Meta-分析的统计学基础-王洪源.pptVIP

Meta－分析的统计学基础;一、概述;什么是循证医学

Evidence-basedmedicine(EBM)

循证医学是在对个体病人制定临床决策时应基于当前最佳的科学研究成果。

循证医学是最佳的证据、临床经验和病人价值的有机结合。;用X线进行乳腺癌筛查

Mammographyforbreastcancerisanestablishedscreeningmethod;Theauthorsfoundthatnotrialdatawereofhighquality

Twowereofmediumquality,andtherestwerepoorqualityorflawed.

Whentheresultsofthetwomediumqualitytrialswerecombined,theriskratiowas1.00(95%CI0.96,1.05)

Theyconcludedthat

“screeningforbreastcancerwith

mammographyisunjustified”

;;TheUSPreventiveServicesTaskForcereviewedthesamesetoftrials;Themeta-analysisperformedfortheUSPSTFonthemostcurrentpublisheddatafoundthatthepooledeffectsizeofthecombinedtrialswassizableandstatisticallysignificant:

thesummaryrelativerisk(RR)ofbreastcancerdeathamongwomenrandomizedtoscreeninginseventrialsthatincludedwomenolderthan50was0.77(95percentCI,0.67-0.89).;TheUSPSTFrecommendsscreeningmammography,withorwithoutclinicalbreastexamination,every1-2yearsforwomenaged40andolder.;被动吸烟的危害;Hackshawetal.

conductedaverycomprehensive

systematicreviewin1997;;;;;什么是循证医学;循证方法的两个关键方面;循证医学证据的分级;系统性综述;循证医学与传统医疗实践的四个重要区别(1992JAMA);Meta-analysis;;;Howtoreadasystematicreview?;Evaluationofqualityofprimarystudiessetssystematicreviewsapartfromtraditionalreviews

EmpiricresearchshowsthatnotallSRsassessstudyquality:

240SRsfromjournals:48%assessedquality(Moher1999)

480SRsinDARE:52%assessedquality(Petticrew1999)

50SRsonasthma:28%reportedvalidityassessmentcriteria(Jadad2000);Berkeley,2002onqualityofSRsonHIV,publishedduring2001

Qualityassessmentdonein56%ofreviews

Testingforheterogeneitydonein56%ofreviews

NotallSRswithsignificantheterogeneityexploredreasonsforit

Manyreviewsdidnotstatetherationaleforchoiceofmodelsusedforcombiningdata

Only12%ofreviewsevaluatedpubli