Meta分析的统计学基础专家讲座.pptx

Meta－分析统计学基础;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订临床决议时应基于当前最正确科学研究结果。

循证医学是最正确证据、临床经验和病人价值有机结合。;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.;被动吸烟危害;Hackshawetal.

conductedaverycomprehensive

systematicreviewin1997;;;;;什么是循证医学;循证方法两个关键方面;循证医学证据分级;系统性综述;循证医学与传统医疗实践四个主要区分(1992JAMA);Meta-analysis;Systematicreviews/meta-analysesindexedinPubMed–10years;Howtoreadasystematicreview?;Evaluationofqualityofprimarystudiessetssystematicreviewsapartfromtraditionalreviews

EmpiricresearchshowsthatnotallSRsassessstudyquality:

240SRsfromjournals:48%assessedquality(Moher1999)

480SRsinDARE:52%assessedquality(Petticrew1999)

50SRsonasthma:28%reportedvalidityassessmentcriteria(Jadad);Berkeley,onqualityofSRsonHIV,publishedduring

Qualityassessmentdonein56%ofreviews

Testingforheterogeneitydonein56%ofreviews

NotallSRswithsignificantheterogeneityexploredreasonsforit

Manyreviewsdidnotstatetherationaleforchoiceofmodelsusedforcombinin