CONSORT 聲明修訂了提高平行組 RCT 報告質量的建議

BMCMedicalResearchMethodology(2001)1:2 /1471-2288/1/2

BMC1:2MedicalstatementResearchMethodology(2001)

Consensus

TheCONSORTstatement:revisedrecommendationsforimprovingthequalityofreportsofparallelgrouprandomizedtrialsDavidMoher1,KennethFSchulz2andDouglasGAltman3

Address:1UniversityofOttawa,ThomasC.ChalmersCentreforSystematicReviews,Ottawa,Canada,2FamilyHealthInternationalandDept.ofObstetricsandGynecology,SchoolofMedicine,UniversityofNorthCarolinaatChapelHill,NorthCarolina,USAand3ICRFMedicalStatisticsGroupandCentreforStatisticsinMedicine,InstituteofHealthSciences,Oxford,UKfortheCONSORTGroup

E-mail:DavidMoher-dmoher@uottawa.ca;KennethFSchulz-kschulz@;DouglasGAltman-d.altman@icrf.icnet.uk

Published:20April2001

Received:30March2001

BMCMedicalResearchMethodology2001,1:2

Accepted:20April2001

Thisarticleisavailablefrom:/1471-2288/1/2

(c)2001Moheretal,licenseeBioMedCentralLtd.

Abstract

Tocomprehendtheresultsofarandomizedcontrolledtrial(RCT),readersmustunderstanditsdesign,conduct,analysisandinterpretation.Thatgoalcanonlybeachievedthroughcompletetransparencyfromauthors.Despiteseveraldecadesofeducationalefforts,thereportingofRCTsneedsimprovement.InvestigatorsandeditorsdevelopedtheoriginalCONSORT(ConsolidatedStandardsofReportingTrials)statementtohelpauthorsimprovereportingbyusingachecklistandflowdiagram.TherevisedCONSORTstatementpresentedinthispaperincorporatesnewevidenceandaddressessomecriticismsoftheoriginalstatement.

ThechecklistitemspertaintothecontentoftheTitle,Abstract,Introduction,Methods,ResultsandDiscussion.Therevisedchecklistincludes22-itemsselectedbecauseempiricalevidenceindicatesthatnotreportingtheinformationisassociatedwithbiasedestimatesoftreatmenteffectortheinformationisessentialtojudgethereliabilityorrelevanceofthefindings.WeintendedtheflowdiagramtodepictthepassageofparticipantsthroughanRCT.Therevisedflowdiagramdepictsinformationfromfourstagesofatrial(enrolment,interventionallocation,follow-up,andanalysis).Thediagramexplicitlyincludesthenumberofparticipants,foreachinterventiongroup,includedintheprimarydataanalysis.Inclusionofthesenumbersallowsthereadertojudgewhethertheauthorshaveperformedanintention-to-treatanalysis.

Insum,theCONSORTstatementisintendedtoimprovethereportingofanRCT,enablingreaderstounderstandatrial'sconductandtoassessthevalidityofitsresults.

Contributors

FrankDavidoff,MD,AnnalsofInternalMedicine,(Phil-adelphia,PA);SusanEastwood,ELS(D),UniversityofCaliforniaatSanFrancisco,(SanFrancisco,CA);Matth-iasEgger,MD,DepartmentofSocialMedicine,Universi-tyofBristol,(Bristol,UK);DianaElbourne,PhD,LondonSchoolofHygieneandTropicalMedicine,(Lon-don,UK);PeterG?tzsche,MD,NordicCochraneCentre,(Copenhagen,Denmark);SylvanB.Green,PhD,MD,

SchoolofMedicine,CaseWesternReserveUniversity,(Cleveland,OH);LeniGrossman,BA,Merck&Co.,Inc.,(WhitehouseStation,NJ);BarbaraS.Hawkins,MD,WilmerOphthalmologicalInstitute,JohnsHopkinsUni-versity,(Baltimore,MD);RichardHorton,MB,TheLan-cet,(London,UK);WayneB.Jonas,MD,UniformedServicesUniversityoftheHealthSciences,(Bethesda,MD);TerryKlassen,MD,DepartmentofPediatrics,Uni-versityofAlberta,(Edmonton,Alberta);LeahLepage,

BMCMedicalResearchMethodology(2001)1:2

PhD,ThomasC.ChalmersCentreforSystematicRe-views,(Ottawa,ON);ThomasLang,MA,TomLangCommunications,(Lakewood,OH);JeroenLijmer,MD,Dept.ofClinicalEpidemiology,UniversityofAmster-dam,(Amsterdam,TheNetherlands);RickMalone,BS,TAPPharmaceuticals,(LakeForest,IIL);CurtisL.Mein-ert,PhD,JohnsHopkinsUniversity,(Baltimore,MD);MaryMosley,BS,LifeSciencePublishing,(Tokyo,Ja-pan);StuartPocock,PhD,LondonSchoolofHygieneandTropicalMedicine,(London,UK);DrummondRen-nie,JAMA,Chicago,IIL);DavidS.Riley,MD,UniversityofNewMexicoMedicalSchool,(SantaFe,NM);RobertaW.Scherer,MD,Epidemiology&PreventiveMedicine,UniversityofMarylandSchoolofMedicine,(Baltimore,MD);IdaSim,MD,PhD,UniversityofCaliforniaatSanFrancisco,(SanFrancisco,CA);DonnaStroup,PhD,MSc,EpidemiologyProgramOffice,CenterforDiseaseControl&Prevention,(Atlanta,GA).

DavidMoher,KenSchulz,andDougAltmanparticipatedinregularconferencecalls,identifiedparticipants,con-tributedintheCONSORTmeetingsanddraftedthemanuscript.DavidMoherandLeahLepageplannedtheCONSORTmeetings,identifiedandsecuredfunding,in-vitedtheparticipantsandplannedthemeetingagenda.ThemembersoftheCONSORTgrouplistedaboveat-tendedtheconsortmeetingsandprovidedinputintherevisedchecklist,flowdiagramand/ortextofthismanu-script.DavidMoheristheGuarantorofthemanuscript.

Introduction

Areportofarandomizedcontrolledtrial(RCT)shouldconveytothereader,inatransparentmanner,whythestudywasundertaken,andhowitwasconductedandan-alyzed.Forexample,alackofadequatelyreportedrand-omizationhasbeenassociatedwithbiasinestimatingtheeffectivenessofinterventions,[1,2].ToassessthestrengthsandlimitationsofanRCT,readersneedanddeservetoknowthequalityofitsmethodology.

Despiteseveraldecadesofeducationalefforts,RCTsstillarenotbeingreportedadequately,[3–6].Forexample,areview,[5]of122recentlypublishedRCTsthatevaluatedtheeffectivenessofselectiveserotoninreuptakeinhibi-tors(SSRI)asfirst-linemanagementstrategyfordepres-sionfoundthatonlyone(0.8%)paperdescribedrandomizationadequately.InadequatereportingmakestheinterpretationofRCTsdifficultifnotimpossible.Moreover,inadequatereportingbordersonunethicalpracticewhenbiasedresultsreceivefalsecredibility.

HistoryofCONSORT

/1471-2288/1/2

ternationalgroupofclinicaltrialists,statisticians,epide-miologistsandbiomedicaleditors.CONSORThasbeensupportedbyagrowingnumberofmedicalandhealthcarejournals,[8–

11]

andeditorialgroups,includingtheInternationalCommitteeofMedicalJournalEditors,

(ICMJE,TheVancouverGroup),theCouncilofSci-enceEditors(CSE),andtheWorldAssociationofMedi-calEditors(WAME).CONSORTisalsopublishedinDutch,English,French,German,Japanese,andSpan-ish.ItcanbeaccessedtogetherwithotherinformationabouttheCONSORTgroupontheInternet,[13].

TheCONSORTstatementcomprisesachecklistandflowdiagramforreportinganRCT.Forconvenience,thechecklistanddiagramtogetherarecalledsimplyCON-SORT.Theyareprimarilyintendedforuseinwriting,re-viewing,orevaluatingreportsofsimpletwo-groupparallelRCTs.

PreliminaryindicationsarethattheuseofCONSORTdoesindeedhelptoimprovethequalityofreportsofRCTs,[14,15].Inanevaluation,[14]of71publishedRCTs,inthreejournalsin1994,allocationconcealmentwasnotclearlyreportedin61%(n=43)oftheRCTs.Fouryearslater,afterthesethreejournalsrequiredau-thorsreportinganRCTtouseCONSORT,theproportionofpapersinwhichallocationconcealmentwasnotclear-lyreportedhaddroppedto39%(30of77,meandiffer-ence=-22%;95%confidenceintervalofthedifference:-38%,-6%).

TheusefulnessofCONSORTisenhancedbycontinuousmonitoringofthebiomedicalliteraturethatpermitsittobemodifieddependingonthemeritsofmaintaining,ordroppingcurrentitemsandincludingnewitems.Forex-ample,whenMeinert,[16]observedthattheflowdia-gramdidnotprovideimportantinformationaboutthenumberofparticipantswhoenteredeachphaseofanRCT(i.e.,enrollment,treatmentallocation,follow-up,anddataanalysis),thediagramcouldbemodifiedtoac-commodatetheinformation.Thechecklistissimilarlyflexible.

ThisiterativeprocessmakestheCONSORTstatementacontinuallyevolvinginstrument.WhileparticipantsintheCONSORTgroupandtheirdegreeofinvolvementvaryovertime,membersmeetregularlytoreviewtheneedtorefineCONSORT.Atthe1999meetingadecisionwasmadetorevisetheoriginalstatement.Thisreportre-flectschangesdeterminedbyconsensusoftheCON-SORTgroup,partlyinresponsetoemergingevidenceontheimportanceofvariouselementsofRCTs.

Inthemid1990s,twoindependentinitiativestoimprove

thequalityofreportsofRCTsledtothepublicationofthe

CONSORTstatement,[7]whichwasdevelopedbyanin-

BMCMedicalResearchMethodology(2001)1:2 /1471-2288/1/2

Table1:Checklistofitemstoincludewhenreportingarandomizedtrial

PAPERSECTION

Item

Descriptor

Reported

Andtopic

#

onpage#

TITLE&ABSTRACT

1

Howparticipantswereallocatedtointerventions(e.g.,randomallocation","ran-

domized",or"randomlyassigned").

INTRODUCTION

Background

2

Scientificbackgroundandexplanationofrationale.

METHODS

Participants

3

Eligibilitycriteriaforparticipantsandthesettingsandlocationswherethedata

werecollected.

Interventions

4

Precisedetailsoftheinterventionsintendedforeachgroupandhowandwhen

theywereactuallyadministered.

Objectives

5

Specificobjectivesandhypotheses.

Outcomes

6

Clearlydefinedprimaryandsecondaryoutcomemeasuresand,whenapplicable,

anymethodsusedtoenhancethequalityofmeasurements(e.g.,multipleobser-

vations,trainingofassessors).

Samplesize

7

Howsamplesizewasdeterminedand,whenapplicable,explanationofanyinter-

imanalysesandstoppingrules.

Randomization:

Sequencegeneration

8

Methodusedtogeneratetherandomallocationsequence,includingdetailsofany

restriction(e.g.,blocking,stratification).

Allocationconcealment

9

Methodusedtoimplementtherandomallocationsequence(e.g.,numbered

containersorcentraltelephone),clarifyingwhetherthesequencewasconcealed

untilinterventionswereassigned.

Implementation

10

Whogeneratedtheallocationsequence,whoenrolledparticipants,andwhoas-

signedparticipantstotheirgroups.

Blinding(Masking)

11

Whetherornotparticipants,thoseadministeringtheinterventions,andthoseas-

sessingtheoutcomeswereblindedtogroupassignment.Ifdone,howthesuccess

ofblindingwasevaluated.

Statisticalmethods

12

Statisticalmethodsusedtocomparegroupsforprimaryoutcome(s);Methodsfor

additionalanalyses,suchassubgroupanalysesandadjustedanalyses.

RESULTS

Participantflow

13

Flowofparticipantsthrougheachstage(adiagramisstronglyrecommended).

Specifically,foreachgroupreportthenumbersofparticipantsrandomlyassigned,

receivingintendedtreatment,completingthestudyprotocol,andanalyzedfor

theprimaryoutcome.Describeprotocoldeviationsfromstudyasplanned,to-

getherwithreasons.

Recruitment

14

Datesdefiningtheperiodsofrecruitmentandfollow-up.

Baselinedata

15

Baselinedemographicandclinicalcharacteristicsofeachgroup.

Numbersanalyzed

16

Numberofparticipants(denominator)ineachgroupincludedineachanalysisand

whethertheanalysiswasbyintention-to-treat".Statetheresultsinabsolutenum-

berswhenfeasible(e.g.,10/20,not50%).

OutcomesandEstimation

17

Foreachprimaryandsecondaryoutcome,asummaryofresultsforeachgroup,

andtheestimatedeffectsizeanditsprecision(e.g.,95%confidenceinterval).

Ancillaryanalyses

18

Addressmultiplicitybyreportinganyotheranalysesperformed,includingsub-

groupanalysesandadjustedanalyses,indicatingthosepre-specifiedandthoseex-

ploratory.

Adverseevents

19

Allimportantadverseeventsorsideeffectsineachinterventiongroup.

DISCUSSION

Interpretation

20

Interpretationoftheresults,takingintoaccountstudyhypotheses,sourcesofpo-

tentialbiasorimprecisionandthedangersassociatedwithmultiplicityofanalyses

andoutcomes.

Generalizability

21

Generalizability(externalvalidity)ofthetrialfindings.

Overallevidence

22

Generalinterpretationoftheresultsinthecontextofcurrentevidence.

BMCMedicalResearchMethodology(2001)1:2 /1471-2288/1/2

RevisionoftheCONSORTstatement

ThirteenmembersoftheCONSORTgroupmetinMay1999withtheprimaryobjectiveofrevisingtheoriginalCONSORTchecklistandflowdiagram,asneeded.Themeritsofincludingeachitemwerediscussedbythegroupinthelightofcurrentevidence.AsindevelopingtheoriginalCONSORTstatement,ourintentionwastokeeponlythoseitemsdeemedfundamentaltoreportingstandardsforanRCT.Someitemsnotconsideredessen-tialmaywellbehighlydesirableandshouldstillbein-cludedinanRCTreporteventhoughtheyarenotincludedinCONSORT.Suchitemsincludeinstitutionalethicalreviewboardapproval,sourcesoffundingforthetrial,andatrialregistrynumber(as,forexample,theIn-ternationalStandardRandomizedControlledTrialNumber(ISRCTN)usedtoregistertheRCTatitsincep-tion[17].

Shortlyafterthemeetingarevisedversionofthecheck-listwascirculatedtothegroupforadditionalcommentsandfeedback.Revisionstotheflowdiagramsimilarlyweremade.AllofthesechangeswerediscussedwhenCONSORTparticipantsmetinMay2000andtherevisedstatementfinalizedshortlyafterwards.

TherevisedCONSORTstatementincludesa22-itemchecklist(Table1)andaflowdiagram(Figure1).Itspri-maryaimishelpingauthorsimprovethequalityofre-portsofsimpletwo-groupparallelRCTs.However,thebasicphilosophyunderlyingthedevelopmentofthestatementcanbeappliedtoanydesign.Inthisregardad-ditionalstatementsforotherdesignswillbeforthcomingfromthegroup,[13].CONSORTcanalsobeusedbypeerreviewersandeditorstoidentifyreportswithinadequatedescriptionoftrialsandthosewithpotentiallybiasedre-sults,[1,2].

Duringthe1999meetingthegroupalsodiscussedthebenefitsofdevelopinganexplanatorydocumenttoen-hancetheuseanddisseminationofCONSORT.Thedoc-umentispatternedonreportingofstatisticalaspectsofclinicalresearch,

[18]

whichwasdevelopedtohelpfacil-itatetherecommendationsoftheICMJEsUniformRe-quirementsforManuscriptsSubmittedtoBiomedicalJournals.ThreemembersoftheCONSORTgroup(DGA,KFS,DM),withassistancefrommembersonsomechecklistitems,draftedanexplanationandelaborationdocument.Thatdocument,[19]wascirculatedtothegroupforadditionsandrevisionsandwaslastrevisedaf-terreviewatthelatestCONSORTgroupmeeting.

ChangestoCONSORT

Intherevisedchecklistanewcolumnfor"papersectionandtopic"integratesinformationfromthe"subheading"columnthatwascontainedintheoriginalstatement.

The"Wasitreported?"columnhasbeenintegratedintoa"reportedonpage#"column,asrequestedbysomejournals.

Eachitemofthechecklistisnownumberedandthesyntaxandorderhavebeenrevisedtoimprovetheflowofinformation.

The"Title"and"Abstract"arenowcombinedinthefirstitem.

Whilethecontentoftherevisedchecklistissimilartotheoriginalonesomeitemsthatpreviouslywerecom-binedarenowseparate.Forexample,previouslyauthorswereaskedtodescribe"primaryandsecondaryout-come(s)measure(s)andtheminimumimportantdiffer-ence(s),andindicatehowthetargetsamplesizewasprojected".Inthenewversionissuespertainingtoout-comes(item6)andsamplesize(item7)areseparate,en-ablingauthorstobemoreexplicitabouteach.Moreover,someitemsrequestadditionalinformation.Forexam-ple,foroutcomes(item6)authorsareaskedtoreportanymethodsusedtoenhancethequalityofmeasure-ments,suchasmultipleobservations.

Theitemaskingfortheunitofrandomization(e.g.,cluster)hasbeendroppedbecausespecificchecklistshavebeendevelopedforreportingclusterRCTs,[20]andotherdesigntypes,[13]sincepublicationoftheorig-inalchecklist.

Wheneverpossiblenewevidenceisincorporatedintotherevisedchecklist.Forexample,authorsareaskedtobeexplicitaboutwhethertheanalysisreportedisbyin-tention-to-treat(item16).Thisrequestisbasedinpartontheobservations,[21]that(a)authorsdonotade-quatelydescribeandapplyintention-to-treatanalysisand(b)reportsthatdonotprovidethisinformationarelesslikelytoreportotherrelevantinformation,suchaslossestofollow-up,[22].

Therevisedflowdiagramdepictsinformationfromfourstagesofatrial(enrolment,interventionallocation,fol-low-up,andanalysis).Thereviseddiagramexplicitlyin-cludesthenumberofparticipants,foreachinterventiongroup,includedintheprimarydataanalysis.Inclusionofthesenumbersletsthereaderknowwhethertheau-thorshaveperformedanintentiontotreatanalysis,

[21

–

23].

Becausesomeoftheinformationmaynotalwaysbeknownandtoaccommodateotherinformation,thestructureoftheflowdiagrammayneedtobemodifiedforaparticulartrial.Inclusionoftheparticipantflowdi-agraminthereportisstronglyrecommendedbutmaybeunnecessaryforsimpletrials,suchasthosewithoutanyparticipantwithdrawalsordropouts.

BMCMedicalResearchMethodology(2001)1:2 /1471-2288/1/2

Figure1

FlowDiagramoftheprogressthroughthephasesofarandomizedtrial(i.e.,enrollment,interventionallocation,follow-up,anddataanalysis)

BMCMedicalResearchMethodology(2001)1:2

Discussion

Specificallydevelopedtoprovideguidancetoauthorsabouthowtoimprovethequalityofreportingofsimpletwo-groupparallelRCTs,CONSORTencouragestrans-parencywhenreportingthemethodsandresultssothatreportsofRCTscanbeinterpretedbothreadilyandac-curately.However,CONSORTdoesnotaddressotherfacetsofreportingthatalsorequireattention,suchasscientificcontentandreadabilityofRCTreports.SomeauthorsintheirenthusiasmtouseCONSORThavemod-ifiedthechecklist,[24].WerecommendagainstsuchmodificationsbecausetheymaybebasedonadifferentprocessthantheoneusedbytheCONSORTgroup.

TheuseofCONSORTseemstoreduce(ifnoteliminate)inadequatereportingofRCTs[14,15].Potentially,theuseofCONSORTshouldhaveapositiveinfluenceonhowRCTsareconducted.Grantingagencieshavenotedthispotentialrelationshipandinatleastinonecase,[25]haveencouragedgranteestoconsiderintheirapplica-tionhowtheyhavedealtwiththeCONSORTitems.

Theevidence-basedapproachusedtodevelopCON-SORThasalsobeenusedtodevelopstandardsforre-portingmeta-analysesofrandomizedtrials,[26],meta-analysesofobservationalstudies,[27]anddiagnosticstudies(personalcommunication-JeroenLijmer).Healtheconomistshavealsostartedtodevelopreportingstandards,[28]tohelpimprovethequalityoftheirre-ports,[29].Theintentofalloftheseinitiativesistoim-provethequalityofreportingofbiomedicalresearch,

andbydoingsotobringaboutmoreeffectivehealth

care.

TherevisedCONSORTstatementwillreplacetheorigi-naloneinthosejournalsandgroupsthatalreadysupportit.JournalsthatdonotyetsupportCONSORTmaydosobyregisteringontheCONSORTInternetsite,[13].Inor-dertoconveytoauthorstheimportanceofimprovedqualityinthereportingofRCTs,weencouragesupport-ingjournalstoreferencetherevisedCONSORTstate-mentandtheCONSORTInternetaddress,[13]intheir"InstructionstoContributors".Asthejournalspublish-ingtherevisedCONSORTstatementhavewaivedcopy-rightprotection,CONSORTisnowwidelyaccessibletothebiomedicalcommunity.TheCONSORTchecklistandflowdiagramcanalsobeaccessedattheCONSORTIn-ternetsite,[13].

AlackofclarificationofthemeaningandrationaleforeachchecklistitemintheoriginalCONSORTstatementhasbeenremediedwiththedevelopmentoftheCON-SORTexplanationandelaborationdocument,[19]whichcanalsobefoundontheCONSORTInternetsite,[13].Thisdocumentincludesreportingtheevidenceon

/1471-2288/1/2

whichthechecklistitemsarebased,includingtherefer-ences,whichannotatedthechecklistitemsintheprevi-ousversion.WeencouragejournalstoalsoincludereferencetothisdocumentalsointheirInstructionstoContributors.

EmphasizingtheevolvingnatureofCONSORT,theCONSORTgroupinvitesreaderstocommentontheup-datedchecklistandflowdiagramthroughtheCONSORTInternetsite,[13].Commentsandsuggestionswillbecollatedandconsideredatthenextmeetingofthegroupin2001.

Footnote

TherevisedCONSORTstatementisalsopublishedinJAMA(MoherD,SchulzKF,AltmanDG,fortheCON-SORTgroup.TheCONSORTstatement:revisedrecom-mendationsforimprovingthequalityofreportsofparallelgrouprandomizedtrials.JAMA2001;285:1987-1991),TheLancet(MoherD,SchulzKF,Alt-manDG,fortheCONSORTgroup.TheCONSORTstate-ment:revisedrecommendationsforimprovingthequalityofreportsofparallelgrouprandomizedtrialsLancet;2001;357:1191-1194),andAnnalsofInternalMedicine(MoherD,SchulzKF,AltmanDG,fortheCONSORTgroup.TheCONSORTstatement:revisedrecommendationsforimprovingthequalityofreportsofparallelgrouprandomizedtrials.AnnalsofInternalMedicine;2001;134:657-662).Authorscanuseanyoneofthesereferences,aswellasthereferenceinBMCMed-icalResearchMethodology,whencitingCONSORT.

Acknowledgements

Theefforttoimprovethereportingofrandomizedtrials,fromitsbegin-ningswiththeStandardsofReportingTrials(SORT)grouptothecurrentactivitiesoftheConsolidatedStandardsofReportingTrials(CONSORT)group,haveinvolvedalargenumberofpeoplearoundtheglobe.WewishtothankLeahLepageforkeepingeverybodyalllinedupandmovinginthesamedirection.

FinancialsupporttoconvenemeetingsoftheCONSORTgroupwaspro-videdinpartbyAbbottLaboratories,AmericanCollegeofPhysicians,Glax-oWellcome,TheLancet,Merck,theCanadianInstitutesforHealthResearch,NationalLibraryofMedicine,andTAPPharmaceuticals.

References

SchulzKF,ChalmersI,HayesRJ,AltmanDG:

Empiricalevidence

ofbias:dimensionsofmethodologicalqualityassociatedwithestimatesoftreatmenteffectsincontrolledtrials.JAMA1995,273:408-412

MoherD,PhamB,JonesA,CookDJ,JadadAR,MoherM,TugwellP:

Doesthequalityofreportsofrandomisedtrialsaffectesti-matesofinterventionefficacyreportedinmeta-analyses?Lancet1998,352:609-613

JadadAR,BoyleM,CunninghamC,KimM,SchacharR:TreatmentofAttention-Deficit/HyperactivityDisorder.EvidenceReport/TechnologyAssessmentNo.11(PreparedbyMcMasterUniversityunderContractNo.290-97-0017),2000,

ThornleyB,AdamsCE:

Contentandqualityof2000controlledtrialsinschizophreniaover50yearsBMJ1998,317:1181-1184

HotopfM,LewisG,NormandC:

Puttingtrialsontrial-thecosts

andconsequencesofsmalltrialsindepression:asystematic

BMCMedicalResearchMethodology(2001)1:2

reviewofmethodology.JEpidemiolCommunityHealth1997,51:354-358

DickinsonK,BunnF,WentzR,EdwardsP,RobertsI:

Sizeandqual-

ityofrandomisedcontrolledtrialsinheadinjury:reviewofpublishedstudies.BMJ2000,320:1308-1311

BeggCB,ChoMK,EastwoodS,HortonR,MoherD,OlkinI,Rennie

D,SchulzKF,SimelDL,StroupDF:

Improvingthequalityofre-portingofrandomizedcontrolledtrials:theCONSORTstatement.JAMA1996,276:637-639

FreemantleN,MasonJM,HainesA,EcclesMP:

CONSORT:anim-portantsteptowardevidence-basedhealthcare.AnnInternMed1997,126:81-83

AltmanDG:

Betterreportingofrandomizedcontrolledtrials:theCONSORTstatement.BMJ1996,313:570-571

SchulzKF:

Thequestforunbiasedresearch:randomizedclin-icaltrialsandtheCONSORTreportingguidelines.AnnNeurol1997,41:569-573

HustonP,HoeyJ:

CMAJendorsestheCONSORTstatement.CMAJ1996,155:1277-1279

DavidoffF:

NewsfromtheInternationalCommitteeofMedi-calJournalEditors.AnnInternMed2000,133:229-231

2000,[]

MoherD,JonesA,LepageL,:

UseoftheCONSORTstatement

andqualityofreportsofrandomizedtrials:acomparativebeforeandafterevaluation?JAMA2001,285:1992-1995

EggerM,JüniP,BartlettC,:

Thevalueofpatientflowchartsin

reportsofrandomizedcontrolledtrials:bibliographicstudy.JAMA2001,285:1996-1999

MeinertCL:

BeyondCONSORT:Needforimprovedreport-ingstandardsforclinicaltrials.JAMA1998,279:1487-1489

ChalmersI:CurrentControlledTrials:anopportunitytohelpimprovethequalityofclinicalresearch.CurrControlTrialsCardi-ovascMed2000,1:3-8

BailerJCIII,MostellerF:

Guidelinesforstatisticalreportingin

articlesformedicaljournals:amplificationsandexplana-tions.AnnInternMed1988,108:266-273

AltmanDG,SchulzKF,MoherD,EggerM,DavidoffF