系统评价报告规范PRISMA

1

PreferredReportingItemsforSystematicReviewsand

Meta-Analyses:ThePRISMAStatement

DavidMoher

1,2

*,AlessandroLiberati

3,4

,JenniferTetzlaff

1

,

5

,ThePRISMAGroup

"

1OttawaMethodsCentre,OttawaHospitalResearchInstitute,Ottawa,Ontario,Canada,2DepartmentofEpidemiologyandCommunityMedicine,FacultyofMedicine,

`

diModenaeReggioEmilia,Modena,Italy,4CentroCochraneItaliano,IstitutoRicercheFarmacologicheMarioUniversityofOttawa,Ottawa,Ontario,Canada,3Universita

Negri,Milan,Italy,5CentreforStatisticsinMedicine,UniversityofOxford,Oxford,UnitedKingdom

Introduction

Systematicreviewsandmeta-analyseshavebecomeincreasingly

iansreadthemtokeepuptodate

withtheirfield[1,2],andtheyareoftenusedasastartingpointfor

ngagenciesmay

requireasystematicreviewtoensurethereisjustificationfor

furtherresearch[3],andsomehealthcarejournalsaremovingin

thisdirection[4].Aswithallresearch,thevalueofasystematic

reviewdependsonwhatwasdone,whatwasfound,andtheclarity

otherpublications,thereportingqualityof

systematicreviewsvaries,limitingreaders’abilitytoassessthe

strengthsandweaknessesofthosereviews.

1987,Mulrowexamined50reviewarticlespublishedinfourleading

medicaljournalsin1985and1986andfoundthatnonemetalleight

explicitscientificcriteria,suchasaqualityassessmentofincluded

studies[5].In1987,Sacksandcolleagues[6]evaluatedtheadequacy

ofreportingof83meta-analyseson23characteristicsinsixdomains.

Reportingwasgenerallypoor;betweenoneand14characteristics

wereadequatelyreported(mean=7.7;standarddeviation=2.7).A

1996updateofthisstudyfoundlittleimprovement[7].

In1996,toaddressthesuboptimalreportingofmeta-analyses,

aninternationalgroupdevelopedaguidancecalledthe

QUOROMStatement(QUalityOfReportingOfMeta-analyses),

whichfocusedonthereportingofmeta-analysesofrandomized

controlledtrials[8].Inthisarticle,wesummarizearevisionof

theseguidelines,renamedPRISMA(PreferredReportingItems

forSystematicreviewsandMeta-Analyses),whichhavebeen

updatedtoaddressseveralconceptualandpracticaladvancesin

thescienceofsystematicreviews(Box1).

clinicians,medicaleditors,ectiveofthe

OttawameetingwastoreviseandexpandtheQUOROM

checklistandflowdiagram,asneeded.

Theexecutivecommitteecompletedthefollowingtasks,priorto

themeeting:asystematicreviewofstudiesexaminingthequality

ofreportingofsystematicreviews,andacomprehensiveliterature

searchtoidentifymethodologicalandotherarticlesthatmight

informthemeeting,especiallyinrelationtomodifyingchecklist

rnationalsurveyofreviewauthors,consumers,and

groupscommissioningorusingsystematicreviewsandmeta-

analyseswascompleted,includingtheInternationalNetworkof

AgenciesforHealthTechnologyAssessment(INAHTA)andthe

GuidelinesInternationalNetwork(GIN).Thesurveyaimedto

ascertainviewsofQUOROM,includingthemeritsoftheexisting

ultsoftheseactivitieswerepresented

duringthemeetingandaresummarizedonthePRISMAWebsite

(/).

Onlyitemsdeemedessentialwereretainedoraddedtothe

ditionalitemsareneverthelessdesirable,and

reviewauthorsshouldincludethese,ifrelevant[10].Forexample,

itisusefultoindicatewhetherthesystematicreviewisanupdate

[11]ofapreviousreview,andtodescribeanychangesin

proceduresfromthosedescribedintheoriginalprotocol.

Citation:MoherD,LiberatiA,TetzlaffJ,AltmanDG,ThePRISMA

Group(2009)PreferredReportingItemsforSystematicReviewsandMeta-

Analyses:d6(7)::10.1371/

.1000097

PublishedJuly21,2009

Copyright:ßanopen-accessarticledistributedunder

thetermsoftheCreativeCommonsAttributionLicense,whichpermits

unrestricteduse,distribution,andreproductioninanymedium,providedthe

originalauthorandsourcearecredited.

Funding:PRISMAwasfundedbytheCanadianInstitutesofHealthResearch;

`

diModenaeReggioEmilia,Italy;CancerResearchUK;ClinicalEvidenceUniversita

BMJKnowledge;theCochraneCollaboration;andGlaxoSmithKline,

funded,inpart,throughgrantsoftheItalianMinistryofUniversity(COFIN-PRIN

2002prot.2002061749andCOFIN-PRIN2006prot.2006062298).DGAisfunded

ndedbyaUniversityofOttawaResearchChair.

Noneofthesponsorshadanyinvolvementintheplanning,execution,orwrite-up

onally,nofunderplayedaroleindraftingthe

manuscript.

CompetingInterests:Theauthorshavedeclaredthatnocompetinginterests

exist.

Abbreviations:PRISMA,PreferredReportingItemsforSystematicreviewsand

Meta-Analyses;QUOROM,QUalityOfReportingOfMeta-analyses.

*E-mail:dmoher@

"MembershipofthePRISMAGroupisprovidedintheAcknowledgments.

Provenance:Notcommissioned;rto

encouragedisseminationofthePRISMAStatement,thisarticleisfreelyaccessible

onthePLoSMedicineWebsite(/)andwillbealso

publishedintheAnnalsofInternalMedicine,BMJ,JournalofClinicalEpidemiology,

detailsonfurtheruse,seethePRISMAWebsite(-statement.

org/).

Terminology

Theterminologyusedtodescribeasystematicreviewandmeta-

sonforchangingthename

fromQUOROMtoPRISMAwasthedesiretoencompassboth

adoptedthe

definitionsusedbytheCochraneCollaboration[9].Asystematic

reviewisareviewofaclearlyformulatedquestionthatuses

systematicandexplicitmethodstoidentify,select,andcritically

appraiserelevantresearch,andtocollectandanalyzedatafrom

ticalmethods

(meta-analysis)mayormaynotbeusedtoanalyzeandsummarize

-analysisreferstotheuseof

statisticaltechniquesinasystematicreviewtointegratetheresults

ofincludedstudies.

DevelopingthePRISMAStatement

Athree-daymeetingwasheldinOttawa,Canada,inJune2005

with29participants,includingreviewauthors,methodologists,

PLoSMedicine|1July2009|Volume6|Issue7|e1000097

Box1:ConceptualIssuesintheEvolutionfrom

QUOROMtoPRISMA

CompletingaSystematicReviewIsanIterative

ProcessTheconductofasystematicreviewdepends

heavilyonthescopeandqualityofincludedstudies:thus

systematicreviewersmayneedtomodifytheiroriginal

tematicreview

reportingguidelineshouldrecommendthatsuchchanges

canbereportedandexplainedwithoutsuggestingthat

SMAStatement(Items5,11,

16,and23)rom

Cochranereviews,allofwhichshouldhaveaprotocol,

onlyabout10%ofsystematicreviewersreportworking

fromaprotocol[22].Withoutaprotocolthatispublicly

accessible,itisdifficulttojudgebetweenappropriateand

inappropriatemodifications.

ConductandReportingResearchAreDistinct

ConceptsThisdistinctionis,however,less

straightforwardforsystematicreviewsthanfor

assessmentsofthereportingofanindividualstudy,

becausethereportingandconductofsystematicreviews

are,bynature,mple,thefailure

ofasystematicreviewtoreporttheassessmentoftherisk

ofbiasinincludedstudiesmaybeseenasamarkerofpoor

conduct,giventheimportanceofthisactivityinthe

systematicreviewprocess[37].

Study-LevelVersusOutcome-LevelAssessmentof

RiskofBiasForstudiesincludedinasystematicreview,a

thoroughassessmentoftheriskofbiasrequiresbotha

‘‘study-level’’,adequacyofallocation

concealment)and,forsomefeatures,anewerapproach

called‘‘outcome-level’’ome-level

assessmentinvolvesevaluatingthereliabilityandvalidity

ofthedataforeachimportantoutcomebydetermining

themethodsusedtoassessthemineachindividualstudy

[38].Thequalityofevidencemaydifferacrossoutcomes,

evenwithinastudy,suchasbetweenaprimaryefficacy

outcome,whichislikelytobeverycarefullyand

systematicallymeasured,andtheassessmentofserious

harms[39],whichmayrelyonspontaneousreportsby

formationshouldbereportedtoallow

anexplicitassessmentoftheextenttowhichanestimate

ofeffectiscorrect[38].

ImportanceofReportingBiasesDifferenttypesof

reportingbiasesmayhampertheconductand

ivereporting

,publicationbias)[28]aswellas

themorerecentlyempiricallydemonstrated‘‘outcome

reportingbias’’withinindividualstudies[40,41]shouldbe

consideredbyauthorswhenconductingasystematic

theimplicationsof

thesebiasesontheconductandreportingofsystematic

reviewsthemselvesareunclear,somepreviousresearch

hasidentifiedthatselectiveoutcomereportingmayoccur

alsointhecontextofsystematicreviews[42].

Althoughnodirectevidencewasfoundtosupportretainingor

addingsomeitems,evidencefromotherdomainswasbelievedto

mple,Item5asksauthorstoprovide

registrationinformationaboutthesystematicreview,includinga

registrationnumber,ghsystematicreview

registrationisnotyetwidelyavailable[12,13],theparticipating

journalsoftheInternationalCommitteeofMedicalJournal

Editors(ICMJE)[14]nowrequireallclinicaltrialstoberegistered

inanefforttoincreasetransparencyandaccountability[15].

Thoseaspectsarealsolikelytobenefitsystematicreviewers,

possiblyreducingtheriskofanexcessivenumberofreviews

addressingthesamequestion[16,17]andprovidinggreater

transparencywhenupdatingsystematicreviews.

ThePRISMAStatement

ThePRISMAStatementconsistsofa27-itemchecklist(Table1;

seealsoTextS1foradownloadableWordtemplateforresearchers

tore-use)andafour-phaseflowdiagram(Figure1;seealsoFigure

S1foradownloadableWordtemplateforresearcherstore-use).

TheaimofthePRISMAStatementistohelpauthorsimprovethe

focused

onrandomizedtrials,butPRISMAcanalsobeusedasabasisfor

reportingsystematicreviewsofothertypesofresearch,particularly

mayalsobeusefulforcritical

r,thePRISMA

checklistisnotaqualityassessmentinstrumenttogaugethequality

ofasystematicreview.

FromQUOROMtoPRISMA

ThenewPRISMAchecklistdiffersinseveralrespectsfromthe

QUOROMchecklist,andthesubstantivespecificchangesare

lly,thePRISMAchecklist

‘‘decouples’’severalitemspresentintheQUOROMchecklist

and,whereapplicable,severalchecklistitemsarelinkedto

improveconsistencyacrossthesystematicreviewreport.

including

studiesandprovidingreasonsforexcludingothers,thereview

archresultsinrecords.

Oncetheserecordshavebeenscreenedandeligibilitycriteria

applied,berof

includedarticlesmightbesmaller(orlarger)thanthenumberof

studies,becausearticlesmayreportonmultiplestudiesandresults

capturethisinformation,thePRISMAflowdiagramnowrequests

informationonthesephasesofthereviewprocess.

Endorsement

ThePRISMAStatementshouldreplacetheQUOROMState-

thatotherjournalswillsupportPRISMA;theycandosobyregistering

rscoretoauthors,andothers,the

importanceoftransparentreportingofsystematicreviews,we

encouragesupportingjournalstoreferencethePRISMAStatement

andincludethePRISMAWebaddressintheirInstructionsto

inviteeditorialorganizationstoconsiderendorsing

PRISMAandencourageauthorstoadheretoitsprinciples.

ShortlyafterthemeetingadraftofthePRISMAchecklistwas

circulatedtothegroup,includingthoseinvitedtothemeetingbut

sitionfilewascreatedcontaining

commentsandrevisionsfromeachrespondent,andthechecklist

upapprovedthe

checklist,flowdiagram,andthissummarypaper.

PLoSMedicine|2

ThePRISMAExplanationandElaborationPaper

InadditiontothePRISMAStatement,asupportingExplana-

tionandElaborationdocumenthasbeenproduced[18]following

thestyleusedforotherreportingguidelines[19–21].Theprocess

July2009|Volume6|Issue7|e1000097

informationthroughthedifferentphasesofasystematicreview.

doi:10.1371/.1000097.g001

ofcompletingthisdocumentincludeddevelopingalargedatabase

ofexemplarstohighlighthowbesttoreporteachchecklistitem,

andidentifyingacomprehensiveevidencebasetosupportthe

lanationandElaboration

documentwascompletedafterseveralfacetofacemeetingsand

numerousiterationsamongseveralmeetingparticipants,after

whichitwassharedwiththewholegroupforadditionalrevisions

y,thegroupformedadissemination

subcommitteetohelpdisseminateandimplementPRISMA.

Discussion

Thequalityofreportingofsystematicreviewsisstillnot

optimal[22–27].Inarecentreviewof300systematicreviews,

fewauthorsreportedassessingpossiblepublicationbias[22],

eventhoughthereisoverwhelmingevidencebothforits

existence[28]anditsimpactontheresultsofsystematic

reviews[29].Evenwhenthepossibilityofpublicationbiasis

assessed,thereisnoguaranteethatsystematicreviewershave

assessedorinterpreteditappropriately[30].Althoughthe

absenceofreportingsuchanassessmentdoesnotnecessarily

indicatethatitwasnotdone,reportinganassessmentofpossible

publicationbiasislikelytobeamarkerofthethoroughnessof

theconductofthesystematicreview.

Severalapproacheshavebeendevelopedtoconductsystematic

mple,systematic

PLoSMedicine|3

reviewsarenowconductedtoinvestigatecost-effectiveness[31],

diagnostic[32]orprognosticquestions[33],geneticassociations

[34],andpolicymaking[35].Thegeneralconceptsandtopics

coveredbyPRISMAareallrelevanttoanysystematicreview,not

justthosewhoseobjectiveistosummarizethebenefitsandharms

r,somemodificationsofthe

checklistitemsorflowdiagramwillbenecessaryinparticular

mple,assessingtheriskofbiasisakey

concept,buttheitemsusedtoassessthisinadiagnosticrevieware

likelytofocusonissuessuchasthespectrumofpatientsandthe

verificationofdiseasestatus,whichdifferfromreviewsof

wdiagramwillalsoneedadjustmentswhen

reportingindividualpatientdatameta-analysis[36].

Wehavedevelopedanexplanatorydocument[18]toincrease

hchecklistitem,thisdocument

containsanexampleofgoodreporting,arationaleforitsinclusion,

andsupportingevidence,includingreferences,wheneverpossible.

Webelievethisdocumentwillalsoserveasausefulresourcefor

urage

journalstoincludereferencetotheexplanatorydocumentintheir

InstructionstoAuthors.

Likeanyevidence-basedendeavor,PRISMAisaliving

endweinvitereaderstocommentonthe

revisedversion,particularlythenewchecklistandflowdiagram,

usesuchinformationto

informPRISMA’scontinueddevelopment.

July2009|Volume6|Issue7|e1000097

istofitemstoincludewhenreportingasystematicreviewormeta-analysis.

Section/Topic

TITLE

Title

ABSTRACT

Structuredsummary

#ChecklistItem

Reportedon

Page#

1Identifythereportasasystematicreview,meta-analysis,orboth.

2Provideastructuredsummaryincluding,asapplicable:background;objectives;datasources;studyeligibility

criteria,participants,andinterventions;studyappraisalandsynthesismethods;results;limitations;conclusions

andimplicationsofkeyfindings;systematicreviewregistrationnumber.

INTRODUCTION

Rationale

Objectives

METHODS

Protocolandregistration

Eligibilitycriteria

Informationsources

Search

Studyselection

Datacollectionprocess

Dataitems

Riskofbiasinindividual

studies

Summarymeasures

Synthesisofresults

Riskofbiasacrossstudies

Additionalanalyses

RESULTS

Studyselection

Studycharacteristics

Riskofbiaswithinstudies

Resultsofindividualstudies

Synthesisofresults

Riskofbiasacrossstudies

Additionalanalysis

DISCUSSION

Summaryofevidence

Limitations

Conclusions

FUNDING

Funding,supplyofdata);roleoffundersfor

thesystematicreview.

24

25

26

Summarizethemainfindingsincludingthestrengthofevidenceforeachmainoutcome;considertheir

,healthcareproviders,users,andpolicymakers).

,riskofbias),,incompleteretrievalof

identifiedresearch,reportingbias).

Provideageneralinterpretationoftheresultsinthecontextofotherevidence,andimplicationsforfuture

research.

17

18

19

20

21

22

23

Givenumbersofstudiesscreened,assessedforeligibility,andincludedinthereview,withreasonsforexclusions

ateachstage,ideallywithaflowdiagram.

Foreachstudy,,studysize,PICOS,follow-upperiod)

andprovidethecitations.

Presentdataonriskofbiasofeachstudyand,ifavailable,anyoutcome-levelassessment(seeItem12).

Foralloutcomesconsidered(benefitsorharms),present,foreachstudy:(a)simplesummarydataforeach

interventiongroupand(b)effectestimatesandconfidenceintervals,ideallywithaforestplot.

Presentresultsofeachmeta-analysisdone,includingconfidenceintervalsandmeasuresofconsistency.

Presentresultsofanyassessmentofriskofbiasacrossstudies(seeItem15).

Giveresultsofadditionalanalyses,,sensitivityorsubgroupanalyses,meta-regression[seeItem16]).

5

6

7

8

9

10

11

12

13

14

15

16

Indicateifareviewprotocolexists,,Webaddress),and,ifavailable,provide

registrationinformationincludingregistrationnumber.

,PICOS,lengthoffollow-up),yearsconsidered,

language,publicationstatus)usedascriteriaforeligibility,givingrationale.

,databaseswithdatesofcoverage,contactwithstudyauthorstoidentify

additionalstudies)inthesearchanddatelastsearched.

Presentfullelectronicsearchstrategyforatleastonedatabase,includinganylimitsused,suchthatitcouldbe

repeated.

,screening,eligibility,includedinsystematicreview,and,ifapplicable,

includedinthemeta-analysis).

,pilotedforms,independently,induplicate)andany

processesforobtainingandconfirmingdatafrominvestigators.

,PICOS,fundingsources)andanyassumptionsand

simplificationsmade.

Describemethodsusedforassessingriskofbiasofindividualstudies(includingspecificationofwhetherthiswas

doneatthestudyoroutcomelevel),andhowthisinformationistobeusedinanydatasynthesis.

,riskratio,differenceinmeans).

Describethemethodsofhandlingdataandcombiningresultsofstudies,ifdone,includingmeasuresof

,I

2

)foreachmeta-analysis.

Spe,publicationbias,selective

reportingwithinstudies).

,sensitivityorsubgroupanalyses,meta-regression),ifdone,

indicatingwhichwerepre-specified.

3

4

Describetherationaleforthereviewinthecontextofwhatisalreadyknown.

Provideanexplicitstatementofquestionsbeingaddressedwithreferencetoparticipants,interventions,

comparisons,outcomes,andstudydesign(PICOS).

doi:10.1371/.1000097.t001

PLoSMedicine|4July2009|Volume6|Issue7|e1000097

ntivespecificchangesbetweentheQUOROMchecklistandthePRISMAchecklist(atickindicatesthepresenceof

thetopicinQUOROMorPRISMA).

Section/Topic

Abstract

Introduction

ItemQUOROM

!

PRISMA

!

!

Comment

r,PRISMAisnot

specificaboutformat.

Thisnewitem(4)addressestheexplicitquestionthereviewaddressesusingthePICO

reportingsystem(whichdescribestheparticipants,interventions,comparisons,and

outcome(s)ofthesystematicreview),togetherwiththespecificationofthetypeof

studydesign(PICOS);theitemislinkedtoItems6,11,and18ofthechecklist.

Thisnewitem(5)asksauthorstoreportwhetherthereviewhasaprotocolandifso

howitcanbeaccessed.

AlthoughreportingthesearchispresentinbothQUOROMandPRISMAchecklists,

PRISMAasksauthorstoprovideafulldescriptionofatleastoneelectronicsearch

strategy(Item8).Withoutsuchinformationitisimpossibletorepeattheauthors’

search.

Renamedfrom‘‘qualityassessment’’em(12)islinkedwith

reportingthisinformationintheresults(Item19).Thenewconceptof‘‘outcome-

level’’assessmenthasbeenintroduced.

Thisnewitem(15)asksauthorstodescribeanyassessmentsofriskofbiasinthe

review,emislinked

withreportingthisinformationintheresults(Item22).

AlthoughbothQUOROMandPRISMAchecklistsaddressthediscussionsection,

PRISMAdevotesthreeitems(24–26)MAthemaintypesof

limitationsareexplicitlystatedandtheirdiscussionrequired.

Thisnewitem(27)asksauthorstoprovideinformationonanysourcesoffundingfor

thesystematicreview.

Objective

Methods

Methods

Protocol

Search!

!

!

MethodsAssessmentof

riskofbiasin

includedstudies

Assessmentof

riskofbiasacross

studies

!!

Methods!

Discussion!!

Funding!

doi:10.1371/.1000097.t002

SupportingInformation

Flowofinformationthroughthedifferent

phasesofasystematicreview(downloadabletemplate

documentforresearcherstore-use).

Foundat:doi:10.1371/.1000097.s001(0.08MB

DOC)

FigureS1

Checklistofitemstoincludewhenreportinga

systematicreviewormeta-analysis(downloadabletem-

platedocumentforresearcherstore-use).

Foundat:doi:10.1371/.1000097.s002(0.04MB

DOC)

TextS1

Acknowledgments

ThefollowingpeoplecontributedtothePRISMAStatement:Doug

Altman,DSc,CentreforStatisticsinMedicine(Oxford,UK);Gerd

Antes,PhD,UniversityHospitalFreiburg(Freiburg,Germany);David

Atkins,MD,MPH,HealthServicesResearchandDevelopmentService,

VeteransHealthAdministration(Washington,D.C.,US);Virginia

Barbour,MRCP,DPhil,PLoSMedicine(Cambridge,UK);NickBarrow-

man,PhD,Children’sHospitalofEasternOntario(Ottawa,Canada);

,ScD,Johnson&JohnsonPharmaceuticalResearchand

Development(Titusville,NewJersey,US);JocalynClark,PhD,PLoS

Medicine(atthetimeofwriting,BMJ,London,UK);MikeClarke,PhD,

UKCochraneCentre(Oxford,UK)andSchoolofNursingand

Midwifery,TrinityCollege(Dublin,Ireland);DeborahCook,MD,

DepartmentsofMedicine,ClinicalEpidemiologyandBiostatistics,

McMasterUniversity(Hamilton,Canada);RobertoD’Amico,PhD,

Universita`diModenaeReggioEmilia(Modena,Italy)andCentro

CochraneItaliano,IstitutoRicercheFarmacologicheMarioNegri

(Milan,Italy);,PhD,UniversityofBirmingham

(Birmingham,UK);aux,MD,PhD,Departmentsof

Medicine,ClinicalEpidemiologyandBiostatistics,McMasterUniversity

(Hamilton,Canada);KayDickersin,PhD,JohnsHopkinsBloomberg

SchoolofPublicHealth(Baltimore,Maryland,US);MatthiasEgger,

MD,DepartmentofSocialandPreventiveMedicine,UniversityofBern

(Bern,Switzerland);EdzardErnst,MD,PhD,FRCP,FRCP(Edin),

PeninsulaMedicalSchool(Exeter,UK);PeterC.Gøtzsche,MD,MSc,

TheNordicCochraneCentre(Copenhagen,Denmark);JeremyGrim-

shaw,MBChB,PhD,FRCFP,OttawaHospitalResearchInstitute

(Ottawa,Canada);GordonGuyatt,MD,DepartmentsofMedicine,

ClinicalEpidemiologyandBiostatistics,McMasterUniversity(Hamilton,

Canada);JulianHiggins,PhD,MRCBiostatisticsUnit(Cambridge,UK);

dis,MD,UniversityofIoanninaCampus(Ioannina,

Greece);JosKleijnen,MD,PhD,KleijnenSystematicReviewsLtd

(York,UK)andSchoolforPublicHealthandPrimaryCare(CAPHRI),

UniversityofMaastricht(Maastricht,Netherlands);TomLang,MA,

TomLangCommunicationsandTraining(Davis,California,US);

AlessandroLiberati,MD,Universita`diModenaeReggioEmilia

(Modena,Italy)andCentroCochraneItaliano,IstitutoRicerche

FarmacologicheMarioNegri(Milan,Italy);NicolaMagrini,MD,NHS

CentrefortheEvaluationoftheEffectivenessofHealthCare–CeVEAS

(Modena,Italy);DavidMcNamee,PhD,TheLancet(London,UK);

LorenzoMoja,MD,MSc,CentroCochraneItaliano,IstitutoRicerche

FarmacologicheMarioNegri(Milan,Italy);DavidMoher,PhD,Ottawa

MethodsCentre,OttawaHospitalResearchInstitute(Ottawa,Canada);

CynthiaMulrow,MD,MSc,AnnalsofInternalMedicine(Philadelphia,

Pennsylvania,US);MaryannNapoli,CenterforMedicalConsumers

(NewYork,NewYork,US);AndyOxman,MD,NorwegianHealth

ServicesResearchCentre(Oslo,Norway);Ba’Pham,MMath,Toronto

HealthEconomicsandTechnologyAssessmentCollaborative(Toronto,

Canada)(atthetimeofthefirstmeetingofthegroup,GlaxoSmithKline

Canada,Mississauga,Canada);DrummondRennie,MD,FRCP,FACP,

UniversityofCaliforniaSanFrancisco(SanFrancisco,California,US);

MargaretSampson,MLIS,Children’sHospitalofEasternOntario

(Ottawa,Canada);,PhD,MBA,FamilyHealth

International(Durham,NorthCarolina,US);le,MD,

PhD,SouthernCaliforniaEvidenceBasedPracticeCenter(Santa

Monica,California,US);JenniferTetzlaff,BSc,OttawaMethods

Centre,OttawaHospitalResearchInstitute(Ottawa,Canada);David

Tovey,FRCGP,TheCochraneLibrary,CochraneCollaboration

PLoSMedicine|5July2009|Volume6|Issue7|e1000097

(Oxford,UK)(atthetimeofthefirstmeetingofthegroup,BMJ,

London,UK);PeterTugwell,MD,MSc,FRCPC,Instituteof

PopulationHealth,UniversityofOttawa(Ottawa,Canada).

AuthorContributions

ICMJEcriteriaforauthorshipreadandmet:he

firstdraftofthepaper:butedtothewritingofthe

paper:ipatedinregularconferencecalls,identified

theparticipants,securedfunds,plannedthemeeting,participatedinthe

meeting,anddraftedthemanuscript:ipatedin

identifyingtheevidencebaseforPRISMA,refiningthechecklist,and

draftingthemanuscript:iththerecommendations:DMALJT

DGA.

References

D,CookDJ,GuyattGH(1994)Users’guidestothemedicalliterature.

272:1367–1371.

erGH,VolminkJ,IoannidisJP(2003)Numberofpublishedsystematic

reviewsandglobalburdenofdisease:327:1083–1084.

anInstitutesofHealthResearch(2006)Randomizedcontrolledtrials

registration/applicationchecklist(12/2006).Available:.

ca/e/documents/rct_reg_ed19May2009.

,HortonR(2005)366:107.

CD(1987)Themedicalreviewarticle:ern

Med106:485–488.

S,BerrierJ,ReitmanD,Ancona-BerkVA,ChalmersTC(1987)Meta-

lJMed316:450–455.

S,ReitmanD,PaganoD,KupelnickB(1996)Meta-analysis:Anupdate.

MtSinaiJMed63:216–224.

,CookDJ,EastwoodS,OlkinI,RennieD,etal.(1994)Improvingthe

qualityofreportingofmeta-analysisofrandomizedcontrolledtrials:The

354:1896–1900.

,HigginsJ,eds(2005)nehandbookforsystematic

ble:

/resources/ed19May2009.

D,TilburtJ(2008)Valuejudgmentsintheanalysisandsynthesisof

pidemiol61:521–524.

,TsertsvadzeA(2006)Systematicreviews:Whenisanupdatean

update?Lancet367:881–883.

sityofYork(2009)ble:

/inst/crd/.Accessed19May2009.

nnaBriggsInstitute(2008)Protocols&ble:

/pubs/systematic_reviews_ed

19May2009.

lisC,DrazanJM,FrizelleFA,HaugC,HoeyJ,etal.(2004)Clinical

trialregistration:AstatementfromtheInternationalCommitteeofMedical

171:606–607.

ngtonCJ,KendallT,FonagyP,CottrellD,CotgroveA,etal.(2004)

Selectiveserotoninreuptakeinhibitorsinchildhooddepression:Systematic

363:1341–1345.

wSM,McAlisterFA,MannsBJ,GhaliWA(2006)Acetylcysteineinthe

preventionofcontrast-inducednephropathy:Acasestudyofthepitfallsinthe

ternMed166:161–166.

-ZoccaiGG,LotrionteM,AbbateA,TestaL,RemigiE,etal.(2006)

CompliancewithQUOROMandqualityofreportingofoverlappingmeta-

analysesontheroleofacetylcysteineinthepreventionofcontrastassociated

nephropathy:332:202–209.

tiA,AltmanDG,TetzlaffJ,MulrowC,GøtzscheP,etal.(2009)The

PRISMAstatementforreportingsystematicreviewsandmeta-analysesofstudies

thatevaluatehealthcareinterventions:d

6::10.1371/.1000100.

DG,SchulzKR,MoherD,EggerM,DavidoffF,etal.(2001)The

revisedCONSORTstatementforreportingrandomizedtrials:Explanationand

ernMed134:663–694.

tPM,ReitsmaJB,BrunsDE,GatsonisCA,GlasziouPP,etal.(2003)

Towardscompleteandaccuratereportingofstudiesofdiagnosticaccuracy:The

ernMed138:W1–W12.

brouckeJP,vonElmE,AltmanDG,GøtzschePC,MulrowCD,etal.

(2007)StrengtheningtheReportingofObservationalStudiesinEpidemiology

(STROBE):ernMed147:W163–W194.

,TetzlaffJ,TriccoAC,SampsonM,AltmanDG(2007)Epidemiology

d4:e78.

doi:10.1371/.0040078.

riM,MorrowF,KulkarniAV,TornettaP(2001)Meta-analysesin

orthopaedicsurgery:oint

SurgAm83-A:15–24.

D,TraversA,DorganM,SlaterL,RoweBH(2001)Evaluatingthe

rg

Med38:518–526.

dsD(2004)sed

Dent5:17.

,HalpernSH,MalikN,JadadAR,TramerMR,etal.(2001)

Examiningtheevidenceinanesthesialiterature:Acriticalappraisalofsystematic

Analg92:700–709.

yA,BagshawSM,FerlandA,MannsB,LauplandKB(2005)A

systematicevaluationofthequalityofmeta-analysesinthecriticalcare

re9:R575–R582.

sinK(2005)Publicationbias:Recognizingtheproblem,understandingits

originsandscope,:RothsteinHR,SuttonAJ,

BorensteinM,ationbiasinmeta-analysis-Prevention,assessmentand

ster(UK):JohnWiley&11–33.

AJ(2005)Evidenceconcerningtheconsequencesofpublicationand

:RothsteinHR,SuttonAJ,BorensteinM,ationbias

inmeta-analysis-Prevention,ster(UK):John

Wiley&175–192.

,IoannidisJP,TerrinN,SchmidCH,OlkinI(2006)Thecaseofthe

333:597–600.

umU,ChopraCL,HuangG,ScheimanJM,ChernewME,etal.(2001)

Aspirinasanadjuncttoscreeningforpreventionofsporadiccolorectalcancer:A

ernMed135:769–781.

J(2001)Systematicreviewsinhealthcare:Systematicreviewsof

323:157–162.

DG(2001)Systematicreviewsofevaluationsofprognosticvariables.

BMJ323:224–228.

disJP,NtzaniEE,TrikalinosTA,Contopoulos-IoannidisDG(2001)

et29:306–309.

,DaviesH,OxmanA,DenisJ,Golden-BiddleK,etal.(2005)Towards

systematicreviewsthatinformhealthcaremanagementandpolicy-making.

JHealthServResPolicy10:35–48.

tLA,ClarkeMJ(1995)Practicalmethodologyofmeta-analyses

(overviews)neWorkingGroup.

StatMed14:2057–2079.

,TelaroE,D’AmicoR,MoschettiI,CoeL,etal.(2005)Assessmentof

methodologicalqualityofprimarystudiesbysystematicreviews:Resultsofthe

330:1053–1055.

GH,OxmanAD,VistGE,KunzR,Falck-YtterY,etal.(2008)

GRADE:Anemergingconsensusonratingqualityofevidenceandstrengthof

336:924–926.

mannHJ,JaeschkeR,CookDJ,BriaWF,El-SolhAA,etal.(2006)An

officialATSstatement:Gradingthequalityofevidenceandstrengthof

pirCrit

CareMed174:605–614.

,HrobjartssonA,HaahrMT,GøtzschePC,AltmanDG(2004)

Empiricalevidenceforselectivereportingofoutcomesinrandomizedtrials:

291:2457–2465.

,Krleza-JericK,SchmidI,AltmanDG(2004)Outcomereporting

biasinrandomizedtrialsfundedbytheCanadianInstitutesofHealthResearch.

CMAJ171:735–740.

CA,MiddletonP,HopewellS(2002)Publishingprotocolsofsystematic

reviews:287:

2831–2834.

PLoSMedicine|6July2009|Volume6|Issue7|e1000097