流行病學(xué)課件-2014級臨床from11ebm

健 提案2健 2015/10/212案Ajuniormedicalresidentworkinginateaching43-year-oldpreviouslywellman,awitnessedgrandmal(癲癇大發(fā)作hadneverhadaseizurebefore,hadnothadanyrecentheadtrauma,hadnothadalcoholonthedayoftheseizure,physicalexaminationarenormal,acomputedtomographicheadscaniscompleynormal,andanelectroencephalogramshowsonlynonspecificfindingsThepatientisgivenaloadingdoseofphenytoin(苯妥英鈉intravenouslyandthedrugisThepatientisveryconcernedabouthisriskofrecurrence(癲癇再次發(fā)作風(fēng)險Howmightthejuniorresident3Source:JAMA.1992;268(17):2420-3健 2015/10/21SolutionThejuniorresidentaskedherseniorresidentandwastoldthattheriskofseizurerecurrenceishigh(thoughhecouldnotputanexactnumberonit)Thejuniorresidentconveyedthisinformationtothepatient,emphasizingtothepatientnottodrive,tocontinuehismedication,andtoseehisfamilyphysicianinfollow-upThepatientleavesinastateofvaguetrepidation恐懼abouthisriskofsubsequent4健 2015/10/214SolutionShegoesthelibrary,searchesliture,25relevantScanningthetitles,oneTheresultsareapplicabletoherRiskofrecurrenceat1yearis43-51%,andat3years51-Afteraseizure-periodof18months,hisriskofrecurrenceSheconveysthisinformationtopatient, takinghismedication,seeingdoctorReassessingtheneedformedicationifseizure-for18Thepatientleaveswithaclearideaofhislikely5健 2015/10/215GeneralmedicalSpecialtyJournal6健 2015/10/2167健 2015/10/217循證8Evidence-based8健 2015/10/21循證pathophysiologicrationaleassufficientgroundsforEBMstressestheexaminationofevidencefromclinicalEBMrequiresnewskillsofthephysician,includingefficientli searchingandevidenceevaluating9健 2015/10/219循證Conscientious:認真的；盡責(zé)的；本著良心Explicit:明確的；清楚的；直率Judicious:明智的；頭腦精明的；判斷正確健 2015/10/21 循證ElliottM.Antman;JosephLau;BruceKupelnick;FrederickMosler;ThomasC.AAComparisonofResultsofJAMA.1992;268(2):240-評價心梗治療RCT積累情況與臨床專家形成推薦意見的“間隔對對于臨床醫(yī)生而言，找到并分析治療試驗相關(guān)的研是非的健 2015/10/21循證couldbestudiedServices”in1972promotingEBM健 2015/10/21循證健 2015/10/21David

循證McMasterUniversity 健 2015/10/21 循證醫(yī)學(xué)作用健 2015/10/21循證EstablishingtheCochraneCollaborationtocollateandsummarizeevidencefromclinicaltrials健 2015/10/21循證Settingmethodologicalandpublicationstandardsforprimaryandsecondaryresearch健 2015/10/21 循證Buildingnationalandinternationalinfrastructuresfordeveloandupdatingclinicalpracticeguidelines健 2015/10/21循證Theevidencebased“qualitymark”hasmisappropriatedbyvested健 2015/10/21循證Toomuch健 2015/10/21循證Statisticallysignificantbenefitsmaybemarginalinclinicalpractice健 2015/10/21循證Poorfitfor健 2015/10/21最最健 2015/10/21100項隨訪研究10年生存能否通過100項研究得出40%結(jié)論呢能否通過100項研究得出40%結(jié)論呢系統(tǒng)綜述（SystematicAsystematicreviewattemptstocollateallempiricalevidencethatfitspre-specifiedeligibilitycriteriainordertoansweraspecificresearchquestionItusesexplicit,systematicmethodsthatareselectedwithaviewtominimizingbias,thusprovidingmorereliablefindingsfromwhichconclusionscanbedrawnanddecisionsmadeSource:Cochrane健 2015/10/21系統(tǒng)綜述（SystematicKeyAclearlystatedsetofpre-definedeligibilitycriteriaforAnexplicit,reproducibleAsystematicsearchthatattemptstoidentifyAnassessmentofthevalidityoftheAsystematicpresentation,andsynthesis,ofcharacteristicsandfindingsoftheincluded健 2015/10/21Meta-ysisistheuseofstatisticalmethodstosummarizetheresultsofindependentstudiesMeta-ysescanprovidemorepreciseestimatesoftheeffectsofhealthcarethanthosederivedfromtheindividualstudiesSource:Cochrane健 2015/10/21系統(tǒng)綜述和meta系統(tǒng)綜述不一定包含Meta健 2015/10/21系統(tǒng)綜述和metaCochrane系統(tǒng)綜述和Meta分析實健 2015/10/21Cochrane系統(tǒng)綜述與其他系統(tǒng)綜述比Cochrane系統(tǒng)綜只收錄RCT綜述個別綜述參考價值有限其他系統(tǒng)綜Lancet,JAMA,BMJ均接收RCT系統(tǒng)綜絕大多數(shù)專業(yè)領(lǐng)域接收RCT系統(tǒng)綜絕大多數(shù)按照“原始論著”出，亦有按“綜述”健 2015/10/21RCT系統(tǒng)綜述與非RCT系統(tǒng)綜述比RCT偏倚少，更受有時RCT難開展，甚至不能開展，以致循證醫(yī)學(xué)缺并非所有醫(yī)學(xué)均需要RCT：如“疾病預(yù)后情況”、“診斷試驗非RCT有時是唯一來除Cochrane館外，絕大多數(shù)接收非RCT綜常見各類原始研究，均可以通過系統(tǒng)綜述予以總健 2015/10/21非RCT系統(tǒng)綜述實健 2015/10/21 系統(tǒng)綜述和meta提出科 制問 計

選文定期更結(jié)定期更結(jié)評處健 2015/10/21提出明確開展系統(tǒng)綜述和Meta分析必要性，常見原因匯總研究Reducealargeamountofinformationtoa回答“不確定性已的研究樣本量通常較小，結(jié)果不穩(wěn)已的結(jié)果差異較大，匯總有助于尋找原全面了解某領(lǐng)域研究進展，為深入研究打基健 2015/10/21制定應(yīng)包含Meta分析的全部環(huán)節(jié)，盡可能計劃有助于理順按計劃執(zhí)行，可防止“健 2015/10/21非Cochrane健 2015/10/21確定檢

檢索、篩選文：過多/過數(shù)據(jù)庫 /全文、熟和，可參考既往類似系統(tǒng)綜和，語言、雙人檢逆向驗證，反復(fù)修最終確定之后，再行檢規(guī)范/優(yōu)化篩選流文獻軟件雙人同步健 2015/10/21評價定量評選擇成熟量TheNewcastle-OttawaScale(NOS):non-randomizedJadadScale:randomized定性評偏倚：選擇、信息、混雙人同評：不一致時第三人確定(Senior)或協(xié)商確健 2015/10/21提取基本信描述性常用Excel提取（示例統(tǒng)計學(xué)分析所需擬合并的分層匯總所需的分層按照擬用分析軟件，確定數(shù)據(jù)整理形式（R軟件示例健 2015/10/21統(tǒng)計統(tǒng)計學(xué)分 目的是為了獲得合并效應(yīng)合并前提研究設(shè)計要相對同質(zhì)：納入排除標準擬合并的效應(yīng)指標相對同質(zhì)：同質(zhì)性檢驗（I2統(tǒng)計量），卡方檢驗P健 2015/10/21 統(tǒng)計同不同固定效應(yīng)模型計算合并效隨機效應(yīng)模型計算合并效亦可用隨機效應(yīng)模型計算合并效應(yīng)分層（亞組）分Meta回歸放棄Meta健 2015/10/21統(tǒng)計效應(yīng)值合并（）基礎(chǔ)軟件可直接輸出（Revman,Stata,R等健 2015/10/21敏感性

統(tǒng)計相對主分析相對穩(wěn)健性（Robustness）而調(diào)整主分析策略，對比調(diào)整后分析結(jié)果與原結(jié)果的差異，進而探原結(jié)果的穩(wěn)健性（或敏感性），間接評價了“調(diào)整”策略的影健 2015/10/21報告偏倚（ReportingBiasesthatmighthaveinfluencedidentificationtrialsarecollectivelyknownasreportingOftenmisusedaspublicationFourmajor 熱點問題快，非熱點問題作者健 2015/10/21報告觀察漏斗圖（FunnelIntheabsenceofreportingbias,largeststudieswillbeplottedneartheaverage,smallerstudi