Antioxidant supplements for prevention of mortality in healthy participants and patients with various diseases (Review) Bjelakovic G, NikolovaD, Gluud LL, Simonetti RG, Gluud C ThisisareprintofaCochranereview,preparedandmaintainedbyTheCochraneCollaborationandpublishedinTheCochraneLibrary 2009,Issue3 http://www.thecochranelibrary.com Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. TABLE OF CONTENTS HEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 PLAINLANGUAGESUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Figure1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Figure2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Figure3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Figure4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 AUTHORS’CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 CHARACTERISTICSOFSTUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 DATAANDANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 Analysis1.1.Comparison1Antioxidantsversusplacebo/nointervention,Outcome1Mortalityintrialswithaloworhigh riskofbias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194 Analysis1.2.Comparison1Antioxidantsversusplacebo/nointervention,Outcome2Mortalityinprimaryandsecondary preventiontrialswithaloworhighriskofbias. . . . . . . . . . . . . . . . . . . . . . . 197 Analysis1.3.Comparison1Antioxidantsversusplacebo/nointervention,Outcome3Mortalityafterexcludingtrials administratingextrasupplementsintheantioxidantgroup. . . . . . . . . . . . . . . . . . . 200 Analysis1.4.Comparison1Antioxidantsversusplacebo/nointervention,Outcome4Mortalityafterexcludingtrialswith extrasupplementsforbothinterventiongroups. . . . . . . . . . . . . . . . . . . . . . . 202 Analysis1.5.Comparison1Antioxidantsversusplacebo/nointervention,Outcome5Mortalityafterexcludingfactorial trialswithpotentialconfounding. . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 Analysis1.6.Comparison1Antioxidantsversusplacebo/nointervention,Outcome6Mortalityafterexcludingfactorial trialswithpotentialconfoundingandtrialswithextrasupplements. . . . . . . . . . . . . . . . 207 Analysis1.7.Comparison1Antioxidantsversusplacebo/nointervention,Outcome7Mortalityafterexcludingtrialswith anypotentialconfounding. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209 Analysis1.8.Comparison1Antioxidantsversusplacebo/nointervention,Outcome8Mortalityinbeta-carotenetrialswith aloworhighriskofbias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210 Analysis1.9.Comparison1Antioxidantsversusplacebo/nointervention,Outcome9MortalityinvitaminAtrialswitha loworhighriskofbias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 Analysis1.10.Comparison1Antioxidantsversusplacebo/nointervention,Outcome10MortalityinvitaminCtrialswith aloworhighriskofbias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 Analysis1.11.Comparison1Antioxidantsversusplacebo/nointervention,Outcome11MortalityinvitaminEtrialswith aloworhighriskofbias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 Analysis1.12.Comparison1Antioxidantsversusplacebo/nointervention,Outcome12Mortalityinseleniumtrialswitha loworhighriskofbias. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 Analysis1.13.Comparison1Antioxidantsversusplacebo/nointervention,Outcome13Mortalityinlow-biasriskbeta- carotenetrialswithoutselenium. . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 Analysis1.14.Comparison1Antioxidantsversusplacebo/nointervention,Outcome14Mortalityinlow-biasriskvitamin Atrialswithoutselenium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 Analysis1.15.Comparison1Antioxidantsversusplacebo/nointervention,Outcome15Mortalityinlow-biasriskvitamin Ctrialswithoutselenium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 Analysis1.16.Comparison1Antioxidantsversusplacebo/nointervention,Outcome16Mortalityinlow-biasriskvitamin Etrialswithoutselenium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 FEEDBACK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) i Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. WHAT’SNEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 HISTORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 CONTRIBUTIONSOFAUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 DECLARATIONSOFINTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 SOURCESOFSUPPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 NOTES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 INDEXTERMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) ii Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. [InterventionReview] Antioxidant supplements for prevention of mortality in healthy participants and patients with various diseases GoranBjelakovic1,DimitrinkaNikolova1,LiseLotteGluud1,RosaGSimonetti2,ChristianGluud1 1CochraneHepato-BiliaryGroup,CopenhagenTrialUnit,CentreforClinicalInterventionResearch,Department3344,Rigshospitalet, CopenhagenUniversityHospital,Copenhagen,Denmark.2DivisionediMedicina,OspedaleV.Cervello,Palermo,Italy Contact address: Goran Bjelakovic, Cochrane Hepato-Biliary Group, Copenhagen Trial Unit, Centre for Clinical Intervention Research, Department 3344, Rigshospitalet, Copenhagen University Hospital, Blegdamsvej 9, Copenhagen, DK-2100, Denmark. [email protected].(Editorialgroup:CochraneHepato-BiliaryGroup.) CochraneDatabaseofSystematicReviews,Issue3,2009(Statusinthisissue:Edited,commented) Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. DOI:10.1002/14651858.CD007176 Thisversionfirstpublishedonline:23April2008inIssue2,2008.Re-publishedonlinewithedits:8July2009inIssue3,2009. Lastassessedasup-to-date: 19February2008.(Helpdocument- DatesandStatusesexplained) This record should be cited as: Bjelakovic G, Nikolova D, Gluud LL, Simonetti RG, Gluud C. Antioxidant supplements for preventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases.CochraneDatabaseofSystematicReviews2008,Issue 2.Art.No.:CD007176.DOI:10.1002/14651858.CD007176. ABSTRACT Background Animalandphysiologicalresearchaswellasobservationalstudiessuggestthatantioxidantsupplementsmayimprovesurvival. Objectives Toassesstheeffectofantioxidantsupplementsonmortalityinprimaryorsecondarypreventionrandomisedclinicaltrials. Searchstrategy WesearchedTheCochraneLibrary(Issue3,2005),MEDLINE(1966toOctober2005),EMBASE(1985toOctober2005),andthe ScienceCitationIndexExpanded(1945toOctober2005).Wescannedbibliographiesofrelevantpublicationsandwrotetopharmaceutical companiesforadditionaltrials. Selectioncriteria Weincludedallprimaryandsecondarypreventionrandomisedclinicaltrialsonantioxidantsupplements(beta-carotene,vitaminA, vitaminC,vitaminE,andselenium)versusplaceboornointervention.Includedparticipantswereeitherhealthy(primaryprevention trials)orhadanydisease(secondarypreventiontrials). Datacollectionandanalysis Threeauthorsextracteddata.Trialswithadequaterandomisation,blinding,andfollow-upwereclassifiedashavingalowriskofbias. Random-effects and fixed-effect meta-analyses were performed. Random-effects meta-regression analyses were performed to assess sourcesofintertrialheterogeneity. Mainresults Sixty-sevenrandomisedtrialswith232,550participantswereincluded.Forty-seventrialsincluding180,938participantshadlowrisk of bias. Twenty-one trialsincluded 164,439 healthyparticipants. Forty-six trials included 68111 participants with various diseases (gastrointestinal, cardiovascular, neurological, ocular, dermatological, rheumatoid, renal, endocrinological, or unspecified). Overall, Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) 1 Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. theantioxidant supplementshadnosignificant effectonmortalityinarandom-effectsmeta-analysis(relativerisk[RR]1.02, 95% confidenceinterval[CI]0.99to1.06),butsignificantlyincreasedmortalityinafixed-effectmodel(RR1.04,95%CI1.02to1.06). In meta-regression analysis, the risk of bias and type of antioxidant supplement were the only significant predictors of intertrial heterogeneity.Inthetrialswithalowriskofbias, theantioxidant supplementssignificantly increasedmortality(RR1.05, 95%CI 1.02to1.08).Whenthedifferentantioxidantswereassessedseparately,analysesincludingtrialswithalowriskofbiasandexcluding seleniumtrialsfoundsignificantlyincreasedmortalitybyvitaminA(RR1.16,95%CI1.10to1.24),beta-carotene(RR1.07,95% CI1.02to1.11),andvitaminE(RR1.04,95%CI1.01to1.07),butnosignificantdetrimentaleffectofvitaminC(RR1.06,95% CI0.94to1.20).Low-biasrisktrialsonseleniumfoundnosignificanteffectonmortality(RR0.90,95%CI0.80to1.01). Authors’conclusions Wefoundnoevidencetosupportantioxidantsupplementsforprimaryorsecondaryprevention.VitaminA,beta-carotene,andvitamin Emay increase mortality.Future randomised trialscouldevaluatethepotential effectsof vitamin Candseleniumforprimary and secondary prevention. Such trials should be closely monitored for potential harmful effects. Antioxidant supplements need to be consideredmedicinalproductsandshouldundergosufficientevaluationbeforemarketing. PLAIN LANGUAGE SUMMARY Noevidencetosupportantioxidantsupplementstopreventmortalityinhealthypeopleorpatientswithvariousdiseases Previousresearchonanimalandphysiologicalmodelssuggestthatantioxidantsupplementshavebeneficialeffectsthatmayprolonglife. Someobservationalstudiesalsosuggestthatantioxidantsupplementsmayprolonglife,whereasotherobservationalstudiesdemonstrate neutralorharmfuleffects.Randomisedtrialshavelargelybeenneutral.Weneedevidencefromrandomisedtrialstodecideifantioxidant supplementsshouldbeusedforprevention. BACKGROUND key physiological processes. The results of our recent system- Oxidativestressmayplayroleinthepathogenesisofcancerand atic review and meta-analyses of the role of antioxidant sup- cardiovasculardisease,theleadingcausesofdeathinmiddle-and plementsforpreventionofgastrointestinal cancerswereunfore- high-incomecountries(Sies1985;Halliwell1999).Dietprovides seen(Bjelakovic2004).Wefoundthatantioxidant supplements numerousvitaminsandtraceelementsthatareessentialforgood significantly increased mortality in the antioxidant group with health.Severalobservationalstudieshaveshownasignificantpos- the fixed-effect model meta-analysis but not with the random- itiveassociationbetweenhigherintakeoffruitsandvegetablesand effects meta-analysis (Bjelakovic 2004). The effect of antioxi- reducedriskofchronicdiseases(Block1992;Ames1993;Willcox dant supplementson mortality hasalsobeen assessed in several 2004). However, exactly which specific dietary constituents of large trials on primary and secondary prevention of diseases ( fruitsandvegetablesmightbebeneficialisnotclear.Furthermore, HPS2002Low;ATBC2003Low;CARET2004Low;SUVIMAX causal inferences are hard to establish from observational stud- 2004Low;HOPETOO2005Low;WHS2005Low).Theresults ies.Antioxidantshaveattractedmostattentionaspromisingpre- oftheindividualtrialsareequivocal.Furthermore,noneofthetri- ventiveagents.Fruitsandvegetablesaresourcesofnumerousmi- alshadsufficientstatisticalpowertoidentifytheeffectofantioxi- cronutrientsandsomeofthese,includingbeta-carotene,vitamin dantsonmortality.Accordingly,weperformedasystematicreview A,vitaminC,vitaminE,andseleniumhaveantioxidantpotential. ofrandomisedtrialsonantioxidantsupplementsforprimaryand Manypeopletakeantioxidantsupplementsbelievingtoimprove secondaryprevention. theirhealth(Balluz2000;Millen2004;Radimer2004;Nichter 2006). OBJECTIVES Whetherantioxidantsupplementsarebeneficialorharmfulisun- certain(Herbert1997;Caraballoso2003;Vivekananthan2003; Ouraimwastoassesstheeffectofantioxidantsupplements(beta- Bjelakovic2004;Stanner2004;Miller2005;Berger2005).An- carotene,vitaminA,vitaminC,vitaminE,andselenium)onover- tioxidants may play dual roles, acting as double-edged swords allmortalityinprimaryorsecondarypreventionrandomisedclin- (Bjelakovic 2007b). Excessive antioxidants can adversely affect icaltrials. Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) 2 Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. METHODS ofrelevantpublicationsforadditionaltrials.Allsearchstrategies aregiveninAppendix1. Wesentlettersbypostore-mailtomajormanufacturersofan- Criteriaforconsideringstudiesforthisreview tioxidantsupplements,ie,CBHinChina,DSMinSwitzerland, CVC4healthinUSA,andBASFinGermany,askingforunpub- lishedrandomisedtrials.Noreplywasreceivedfromanyof the Typesofstudies contactedmanufacturers. Allprimaryandsecondarypreventionrandomisedclinicaltrials, irrespectiveoftrialdesign,blinding, publicationstatus,publica- tion year,or language. Fromcross-overtrials, onlythefirsttrial periodwasconsidered. Datacollectionandanalysis Thepresentreviewisbasedonourprotocolonantioxidantsup- Typesofparticipants plementsforpreventinggastrointestinalcancers(Bjelakovic2003) Adultparticipants(age18yearsorover)whowere adoptedtoassessoverallmortality.Anabbreviatedversionofthe reviewhaspreviouslybeenpublished(Bjelakovic2007a). • healthyparticipantsorwererecruitedamongthegeneral Inclusioncriteriaapplication population(primaryprevention); Two of thethreeauthors (GBand DN or RGS) independently • diagnosedwithaspecificdiseaseinastablephase(sec- assessedtrialeligibilitywithoutblindingofthestudyauthors.We ondaryprevention). listedexcludedtrialswiththereasonsforexclusion.Disagreement wasresolvedbydiscussion orinconsultation with LLGorCG. We excluded tertiary prevention trials, ie, randomised trials in Wecontactedauthorsofthetrialsformissinginformation. whichantioxidantsupplementswereusedtotreataspecificdisease Dataextraction or nutritional defects,liketrialswith patientswith acute, infec- Participantcharacteristics,diagnosis,andinterventions tious,ormalignantdiseases(exceptnon-melanomaskincancer). Fromeachtrialwerecordedfirstauthor;countryoforigin;country Weexcludedtrialsincludingchildrenandpregnantwomensince incomecategory(low,middle,high)(WorldBank2006);number theymaybeinneedofcertainantioxidantsupplements. ofparticipants;characteristicsofparticipants:agerange(meanor median)andsexratio;participationrate;dropoutrate;trialdesign Typesofinterventions (parallel,factorial,orcrossover);typeofantioxidant;dose;dura- Weconsideredforinclusiontrialsthatcomparedantioxidantsup- tion of supplementation; duration of follow-up (ie, duration of plements(ie,beta-carotene,vitaminA,vitaminC,vitaminE,and interventionpluspost-interventionfollow-up);andco-interven- selenium)atanydose,duration,androuteofadministrationver- tions. susplaceboornointervention. Theantioxidantscouldhavebeenadministered Trialcharacteristics • separatelyorinanycombinationamongthemselves;or We recorded the date, location, sponsor of the trial (known or • incombinationwithothervitamins;or unknownandtypeofsponsor)aswellaspublicationstatus. • incombinationwithtraceelementswithoutantioxidant Assessmentofmethodologicalquality Wedefinedthemethodologicalqualityastheconfidencethatthe function. design and reportrestrictbias in the intervention comparison ( Concomitantinterventionswereallowedifusedequallyinboth Schulz1995;Moher1998;Kjaergard2001).Due totheriskof interventionarmsofthetrial. overestimation of intervention effects in randomised trials with inadequate methodological quality (Schulz 1995; Moher 1998; Typesofoutcomemeasures Kjaergard2001),weassessedtheinfluenceofmethodologicalqual- ityofthefourcomponentsbelowasreportedinthetrials.When Oursoleoutcomemeasurewasall-causemortality. this information was not available, we asked the authors of the trialpublicationstoprovideit. Searchmethodsforidentificationofstudies Generationoftheallocationsequence We searched the Cochrane Central Register of Controlled Trials (CENTRAL)inTheCochraneLibrary(Issue3,2005),MEDLINE • Adequate,iftheallocationsequencewasgeneratedby (1966 to October 2005), EMBASE (Excerpta MedicaDatabase) acomputerorrandomnumbertable.Drawingoflots, (1985toOctober2005),andtheScienceCitationIndexExpanded tossing of a coin, shuffling of cards, or throwing dice (1945toOctober2005)(Royle2003).Wescannedbibliographies was considered as adequate if a person who was not Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) 3 Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. otherwise involved in the recruitment of participants Weperformedthemeta-analysesaccordingtotherecommenda- performedtheprocedure. tionsofTheCochraneHandbookforSystematicReviewsofInterven- • Unclear, if thetrial was described asrandomised, but tions(Higgins2006).Forthestatisticalanalyses,weusedRevMan themethodusedfortheallocationsequencegeneration Analyses(RevMan2003),STATA8.2(STATACorp,CollegeSta- wasnotdescribed. tion,Tex),SigmaStat3.0(SPSSInc,Chicago,Ill),andStats-Di- • Inadequate, ifasysteminvolvingdates,names,orad- rect(StatsDirectLtd,Altrincham,England). mittance numbers wereused for theallocation of pa- We analysed the data with both random-effects (DerSimonian tients. 1986)andfixed-effect(DeMets1987)modelmeta-analyses.We presentedtheresultsofrandom-effectsmodelanalyses.Whensta- Allocationconcealment tisticallysignificantresultsareobtainedineithertherandom-or • Adequate,iftheallocationofpatientsinvolvedacen- fixed-effectmodel,wepresentbothanalyses.Resultsarepresented astherelativerisk(RR)with95%confidenceintervals(CI).Weas- tral independentunit, on-site lockedcomputer,iden- sessedheterogeneitywithI2,whichdescribesthepercentageofto- ticallyappearingnumbereddrugbottlesorcontainers talvariationacrossstudiesduetoheterogeneityratherthanchance preparedbyanindependentpharmacistorinvestigator, (Higgins2002). orsealedenvelopes. • Unclear, if thetrial was described asrandomised, but Random-effectsmeta-regressionanalyseswereperformedtoassess potentialcovariatesthatcouldpredictintertrialheterogeneity,ie, themethodusedtoconcealtheallocationwasnotde- whichcovariatesthatwerestatisticallyassociatedwithestimated scribed. • Inadequate,iftheallocationsequencewasknowntothe interventioneffects.Theincludedcovariateswerebiasrisk(lowor high),typeanddoseofsupplement,singleorcombinedantiox- investigators whoassignedparticipantsorifthestudy idant experimental supplement regimen, duration of treatment, wasquasi-randomised. andtypeofprevention(primaryorsecondary).Wealsoperformed Blinding(ormasking) subgroupanalysescomparingtheprimaryandsecondarypreven- tiontrials.Furthermore,weperformedsensitivityanalysesexclud- • Adequate,ifthetrialwasdescribedasdoubleblindand ing trials using small dose antioxidant supplements in both the the method of blinding involved identical placebo or experimentalandcontrolstudygroups.Theexclusionoftrialsus- activedrugs. ingsmalldoseantioxidantsupplementsinboththeexperimental • Unclear,ifthetrialwasdescribedasdoubleblind,but andcontrolstudygroupswasbasedonthefactthatadditionof, themethodofblindingwasnotdescribed. eg,avitaminpillcouldbeaconfounder.Weobservedthatsele- • Notperformed,ifthetrialwasnotdoubleblind. niumseemedtohaveabeneficialeffectongastrointestinalcancer Follow-up development(Bjelakovic2004).Thesensitivityanalysisremoving seleniumtrialsfromouranalysistoevaluatetheirinfluenceonour • Adequate,ifthenumbersandreasonsfordropoutsand conclusionswasthereforenotaposthocdecision. withdrawals in allintervention groups weredescribed Theinfluenceoftrialswithzeroeventsinthetreatmentorcontrol or if it was specified that there were no dropouts or groupwasassessedbyre-calculatingtherandom-effectsmeta-anal- withdrawals. yseswith 0.5, 0.05, and0.005 continuity corrections (Sweeting • Unclear, if the report gave the impression that there 2004;Bradburn2007).Wealsoperformedadditionalmeta-anal- hadbeennodropoutsorwithdrawals,butthiswasnot ysesincludingonelargehypotheticaltrialwithoneeventinthe specificallystated. treatmentandcontrolgroupandasamplesizecorrespondingto • Inadequate,ifthenumberorreasonsfordropoutsand thetotalnumberofparticipantsinthezeroeventstrials. withdrawalswerenotdescribed. Allouranalysesfollowedtheintention-to-treatprinciple.Weac- counted all of the participants for each trial and performedthe Trialswith adequate generation oftheallocationsequence, ade- analysesirrespectiveofhowtheoriginaltrialistshadanalysedthe quate allocation concealment, adequate blinding, and adequate data.Participantslosttofollow-upwereconsideredsurvivors.For follow-upwereconsideredlow-biasrisktrials(highmethodologi- trialswithafactorialdesign,webasedourresultson’atthemargins’ calquality)(Kjaergard2001;Gluud2006a).Weappended’Low’ analysis, comparing allgroups thatreceivedantioxidant supple- tothereferencesofthesetrials.Trialswithoneormoreunclearor mentswithgroupsthatdidnotreceiveantioxidantsupplements inadequatequalitycomponentswereclassifiedashigh-biasrisktri- (McAlister 2003).Thisentailsarisk of interaction betweenthe als(lowmethodologicalquality)(Kjaergard2001;Gluud2006a). antioxidantandtheotherintervention(s)assessed,whethersignif- We also reported on whether the investigators had performed icantornotintheindividualtrial.Duetotheriskofconfounding a sample-size calculation and used intention-to-treat analysis ( infactorialtrialsassessingotherinterventions,weconductedpost- Gluud2001). hocsensitivity analysis including onlyfactorial trialdata, which Statisticalanalyses Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) 4 Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. couldnotbeaffectedofsuchconfounding (ie,’inside thetable’ analysis)(McAlister2003).Todeterminetheeffectofasinglean- tioxidantwealsoperformed’insidethetable’analysis(McAlister 2003)inwhichwecomparedthesingleantioxidantintervention with the placebo or no intervention. In the trials with parallel group design with more than two arms and additional therapy, we compared only antioxidant intervention with placebo or no intervention.Forcross-overtrialsweincludedonlydatafromthe firstperiod(Higgins2006). Comparison of intervention effects was conducted with test of interaction(Altman2003). Weperformedadjustedrankcorrelation(Begg1994)andregres- sionasymmetrytest(Egger1997)fordetectionofbias.AP<0.10 wasconsideredsignificant. RESULTS Descriptionofstudies See:Characteristicsofincludedstudies;Characteristicsofexcluded studies;Characteristicsofongoingstudies. Searchresults Databasesearchesyielded16111references(Figure1).Exclusion ofduplicatesandirrelevantreferencesleft1201referencesdescrib- ing815trials.Toobtainadditionalinformationwewrotetoau- thorsofabout500eligibletrials.Morethanonehundredauthors responded. Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) 5 Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. Figure1. FlowdiagramofidentificationofrandomisedtrialsforInclusion Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) 6 Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd. Weexcluded815referencesdealingwith748studies.Afterfur- ther evaluation we excluded 339 studies because they were not randomisedtrialsordidnotfulfilourinclusioncriteria.There- maining409wererandomisedtrialsexaminingantioxidantsup- plements. Four of these were still ongoing. The authors of the 405trialsdidnotreportdataonmortality(thesetrialsareshown athttp://ctu.rh.dk).ThemajorityoftheseweresmallphaseIor phaseIItrialswithshortdurationoffollow-upwithoutassessment ofclinicaloutcomemeasures.Wecontactedtheauthorsandabout onefifthofthemconfirmedthatmortalitywasindeedzero. Weincluded386referencesdescribing67randomisedtrialsful- fillingourinclusioncriteriaandabletoprovidedataforouranal- yses(Table1;Table2)(http://ctu.rh.dk).Thiscorresponds toa medianof6referencesperincludedtrial(range1to44references pertrial). Table1. Characteristicsofincludedtrialswithlowriskofbias Trial Design Num- Meanage Suppl Beta- Vitamin A VitaminC Vitamin E Selenium berpartic- (Follow- carotene (IU) (mg) (IU) (µg) ipants up)-y (mg) SCPS Parallel 1805 NA 5(5) 50 1990 Murphy Parallel 109 NA 0.003 200000 1992 (0.25) NIT2 Parallel 3318 54 6(6) 15 10000 180 60 50 1993 PPS1994 2x2 864 61 4(4) 25 1000 440 Pike1995 Parallel 47 69 1(1) 2666 90 45 AMDS Parallel 71 72 1.5(1.5) 12 750 200 50 1996 CHAOS Parallel 2002 62 1.4(1.4) 600 1996 NPCT Parallel 1312 63 4.5(7.4) 200 1996 PHS1996 2x2 22071 53 12(12.9) 25 Antioxidantsupplementsforpreventionofmortalityinhealthyparticipantsandpatientswithvariousdiseases(Review) 7 Copyright©2009TheCochraneCollaboration.PublishedbyJohnWiley&Sons,Ltd.