还剩4页未读，继续阅读
文本内容:
Improving theReporting Qualityof NonrandomizedEvaluationsof Behavioraland Public Health Interventions:The TRENDStatementDeveloping anevidence basefor makingpublic DonC.DesJarlais,PhD,CynthiaLyles,PhD,NicoleCrepaz,PhD,andtheTREND Grouphealthdecisions willrequire using data fromevaluation studies withrandomized andnonrandomizeddesigns.Assessingindividualstudies argumentthat evidence-based publicimprove the reporting qualityof agreed-upon framework for theandusing studies in quantitativeresearch syntheseshealthwill neces-sarily involvethe thesetypes ofstudies.The transparent reporting ofrequiretransparent reporting of thestudy,withuse of research designsother thantransparency,or clarity,in the nonrandomized research evaluations.suf-ficient detailand clarity toread-ily seedifferencesRCTs.Most important,they arguereporting ofindividual stud-ies isThe HIV/AIDS Preventionandsimilarities amongstudiesin the samearea.Thethat RCTsare oftennot practicalor key.Sufficient detailandclarityin ResearchSynthesis PRSteam ofConsolidatedStandards ofReporting Trialsnotethical forevaluating manythereportallow readersto theCenters forDisease Con-trol andCONSORT statement providesguidelines forpub-lic health interventions andunderstand theconduct andfindings PreventionCDC has beentransparentre-portingofrandomizedclinicaltrials.dis-cuss methodsfor drawingcausal of the interventionstudy andhow thesynthesizing evidencefromWe presentthe initialversion of the Transparentinferencesfrom nonrandomizedstudy wasdifferent fromor similartoReporting ofEvaluations withNonrandomizedHIV behavioralintervention studiesDesignsTREND statement.These guidelinesevaluation designsplausibility”other studiesin thefield.involving RCTand non-randomizedemphasize the reporting of theories usedand and“adequacy”designs in their Furthermore,evidcnce-bascddesigns.The PRSteam foundthatdescriptions of intervention andcomparison terminology.practice mayoften relyonmany studyreports failedto includeconditions,researchdesign,andmethodsofadjustingAlso inthis issue,Donner andmetaanalyses oflarge numbersof7critical informatione.g.,for possiblebiases inevaluation studiesthat use8studies,some ofwhich mayreportintervention timingand dosage,Klar,Munay etal.,and Var-nell9nonrandomized designs.Am JPublicHealth.negative results.Metaanalysiset al.provide overviewsof theeffect sizedata necessaryfor2004;94:361-366requires full reporting ofmethodsbenefits andpitfalls of the group-research synthe-OVER THEPAST SEVERALdecades,and outcomesto en-able assessmentrandomizedtrial,which,in some[3—17a strongmovement towardses.To improve their ability toof comparabilityof differentstudies.situations,may bea reasonableevidence-based medicine[—3synthesize HIVbehavioralInadequate,or nontransparent,alternative to the RCT.There arehasemerged.In thecontext ofpreventionresearch,the PRSteamreporting maymake itdifficult toalsoa widevari-ety ofevidence-based medicine,clinicalconvened theCDCs Jour-nal Editorsunder-stand thevariables that affectnonrandomized evalua-tion designsdecisionsare based on thebestMeeting inAtlanta,Ga,on Julyin-tervention outcomesand thethatcan contributeimportant dataonavailable scientificdata ratherthan24-25,2003;this meetingwascentral elementsin interventionthe efficacy oreffectiveness ofoncustomary practicesor theattendedby editorsandinterventions,such assuccess orfailure overmultiplepersonal beliefsof thehealth carerepresentativesof18jour-nals thatquasi-experimental designs,10studies.provider.There isnow aparallelpublish HIVbehavioral interventionnonrandomizedtrials,and naturalmovementtoward evi-dence-studies acomplete listof theexperiments.Includ-ing thesetypes Inrecent years,efforts havebeen45journals isavailable from the authorsbased public healthprac-tices.”The ofdesigns indeveloping madeto improvethe qual-ity ofmovementis in-tended toutilize theevidcnce-bascd recommendationsreporting of RCTs.The Consolidatedand athttp://.The maingoals of thebest availablescientific knowledgecan providea moreintegrated pictureStandards ofReporting Trialsmeeting wereto1communicate theasthe foundationfor publicof theexisting evidenceand couldCONSORTstatement11provides ausefulness andimportance ofhealth-related decisionmaking.help tostrengthen public health22-itcm checklistand subjectflow ad-equate reporting standards,In thecontext ofevidencebased prac-tice.Excluding datacollected chartfor thetransparent reporting of2reach consensuson reportingmedicine,the randomizedcontrolled undersuch designswould RCTs.This statement hasbeenstandards forbehavioraltrial RCTis usuallyconsidered ofun-doubtcdly biasthe evidencebase adoptedas aframeworkfor the interventions,3develop achecklistgreatest eviden-tiary valuefor towardinterventions that are“easier”reportingofRCTs bya largenumber of reporting standardsto guideassessingtheeffi-cacy ofto evaluatebut notnecessarily moreof medical,clinical,and authorsand journalreviewers,and4interventions.Indeed,the preferenceeffective orcost-effective.psychological journals153,develop strategiesto disseminatethefor thisdesign issufficiently strongaccording to.org,as ofSeptember resultingreportingstandards.that whenempirical evidencefrom Ifnonrandomized designs are to16,
2003.Use ofthe CONSORTThe discussionsat themcet-ingRCTs isavailable,tweakef,designs besystematically used in buildingstatementhasimproved thequality broadenedto includestandardizedare oftenconsidered to be oflittle orevidence-based public health ofRCT reportsover thepast severalreportingof behavioral and publicno evidentiaryvalue.In thisissue,practices,it willbe neces-sary to12health interventions in general,years.〜There isyet,however,noVictora etal.6make astrong ratherthan focusingonly onHIVbehavioral inter-ventions.There was a definedintervention thatis beingwas used.To minimizeconfusion,it havingadditional informationfrom astrongcon-sensus atthe meetingin studiedand2a researchdesign thatwould behelpful for authors topublished studyprovided onaregard tomore standardizedand providesfor anassessment ofthe specifythe designthey intended to journaPsWeb siteor on antrans-parentreportingofresearchefficacy oreffectiveness ofthe use,particularly themethod ofevaluationsusing otherthan intervention.Thus,our proposedassignment,and anyvariations orran-domized designs,particularly checklistemphasizes description of deviationsfromthedesign.those withsome formofthe intervention,including thecom-parison group.This agreementtheoretical base;descriptionofthe Itis importantto notethat thewasreached with the realizationthat comparison condition;fullreportingTREND checklist is notintended toadditionalinput would be neededof outcomes;and incluserve asa criterionfor evaluatingfroma widevariety ofresearchers,sion ofinformation relatedtothepapers forpublication.Rather,it isotherjournal edi-tors,and designneeded toassess possibleintended toimprovethequality ofpractitionersin thepublic healthbiases inthe outcomedata.Brief datareporting inpeer-reviewedfield beforethe commentsmay behelpful for a fewpublications sothat theconduct andadoptionof afinal set ofreportingofthe items included inthe proposed thefindings ofresearch arestandards.TREND checklist.transparent.As thevolume ofpublicTable1presents aproposed healthliter-ature isconsistentlychecklist-the TransparentReporting•Use oftheory item
2.Behavioral expanding,research synthesisofEvaluations withNonrandomized andsocial sciencetheories providea becomesan importanttool forDesignsTREND checklist-for frameworkfor generatingcreating acumulative bodyofreporting standardsofbehavioraland cumulativeknowledge.Thus,it knowledge and makingpublichealthinterventionwouldbevery helpfulto includeevidence-based recommendationsofevaluations involvingreferences tothe theoret-ical basesof effectiveinterventions.Reportingnonrandomized designs.The the intervention beingevaluated.standards willhelp ensurethat fewerTRENDchecklist ismeant to be Thiswould per-mit identificationofinterventiontrials withconsistentwiththe CONSORT theoriesthat areuseful indeveloping nonrandomizeddesigns aremissingchecklist forthereportingofRCTs.inter-ventionsindifferent fields.information criticalfor researchItemspresented in boldface typein Someinterventions,however,are synthesisand thatcomparablethe tableare particularlyrelevant tobasedonatheoretical needsinformation acrossstudies canbebehavioral andpublichealthas-sessments orsimply themore easilyconsolidated andinterventionstud-ies,whether ornot experi-ence ofthe individualswho translatedinto generalizablerandomizeddesignsareused.Thus,de-signed the intervention.In theseknowledgeandpractice.we wouldsuggest thatthey beused situations,a posthoc applicationof aWe recognizeseveral challengestoexpand theinformation requestedtheory is not likelyto behelpful.in promotingand disseminatingreportingstandards forbyCONSORT forRCTs ofbehavioralandpublichealth•Description oftheinterventionnonrandomized interventionevaluations.Most important,theinterventions.Some oftheitems8,condition and the servicesprovidedTREND checklistis onlya suggested10,and15presented intheproposedin acomparisonconditionitem
4.setofguidelines andshould beTRENDchecklist are not relevantto Althoughspace islimited in manyconsidered awork inprogress.It isRCTsand,thus,not includedinthejournals,it isstill criticalto providehighlylikely thatimprovements willCONSORT checklist,but theyare sufficientdetail sothatareader hasbe necessary;moreover,adaptationsextremely relevantto anunderstanding ofthe contentandmay beneeded torefine thenonrandomizeddesigns.We alsodelivery ofboth theexperimen-talstandards forspecific fieldsofrefer readersto CONSORTintervention andthe servicesin theinterventionresearch,and additionalelaborationre-ports thatprovide comparisoncondition.For example,specifications forspecific typesofrationales andexamples foritems inusual careisnota helpfulnonrandomizedevaluation designsTable1thatareshared withthe descriptionof acomparisonare likelytobeneeded.Further-more,CONSORTchecklist.condition.page limitationsinmanyjournalsThe TRENDchecklistisproposed•Description ofthe researchdesigncreate strongpressure towardshorterfor interventionevaluationstudiesitem
8.We recognize that therecanrather thanlonger articles.Someusing nonrandomizeddesigns,not bemeaningful disagreementaboutalternatives wererecommended inforall researchusing nonrandomizedwhat researchdesign wasactuallythe CDCsJournal EditorsMeetingdesigns.Intervention evaluationusedinan interventionstudy,to resolvethe spaceissue,such asstudieswould necessarilyinclude1including whetheran RCTdesignTABLE1—The TRENDChecklist Version
1.0PaperSection/Topic ItemNo.Descriptor ExamplesFromHIVBehavioralPreventionResearch1Exampletitle:Anonrandomizedtrialofaclinic-basedHIVcounselinginterventionforAfricanTitleandabstract•InformationonhowunitswereallocatedtointerventionsAmericanfemaledrugusers•Structuredabstractrecommended•Informationontargetpopulationorstudysample•ScientificbackgroundandexplanationofrationaleIntroduction•TheoriesusedindesigningbehavioralinterventionsBackground2Exampletheoryused:thecommunity-basedAIDSinterventionwasbasedonsociallearningtheoryMethodsParticipants3•日igibility criteriafor participants,including criteriaat differentlevels inrecruitment/samplingplane.g.,cities,clinics,subjectsExamplesamplingmethod:usinganalphanumericsortedlistofpossiblevenuesandtimesfor•Method ofrecruitment e.g.,referral,self-selection,including thesamplingidentifyingeligiblesubjects,everytenthvenue-timeunitwasselectedforthelocationandmethodifasystematicsamplingplanwasimplementedtimingofrecruitmentExamplesrecruitmentsetting:subjectswereapproachedbypeeropinionleadersduring•Recruitmentsettingconversationsatgaybars•SettingsandlocationswherethedatawerecollectedInterventions4•Details ofthe interventionsintended foreach studycondition andhow andwhentheywereactuallyadministered,specificallyincluding:Content:whatwasgivenDeliverymethod:howwasthecontentgivenUnit ofdelivery:how weresubjects groupedduring deliveryDeliverer:whodeliveredtheinterventionSetting:wherewastheinterventiondelivered Exampleunitofdelivery:theinterventionwasdeliveredtosmallgroupsof5-8subjectsExposurequantityandduration:howmanysessionsorepisodesoreventswereExamplessetting:theinterventionwasdeliveredinthebars;theinterventionwasdeliveredintheintendedtobedeliveredHowlongweretheyintendedtolast waitingroomsofsexuallytransmitteddiseaseclinicsTimespan:how longwas it intendedtotaketo delivertheinterventionto eachExamplesexposurequantityandduration:theinterventionwasdeliveredinfive1-hoursessions;unit theintervention consistedof standardHIV counselingand testingpretest andposttestcounselingsessions,eachabout30minutesActivitiestoincreasecomplianceoradherencee.g.,incentives Examplestime span:each interventionsession wastobedelivered infive1-hour sessionsonceaweekfor5weeks;theinterventionwastobedeliveredovera1-monthperiod.Example activitiesto increasecompliance oradherence:bus tokensand foodstamps wereprovidedObjectives5•Specificobjectivesandhypotheses•ClearlydefinedprimaryandsecondaryoutcomemeasuresOutcomes6・Methods used to collectdata andany methodsusedtoenhance thequality ofExamplesmethodusedtocollectdata:self-reportofbehavioraldatausingaface-to-facemeasurements interviewer-administeredquestionnaire;audio-computer-assistedself-administered•Information onvalidated instrumentssuch aspsychometric andbiometric instrumentpropertiesSamplesize7・How samplesize wasdetermined and,when applicable,explanation ofanyinterimanalysesandstoppingrules•Unitofassignmenttheunitbeingassignedtostudycondition,Assignment8Example1assignmentmethod:subjectswereassignedtostudyconditionsmethod e.g.,individual,group,communityusinganalternatingsequencewhereineveryotherindividualenrollede.g.,1,3,5,etc.was•Methodusedtoassignunitstostudyconditions,includingdetailsofanyrestrictionassignedtothe interventionconditionandthe alternatesubjectsenrollede.g.,2,4,6,etc.e.g.,blocking,stratification,minimizationwereassignedtothecomparisoncondition・Inclusion ofaspects employedto helpminimize potentialbias induceddue tononrandomizatione.g.,matchingExample2assignmentmethod:foroddweekse.g.1,3,5,subjectsattendingthecliniconMonday,Wednesday,andFridaywereassignedtotheinterventionconditionandthoseattendingthecliniconTuesdayandThursdaywereassignedtothecomparisoncondition;thisassignmentwasreversedforevenweeksContinuedTABLE1—ContinuedBlinding9•Whetherornot participants,thoseadministeringtheinterventions,and Exampleblinding:the staffmember performingthe assessmentsmaskingthoseassessingtheoutcomeswereblindedtostudyconditionassignment;if wasnot involvedin implementingany aspectoftheintervention andso,statementregardinghowtheblindingwasaccomplishedandhowitwas knewthe participantsonly bytheir studyidentifier numberassessed•Descriptionofthesmallestunitthatisbeinganalyzedtoassessintervention Example1unitofanalysis:sincegroupsofindividualswereassignedtostudyconditions,theUnitofanalysis10effectse.g.,individual,group,orcommunity analyseswereperformedatthegrouplevel,wheremixedeffectsmodelswereusedto•Iftheunitofanalysisdiffersfromtheunitofassignment,theanalytical accountforrandomsubjecteffectswithineachgroupExample2unitofanalysis:sinceanalyseswereperformedattheindividuallevelandcommunitiesmethodusedtoaccountforthise.g.,adjustingthestandarderrorwererandomized,apriorestimateoftheintraclasscorrelationcoefficientwasusedtoadjusttheestimatesbythedesigneffectorusingmultilevelanalysisstandarderrorestimatesbeforecalculatingconfidenceintervals11•Statisticalmethodsusedtocomparestudygroupsforprimaryoutcomes,Statisticalmethods includingcomplexmethodsforcorrelateddata•Statisticalmethodsusedforadditionalanalyses,suchassubgroupanalysesandadjustedanalysis•Methodsforimputingmissingdata,ifused•StatisticalsoftwareorprogramsusedResultsParticipantflow12•Flow ofparticipants througheach stageofthestudy:enrollment,assignment,allocationandinterventionexposure,follow-up,analysisadiagramisstronglyrecommendedEnrollment:thenumbersofparticipantsscreenedforeligibility,foundtobeeligibleornoteligible,declinedtobeenrolled,andenrolledinthestudyAssignment:thenumbersofparticipantsassignedtoastudyconditionAllocationandinterventionexposure:thenumberofparticipantsassignedtoeachstudyconditionandthenumberofparticipantswhoreceivedeachinterventionFollow-up:thenumberofparticipantswhocompletedthefollow-upordidnotcompletethefollow-upi.e,,losttofollow-up,bystudyconditionAnalysis:thenumberofparticipantsincludedinorexcludedfromthemainanalysis,bystudycondition•Descriptionofprotocoldeviationsfromstudyasplanned,alongwithreasons・Datesdefiningtheperiodsofrecruitmentandfollow-upRecruitment13•BaselinedemographicandclinicalcharacteristicsofparticipantsineachstudyBaselinedata14condition•Baselinecharacteristicsforeachstudyconditionrelevanttospecificdiseasepreventionresearch ExamplebaselinecharacteristicsspecifictoHIVpreventionresearch:HIVserostatusandHIV•Baselinecomparisonsofthoselosttofollow-upandthoseretained,overallandby testingbehaviorstudycondition•ComparisonbetweenstudypopulationatbaselineandtargetpopulationofinterestBaseline•Dataonstudygroupequivalenceatbaselineandstatisticalmethods Examplebaselineequivalence:theinterventionandcomparisongroupsdidnot15statisticallydifferwithrespecttodemographicdatagender,age,race/ethnicity;equivalence usedtocontrolforbaselinedifferencesP.05foreach,buttheinterventiongroupreportedasignificantlygreaterbaselinefrequencyofinjectiondruguseP=.03;allregressionanalysesincludedbaselinefrequencyofinjectiondruguseasacovariateinthemodelNumbers16•Numberofparticipantsdenominatorincludedineachanalysisforeach Examplenumberofparticipantsincludedintheanalysis:theanalysisofcondomuseincludedonlythosewhoreportedatthe6-monthfollow-uphavinghadvaginaloranalsexinthepast3analyzed studycondition,particularlywhenthedenominatorschangefordifferentmonths75/125forinterventiongroupand35/60forstandardgroupoutcomes;statementoftheresultsinabsolutenumberswhenfeasible•IndicationofExampleintentiontotreat:theprimaryanalysiswasintentiontotreatandincludedallsubjectswhethertheanalysisstrategywasintentiontotreator,ifnot,descriptionofhownoncompliersweretreatedintheanalyses asassignedwithavailable9-monthoutcomedata125of176assignedtotheinterventionand110of164assignedtothestandardconditionContinuedTABLE1—ContinuedOutcomesand17estimation•Foreachprimaryandsecondaryoutcome,asummaryofresultsforeachstudycondition,andtheestimatedeffectsizeandaconfidenceintervaltoindicatetheprecision•Inclusionofnullandnegativefindings•Inclusionofresultsfromtestingprespecifiedcausalpathwaysthroughwhichtheinterventionwasintendedtooperate,ifany•Summaryofotheranalysesperformed,includingsubgrouporrestrictedExampleancillaryanalyses:althoughthestudywasnotpoweredforthishypothesis,analyses,indicatingwhichareprespecifiedorexploratoryanexploratoryanalysisshowsthattheinterventioneffectwasgreateramongwomenthanamongmenalthoughnotstatisticallysignificantAncillary18analyses•SummaryofallimportantadverseeventsorunintendedeffectsineachExampleadverseevents:policecrackeddownonprostitution,whichdrovethetargetstudyconditionincludingsummarymeasures,effectsizeestimates,population,commercialsexworkers,toareasoutsidetherecruitment/samplingareaandconfidenceintervalsAdverseevents19•Interpretationoftheresults,takingintoaccountstudyhypotheses,sourcesofpotentialbias,imprecisionofmeasures,multiplicativeanalyses,andotherlimitationsorweaknessesofthestudyDiscussion•DiscussionofresultstakingintoaccountthemechanismbywhichtheinterventionwasintendedtoworkcausalpathwaysoralternativemechanismsorexplanationsInterpretation20•Discussionofthesuccessofandbarrierstoimplementingtheintervention,fidelityofimplementation•Discussionofresearch,programmatic,orpolicyimplications•Generalizability externalvalidity ofthe trialfindings,taking intoaccount thestudy population,thecharacteristics oftheintervention,length offollow-up,incentives,compliance rates,specificsites/settingsinvolvedinthestudy,andothercontextualissues•GeneralinterpretationoftheresultsinthecontextofcurrentevidenceandcurrenttheoryNote.Maskingblindingofparticipantsorthoseadministeringtheinterventionmaynotberelevantorpossibleformanybehavioralinterventions.Theoriesusedtodesigntheinterventionssee item2couldalso bereportedas partof item
4.The comparisonbetween studypopulation atbaseline andtarget populationof interestseeitem14could alsobe reportedas partofitem
21.Descriptors appearinginboldfaceare specificallyadded,modified,or furtheremphasized fromtheCONSORTstatement.Boldface topicand descriptorsarenotincludedintheCONSORT statementbut arerelevant forbehavioral interventionsusing nonrandomizedexperimental designs.The CONSORTstatement11ortheexplanation documentfor theCONSORTstatement18providesrelevantexamplesforanytopicordescriptorthatisnotinboldface.AstructuredformatofthediscussionispresentedinAnnals ofInternal Medicineinformationforauthors;,accessed September16,
2003.Generalizability21author^dissemination ofthese guidelinesTREND statementor byreferencing statementpresented hereisWeb siteor mustinvolve anongoing dialogueitintheir pub-lication guidelinesfor proposedasafirst steptohaving andmust beextended toa largeauthors warddeveloping standardizedandauthors numberof otherresearchers,and reviewers.To increasetransparent reportingforOverallevidence22send methodologists,and statisticiansaccessibility andease ofuse,thenonrandomizedinterventionadditional informationto relevantacross varioushealth-related revisedversions ofthe TRENDresearchevaluationsin publicresearchsynthesis groupsoraresearch fields.statement willbe postedonanopen health-related fields.■central repository.In aneffort toinitiate thisaccess Website http://.Finally,the processhas sofar dialogue,we inviteall editors,If themovement towardinvolvedonly CDCscientists andreviewers,authors,and readersto evidence-basedpublichealth isto Aboutthe Authorsjournaleditors ina singlemeeting providecomments andfeedback tosucceed,it willbenecessaryto DonC.DesJurlaisiswiththeBaronEdmonddeRothschildChemicalDependencyInstitute,Bethalong withthe preparationof thishelp usrevise thestandards.improve ourabilitytosynthesizeIsraelMedicalCenter,NewYorkCity.CynthiaLylescommentary.Although manyoftheComments canbe sentto,andtheresearch onpublic healthandNicoleCrepazarewiththeCentersforDiseasejournaleditors includedare notableTREND groupwill periodicallyinterventions.As Victoraand ControlandPrevention,Atlanta,Ga.TRENDGroupmembersarelistedbelowintheAcknowledgments.researchers inthe fieldsof HIV,revise theguidelines accord-ingly.colleagues note,this willincludeRequestsforreprintsshouldbesenttoDonC.publichealth,and drugabuse Also,journals areencour-aged tousingdatafrom interventionDesJarlais,PhD,BethIsraelMed-icalCenter,prevention,we real-izethatendorse thiseffort bypublishing evaluationsthat donot involveChemicalDependencyInsti-iule,FirstAveal16St,NewYork,NY10003e-mail:.successful promotionand editorialsor commen-taries onthe randomizeddesigns.The TRENDThisarticlewasacceptedOctober26,
2003.
2.USCochraneCenter.TheCochrane revisedCONSORTstatementforreportingCollaboration.Availableat:.AccessedJanuary5,randomizedtrials:explanationandelaboration.Ann
2004.InternMed.2001;134:663-
694.ContributorsDonDesJarlais,CynthiaLyles,andNicoleCrepaz
3.Evidence-BasedMedicineResourceCenter,developedtheinitialdraftofthearticleandmadethe NewYorkAcademyofMcdi-cinc.Hornepage.successiverevisions.AllmembersoftheTREND Availableat:http://.AccessedJanuary5,
2004.groupcontributedtotheideasinthearticleand
4.AgencyforHealthcareResearchandQuality.reviewedtheinitialdraftandrevisions.AHRQsevidence-basedpracticecenters.Availableat:http://.AccessedJanuary5,
2004.Acknowledgments
5.CentersforDiseaseControlandPrevention.MembersoftheTRENDGroupareasfollows:Guidetocommunitypreventiveservices.Availableat:KamranAbbasi,MBChB,MRCPBMJ,Williamhttp://www.hecommunityguide.org.AccessedBlattner,MDUniversityofMaryland,Baltimore;January5,
2004.JournalofAc-qitiredImmuneDeficiencySyndromes,
6.VictoraCG,HabichtJ-P,BryceJ.JayBernhardt,PhD,MPHEmoryUni-versity;HealthEvidcncc-bascdpublichealth:movingbeyondEducationResearch,BruceBullington,PhD Floridarandomizedtrials.AmJPublicHealth.StateUniversity;JournalofDrugIssues,Raul2004;94:400-
405.Caetano,MD,PhD,MPHUniversityofTexas;Addiction,TerryChambers,BA{Journalof
7.DonnerA,KlarN.PitfallsofandcontroversiesPsychoactiveDrugs,HarrisCooper,PhDDukeinclusterrandomizationtrials.AmJPublicHealth.University;Psycho-logicalBulletin,RoelA.Coutinho,2004:94:416-
422.MD,PhDUniversityofAmsterdam;AIDS,Nicole
8.MurrayDM,VarnellSP.BlitsteinJL.DesignCrepaz,PhD CentersforDiseaseControlandandanalysisofgroup-randomizedtrials:areviewofPrevention,JohnD.DeLa-mater,PhDUniversityofrecentmethodologicaldevelopments.AinJPublicWisconsin,Madison;JournalofSexResearch,DonHealth.2004;94:423-
432.C.DesJarlais,PhD BethIsraelMedicalCenter;AmericanJournalofPublicHealth},JeffreyH.Herbst,
9.Varnellelal.DesignandAnalysisofPhDCentersforDiseaseControlandPrevention,Group-RandomizedTrials:AReviewofRecentDavidHoltgrave,PhD EmoryUniver-sity,Barry Practices.AmJPublicHealth.2004;94:393-
399.Hong,PhDWashingtonUni-versityatSt.Louis;
10.CampbellDT,StanleyJC.ExperimentalandJournalofConsultingandClinicalPsychology,Qiiasi-EjcperimentalDesignsfarResearch.Boston,AngelaB.Hutchinson,PhD CentersforDiseaseMass:HoughtonMifflinCo;
1963.ControlandPrevention,SethKalich-man,PhDUniversityofConnecticut;HealthPsychology,Linda
11.MoherD,SchulzKF,AltmanDG.TheKay,MPH CentersforDiseaseControlandCONSORTstatement:revisedrecommendationsfbrPrevention,CynthiaLyles,PhDCentersforDisease improvingthequalityofreportsofparallcl-groupControlandPrevention,RobertMcNutt.MDRush randomizedtrials.Lancet.2001;357:1191-
1194.St.Luke,sMedicalCenteratChicago;Journalofthe
12.MoherD,JonesA,LepageL.UseoftheAmericanMedicalAssociation,IhoniasL.Patterson,CONSORTstatementandqualityofreportsofPhDUniversityofCalifor-nia,SanDiego;AIDSandrandomizedtrials:acomparativebefbre-and-afterBehavior,MichaelRoss,PhD,MPH,MA,MHPEdevaluation.JAMA.2001;285:1992-
1995.UniversityofTexas,Houston;AIDSCare,TheoSandforl,PhDColumbiaUniversity;Archivesof
13.ScmaanS,KayL,StrouseD,ctal.AprofileofSexualBehavior,RonStall,PhD,MPH CentersforU.S.-basedtrialsofbehavioralandsocialDis-easeControlandPrevention,FranciscoS.Sy,interventionsfbrHIVrisk-reduction.JAcquirImmuneMD,DrPHCentersforDiseaseControland DeficSyndr.2002;30suppl1:S3O-S5O.Prevention;AIDSEducationandPrevention,Cesar
14.ScmaanS,DesJarlaisDC,SogolowE,etal.AG.Victora,MD.PhD UniversidadeFederaldemeta-analysisoftheeffectofHIVpreventionPclotas,andDavidVlahov,PhD NewYorkinterventionsonthesexbehaviorsofdrugusersinAcademyofMedicine;AmericanJournaloftheUnitedStates.JAcquirImmuneDeficSyndr.Epidemiology,JournalofUrbanHealth.2OO2;3OsupplI:S73-S
93.
15.JohnsonW,HedgesL.RamirezG,etal.HIVpreventionresearchformenwhohavesexwithmen:asystematicreviewandmeta-analysis.JAcquirWealsothankthefollowingPreventionResearch ImmuneDeficSyndr.2002;30suppl1:S118-S
129.Synthesisteammembersfortheirassistanceinthe
16.MullenPD,RamirezG,StrouseD,HedgesL,meetingpreparationandcontributiontothedataSogolowE.Meta-analysisoftheeffectsofbehaviorreportingpresentation:JuliaBritton,TaneshaGriffin,HIVpreventioninterventionsonthesexualriskAngelaKim,MaryMullins,PaolaMarrero-Gonzalez,behaviorofsexuallyexperiencedadolescentsinU.S.SimaRama,R.ThomasSherba,andSekhartrials.JAcquirImmuneDeficSyndr.2002;30supplThadiparthi.l:S94-S
105.
17.NeumannMS,JohnsonWD,Se-maanS,etal.ReferencesReviewandmeta-analysisofHIVprevention
1.CentreforEvidence-BasedMedicine,interventionresearchforheterosexualadultUniversityDepartmentofPsychiatry,Warneford populationsintheUnitedStates.JAcquirImmuneHospital.Homepage.DeficSyndr.2OO2;3Osuppll:S106-Sl
17.Availableat:.AccessedJanuary5,
2004.
18.AltmanDG,SchultzKF,MoherD,etal.The。
个人认证
优秀文档
获得点赞 0
