.

實(shí)證醫(yī)學(xué)發(fā)展的重點(diǎn)實(shí)證醫(yī)學(xué)發(fā)展的重點(diǎn)，主要有四大範(fàn)疇：

增進(jìn)醫(yī)療決策技能medicaldecisionmakingtechnique改善醫(yī)療資訊擷取技能accessingmedicalinformation嚴(yán)格評(píng)估醫(yī)療資訊及其臨床應(yīng)用assessingthevalidityofmedicalinformation促進(jìn)專(zhuān)家間之合作，以作隨機(jī)對(duì)照試驗(yàn)(RCTrandomizedcontrolledtrials)的研究應(yīng)用最佳實(shí)證文獻(xiàn)，訂出臨床疾病診療的guideline，並促進(jìn)clinicalpathway形成。醫(yī)療資源有限，對(duì)醫(yī)療資源的合理運(yùn)用，必須利用現(xiàn)有的證據(jù)，及使用標(biāo)準(zhǔn)化的規(guī)範(fàn)及流程。大量的醫(yī)學(xué)知識(shí)文獻(xiàn)，使醫(yī)療提供者(healthcare-provider)無(wú)法消化這些知識(shí)。因此在實(shí)驗(yàn)室或臨床研究獲得的知識(shí)，與一般醫(yī)師真正日常行醫(yī)行為之應(yīng)用之間產(chǎn)生不一致的現(xiàn)象。透過(guò)各種疾病的臨床診療指引(clinicalpracticeguideline)，使醫(yī)療不一致性減低，以提高醫(yī)療品質(zhì)。ClinicalPracticeGuidelines醫(yī)院評(píng)鑑暨醫(yī)療品質(zhì)策進(jìn)會(huì)中央健保局實(shí)證醫(yī)學(xué)會(huì)

國(guó)家衛(wèi)生研究院ResourceCentersforGuidelinesNGC-NationalGuidelineClearinghouse

AGREE-appraisalofguidelineresearch&evaluation

GuidelinesInternationalNetwork

SIGN-ScottishIntercollegiateGuidelinesNetwork

臺(tái)灣實(shí)證醫(yī)學(xué)暨臨床診療指引平臺(tái)之建置國(guó)衛(wèi)院郭耿南教授Decisiontreeanalysis實(shí)證醫(yī)學(xué)應(yīng)用之障礙沒(méi)有實(shí)證、找不到實(shí)證實(shí)證之間有矛盾實(shí)證很難去應(yīng)用缺乏時(shí)間與資源缺乏保險(xiǎn)給付臨床診療指引制定費(fèi)時(shí)終身學(xué)習(xí)~醫(yī)學(xué)教育增加實(shí)證醫(yī)學(xué)訓(xùn)練發(fā)現(xiàn)臨床問(wèn)題，進(jìn)而帶動(dòng)研究風(fēng)氣將有限的資源運(yùn)用於具有實(shí)證的醫(yī)療臨床指引制定(學(xué)會(huì)、健保局、及醫(yī)院)

制定各種疾病的臨床診療指引及設(shè)立本土醫(yī)學(xué)資料庫(kù)，也