非苯二氮卓类药疗效荟萃分析中发表偏倚探讨.pdfVIP

BMJ2015;351:h4419doi:10.1136/bmj.h4419(Published14August2015)Page1of2

ENDGAMES

STATISTICALQUESTION

Whatispublicationbiasinameta-analysis?

PhilipSedgwickreaderinmedicalstatisticsandmedicaleducation

InstituteforMedicalandBiomedicalEducation,StGeorge’s,UniversityofLondon,London,UK

Researchersundertookameta-analysisoftheeffectivenessofsubjectivemeasurementsofsleeplatency.Thepurposeofthe

non-benzodiazepinehypnotics,includingeszopiclone,zaleplon,meta-analysiswastocombinethesampleestimatesofthe

andzolpidem.Randomisedcontrolledtrialswereincludediftreatmenteffectsfromeachtrialthatwasidentifiedandtogive

theyweredoubleblind,cebocontrolled,andhadparallelatotalestimateforeachoutcome,therebyreducingalarge

treatmentgroups.Participantswereadultswithprimaryinsomniaamountofinformationtoamanageablequantity.

(transientorchronic).ThemainoutcomeswereTheresearchersmaynothaveidentifiedalltherelevanttrials

polysomnographicandsubjectivemeasurementsofsleepthathadbeenconducted.Ifso,itmightmeanthatthetotal

latency.1

overallestimatesproducedbythemeta-analysiswerebiased

Thirteentrialswereeligibleforinclusion