suggestedwww.pu811.com

RESEARCHARTICLEOpenAccessPerceivedcontrolovercondomuseamongsexworkersinMadagascar:acohortstudyAudreyPettifor1*,AbigailNorrisTurner1,TeresaSwezey1,MariaKhan1,MbolatianaSMRaharinivo2,BodoRandrianasolo2,AnaPenman-Aguilar3,KathleenVanDamme2,DeniseJJamieson3,FriedaBehets1,2AbstractBackground:Women'sperceivedcontrolovercondomusehasbeenfoundtobeanimportantdeterminantofactualcondomuseinsomestudies.
However,manyexistinganalysesusedcross-sectionaldataandlittlequantitativeinformationexiststocharacterizetherelationshipsbetweenperceivedcontrolandactualcondomuseamongsexworkerpopulations.
Methods:Weassessedtheassociationbetweenmeasuresofperceivedcondomusecontrolandself-reporteduseofmalecondomsemployingdatafromalongitudinalpilotstudyamong192sexworkersinMadagascar.
Results:Inmultivariablemodels,alackofperceivedcontrolovercondomusewithamainpartnerandhavingamainpartnereverrefusetouseacondomwhenaskedwerebothassociatedwithanincreasednumberofsexactsunprotectedbycondomsinthepastweekwithamainpartner(RR1.
86;95%CI1.
21-2.
85;RR1.
34;95%CI1.
03-1.
73,respectively).
Conversely,nomeasureofcondomusecontrolwassignificantlyassociatedwithcondomusewithclients.
Conclusion:Perceivedcontrolovercondomusewasanimportantdeterminantofcondomusewithmainpartners,butnotclients,amongsexworkersinMadagascar.
Programsworkingwithsexworkersshouldreachouttomainandcommercialpartnersofsexworkerstoincreasemalecondomuse.
BackgroundMalecondomsareoneofthefeweffectivemethodscur-rentlyavailableforthepreventionofHIV/STIinfection[1-3].
However,manywomenreporthavinglimitedcon-trolovertheuseofmalecondomswiththeirsexpart-ners[4-6].
Womenreportinglowerdecisionmakingpowerwithinrelationshipsmaybelesslikelytousecon-domsthanwomenwithgreaterpower[6,7].
Ameta-analysisofsocialpowerandnormativesupportoncon-domusefoundthatperceptionsofcondomusecontrolgenerallyhadstrongerassociationsamongmembersofsocietalgroupswithlesspower,includingfemales,youngerindividuals,ethnic-minoritiesandthosewithlesseducation[8].
Todate,thereislimitedquantitativeinformationavailableontherelationshipbetweenper-ceivedcontrolovercondomuseandactualcondomusepatternsamongsexworkersfromresource-constrainedcountries.
Further,thereislimitedinformationexaminingtheserelationshipsbypartnertype(forexam-ple,withmainpartnersvs.
clients)orlongitudinally.
Researchconductedamongsexworkersinresourcelim-itedsettingshasfoundthatindividualattitudesandbeliefshavelessinfluenceoncondomusethansocialandstructurallevelfactors[9-12].
ResearchamongsexworkersinTanzaniafoundthatwhensexworkersdecidedtousecondomsontheirownorjointlywithpartnersorclients,theyweresignificantlymorelikelytousecondomsthanwhenthepartnerorclientmadethedecision[13]AmongsexworkersinIndia,anumberofdifferenttypesofpowerwereassociatedwithconsistentcondomuse,includingcontroloverthetypeofsexwithclients,amountchargedandeconomicindependence[9].
Weaimedtoassesstheassociationbetweenvariousmeasuresofcondomusecontrolanduseofmalecon-doms.
Weuseddatafromalongitudinalpilotstudythatassessedtheacceptabilityandfeasibilityofconductingarandomizedtrialtoevaluatetheeffectivenessofthecer-vicaldiaphragmpluscandidatemicrobicideorplacebo*Correspondence:apettif@email.
unc.
edu1DepartmentofEpidemiology,UniversityofNorthCarolina,ChapelHill,USAPettiforetal.
BMCWomen'sHealth2010,10:4http://www.
biomedcentral.
com/1472-6874/10/42010Pettiforetal;licenseeBioMedCentralLtd.
ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(http://creativecommons.
org/licenses/by/2.
0),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.
gelinpreventinginfectionwithC.
trachomatis(CT)andN.
gonorrhoeae(GC)amongsexworkersinMadagascarObjectivesTodescribebaselineindicatorsofcondomusecontrolbytypeofsexpartnerandtoassesswhetherpoorcon-domusecontrolispredictiveofnotusingmalecon-domsamongfemalesexworkersinMadagascar.
MethodsWeuseddatafromafour-arm,partially-masked,rando-mized,prospectivepilotstudytoassesstheacceptabilityofevaluatingthediaphragmpluscandidatemicrobicideorplacebogelinpreventingchlamydialandgonococcalinfectionsamongsexworkersinfourcitiesinMadagas-car(Antananarivo,Antsiranana,Mahajanga,andToama-sina).
Detailsofthepilotstudyaredescribedindetailelsewhere[14].
Briefly,womenwererecruitedastheywereseekingcareforSTIsatpublicclinicsinthefourcommunitiesandthroughacommunity-basedoutreachprogramstaffedbypeercounselors.
Eligibilitycriteriaincludedhavingmorethanthreedifferentsexpartnersinthepastmonth,lessthan100%condomuseinthepasttwoweeks,ages15-55years,intendingtostayintheareaforthenextmonth,notbeingpregnantorplanningapregnancyinthenexttwomonths,noknownallergiestolatexandnophysicalabnormalitythatprecludeddiaphragmuse,andbeingabletogiveinformedconsent.
Althoughself-reportedsexworkwasnotaninclusioncriteriaforthestudy,whenaskediftheyhadeverhadsextoearnmoney,allwomeninthisstudyresponded"yes".
ThestudywasconductedbetweenSeptemberandNovember2005.
Atotalof192eligiblewomenwereenrolled.
Atenrollmentallwomenwereinterviewedusingaques-tionnairewhichcollectedinformationondemographic,reproductiveandsexualbehaviorcharacteristics.
Womenwereaskedtoreturnforfourweeklyfollow-upvisits.
Allwomengavewritteninformedconsentforscreen-ingandagainforenrollment.
TheinstitutionalreviewboardsattheUniversityofNorthCarolinaatChapelHillandattheCentersforDiseaseControlandPreven-tion,aswellastheComitéd'EthiqueauprèsduMinis-tèredelaSantéetduPlanningFamilialinMadagascarapprovedthestudy.
MeasuresAteachfollow-upvisit,womenreportedonthenumberofsexactsinthepastweekduringwhichmalecondomswerenotused.
Participantsansweredquestionsaboutsexualbehaviorsseparatelybypartnertype,andconse-quentlywealsoconductedouranalysesseparatelybypartnertype:1)husbandorboyfriendand2)otherpart-ners(notahusband/boyfriend).
Inthispaperwerefertohusband/boyfriendsas"mainpartners"andotherpartnersas"clients".
Ourmainoutcomeforeachanaly-siswasthenumberofsexactsnotprotectedbyamalecondomduringthepreviousweek.
Themainexposureswereperceivedcontrolovercon-domuse,andcondomusedynamicswiththemainpart-nerandwithclients.
Thequestionnaireincludedsixquestionsaboutcondomusecontrolthatwereaskedseparatelyformainpartnersandforclients.
Themea-sureswereusedasproxiesforgenderpowerwithmainpartnersandclients.
Thesemeasuresincluded:1)whethershehadsuggestedusingacondomatthelastsexact;2)whethershehadeverrefusedtohavesexbecausehewouldnotwearacondom;3)whetherhehadeverbecomeviolentwhenaskedtouseacondom;4)whetherhehadrefusedtouseacondom(everformainpartner,thelasttimetheyhadsexforclient);5)whetherhehadbecomeangryorarguedwhencondomusewassuggested(everformainpartner,thelasttimetheyhadsexforclient);6)herperceivedcontrolovercondomuse,withresponsesdichotomizedasalotorcompleteversusnoneorlittleAnalysisAllanalyseswereconductedusingSTATA(Version9.
0,CollegeStation,TX).
Weusednegativebinomialregressionmodelstoassesstheassociationbetweensixdifferentmeasuresofperceivedcondomusecontrolandthenumberofunprotectedsexactsinthepastweek.
Becausewecol-lectedoutcomedataonthesamecohortofwomenatfourweeklytimepoints,weusedgeneralizedestimatingequations(GEE)[15]toaccountforclusteringwithrobustvarianceestimatorsandanexchangeablecorrela-tionstructure.
Unadjustedandadjustedmodelsweredeveloped.
Formultivariablemodels,variableswereretainedinthefinalmodelbasedonamanualbackwardeliminationprocedure(10%changeineffectestimate)andbasedonapriorihypotheses.
ResultsThemeanageofwomeninthesamplewas30.
2years(standarddeviation(SD):7.
8years).
Mostwomenhadlimitededucation:themeannumberofyearsofschool-ingwas5.
7(SD:3.
0years).
Nearlyhalftheparticipants(45.
5%)reportedneverbeingmarried,37.
6%weredivorcedorseparated,10.
1%werecohabitingwithapartner,1.
6%weremarried,and5.
3%werewidowed.
Themedianweeklyincomeofwomenwas22,000Mala-gasyAriary(approximatelyUSD10.
40)[InterQuartileRange(IQR)15,000-40,000ma:USD$7.
10-$19.
00].
Womenwereaskedwhethertheirhomehadalistoftenitems(tapwaterinhouse/property,hotrunningwater,electricity,flushtoilet,cellphone,TV,refrigerator,microwave,electricstove,car)themediannumberofitemswas1(range0-6).
Pettiforetal.
BMCWomen'sHealth2010,10:4http://www.
biomedcentral.
com/1472-6874/10/4Page2of7MainpartnercharacteristicsatbaselineAtbaseline,lessthanhalf(41.
8%,n=79)ofallwomenreportedthattheycurrentlyhadamainpartner.
Ofthese,most(72.
2%)describedthepartnerasaboyfriend,22.
8%describedhimasahusbandandtheremainingpercentagedescribedhimasafriendor"other".
Ofthosewithmainpartners,themeanlengthoftherela-tionshipwas2.
6years(SD:2.
9years),andpartner'smeanagewas30.
8years(SD:8.
5years).
Themajorityofwomenwithmainpartnersreportedthatthepartnerwasolder:for43.
0%ofwomenthepartnerwas5ormoreyearsolder;for26.
6%,thepartnerwas1-4yearsolder;andfor30.
4%,thepartnerwasthesameageoryoungerthantheparticipant.
Womenreportedameanof2.
3sexacts(SD:1.
9acts)perweekwiththeirmainpartner.
Thevastmajority(70.
8%)ofwomenreportedthattheirmainpartnerprobablyordefinitelyhadsexwithotherwomeninthepastmonth.
Alcoholusebythemainpartnerduringthelastsexactwasreportedby13.
9%ofparticipants,whereas3.
8%ofwomenreportedthattheywereundertheinfluenceofalcoholatthelastsexactwiththeirmainpartner.
ClientcharacteristicsatbaselineAtbaseline,themediannumberofclientsreportedbywomeninthelastweekwas6(IQR4-9).
Morethanhalf(54.
2%)ofwomendescribedtheirlastclientas"non-regular"(someoneshehadsexwithonlyonceorafewtimes)ratherthan"regular"(someoneshehadsexwithonanon-goingbasis).
Mostwomen(68.
2%)reportedthattheymettheirlastpartneronthestreet,withfewerreportingthewoman'shouse(7.
8%),"privateplace",nightclub,andhotel(approximately4%foreach).
Twelvepercentreportedthattheirlastclientwasundertheinfluenceofalcoholwhentheyhadsexwhileonly2%reportedthattheywereundertheinfluenceofalcohol.
CondomuseatbaselineAtbaseline,condomusewasmorecommonwithclientsthanwithmainpartners(Table1).
Morethan40%ofparticipantsreportedneverusingacondomwiththeirmainpartner,comparedto0.
5%whoneverusedmalecondomswithclients.
Condomuseatthelastsexactwithmainpartnerswasreportedby26.
6%ofwomencomparedto61.
7%withclients.
PerceivedcontrolovercondomuseatbaselineAtbaseline,morewomenreportedthattheysuggestedcondomusetotheirclients(69%)thantotheirmainpartners(43%)thelasttimetheyhadsex.
Theyalsoreportedrefusingtohavesexmoreoftenifaclientrefusedtouseacondom,thanifamainpartnerrefused(Table1).
Justover20%ofwomenreportedthataclienthadeverbecomeviolentwhenaskedtouseacondom,comparedto9%whosemainpartnershadeverbeenviolentwhenaskedtouseacondom.
Morethanhalfofthewomen(52.
6%)reportedhavingnoorlittlecontrolovercondomusewiththeirmainpartnerscomparedto40.
2%withclients.
Sexactsunprotectedbycondomsoverstudyfollow-upSummedoverthefullfour-weekfollow-upperiodforallparticipantsinthecohort,thenumberofsexactswherenocondomwasusedwas49.
2%(271/551acts)withmainpartnersand35.
6%(4257/11,958acts)withclients.
AssociationsbetweenindicatorsofcondomusecontrolandactsunprotectedbycondomsMainpartnersInunadjustedanalyses,alackofperceivedcontrolovercondomusewasassociatedwithanincreasedlikelihoodofsexactsunprotectedbymalecondomswithamainpartnerduringthepreviousweek(Table2).
Thisasso-ciationremainedsignificantinmultivariablemodelsadjustingforage,education,maritalstatus,assets,studysite,andstudyarm(RateRatio(RR)1.
86;95%CI1.
21-2.
85)(Table2).
Sexunprotectedbymalecondomswasalsomorelikelyamongwomenwhoreportedthatamainpartnerhadeverrefusedtouseacondomwhenasked(RR1.
34;95%CI1.
03-1.
73).
Surprisingly,womenwhoreportedthatamainpartnerhadeverbecomeangryorarguedwhenaskedtouseacondomandwhoreportedthatamainpartnerhadbecomeviolentwhenaskedtouseacondomweremorelikelytousecon-doms,buttheseassociationswerenotstatisticallysignificant.
ClientsWithregardtounprotectedcondomactswithclients,inunadjustedmodels,womenwhoreportedlittleornocontrolovertheuseofcondomswithclientswereslightlymorelikelytoreportunprotectedsexacts(RR1.
07;95%CI1.
03-1.
11).
Womenwhoreportedneverrefusingsexwithclientswhowouldnotwearcondomswereslightlymorelikelytoreportactsunprotectedbycondomswithclients(RR1.
05;95%CI1.
02-1.
09).
Theseeffectsdidnotremainsignificantinmultivariablemodels,however(Table3).
Inaddition,womenreportedsignificantlyfewerunprotectedactswithnon-regularcli-entcomparedtoregularclients(datanotshown).
DiscussionWefoundimportantdifferencesinreportedcondomusebetweenwomen'smainpartnersandtheirclients.
Mea-suresofgender-basedpower,particularlyperceivedcon-trolovercondomuse,wereassociatedwithexperiencesofunprotectedsexamongmainpartners,whereastheeffectsweremoreattenuatedandnon-significantwithclients.
SexworkinMadagascarisheterogeneous,differenttypesandsocialcategoriesofsexworkhavebeenreported[16,17].
Inaddition,itisnotformallyinstitutio-nalizedthroughbrothelsorpimpsthusmostwomenworkforthemselves[16,17].
ThemajorityofwomeninPettiforetal.
BMCWomen'sHealth2010,10:4http://www.
biomedcentral.
com/1472-6874/10/4Page3of7thisstudyrecruitedclientsonthestreet,marketsorinbarsandoperateinpoorneighborhoods.
Importantly,theexperienceinthisstudyandfromotherresearchwithsexworkersinMadagascaristhatthedistinctionbetweenmainpartners/boyfriend,called"sipas"inMala-gasy,andclientsisoftenfluid[16].
Clientscanbecomesipasandvice-versawhichmakesthedifferentiationbetweenroleswithintheserelationshipsandthe"rules"regardingcondomuseblurred.
Asobservedinthisstudy,unprotectedsexwasmorecommonwithregularclientsthannon-regularclientswhichmaysupportthehypothesisthatsomeclientsbecomeregularpartnersovertimeleadingtomoreintimaterelationshipswherecondomusemaynolongerbewantedorexpected.
Lessthanhalfofwomeninthisstudyreportedhavingamainpartner.
Nevertheless,amongthesewomen,Table1MeasuresofcondomuseandperceivedcondomusecontrolamongwomenbypartnertypeatbaselineamongMalagasysexworkers,2005.
MainPartnerClient%(n=79)%(n=189)FrequencyofcondomuseNever44.
3(35)0.
5(1)Rarely12.
7(10)7.
9(15)Sometimes31.
7(25)43.
9(83)Almostalways10.
1(8)40.
7(77)Always1.
3(1)6.
9(13)Condomuseatlastsexact26.
6(21)61.
7(116)Womansuggestedcondomusewithpartneratlastsexact43.
0(34)69.
2(130)Womaneverrefusedtohavesexwithpartnerbecausehewouldnotuseacondom51.
4(37)79.
7(149)Partnereverbecameviolentwhenaskedtouseacondom9.
0(7)21.
2(40)PerceivedcontrolovercondomuseNone/little52.
6(41)40.
2(76)Alot/complete47.
4(37)59.
8(113)Partnerrefusedtouseacondomwhenasked*65.
8(52)32.
3(43)Partnerbecameangryorarguedwhencondomusesuggested*44.
3(35)13.
5(18)*exactwordingofthequestionandtimeperiodvariedbypartnertype–seetables2&3forexactwording.
Notethatformainpartnersthesequestionswere'ever'whileforclientsitwasthe'lastclient'**Totalsvaryslightlyduetomissingdata.
Table2UnadjustedandAdjustedRateRatiosand95%ConfidenceIntervalsforassociationsbetweenmeasuresofperceivedcondomusecontrolandthenumberofsexactsunprotectedbyamalecondomwithmainpartnersinthepastweekamongMalagasysexworkers,2005.
MeasureofperceivedcondomusecontrolUnadjustedRR(95%CI)AdjustedRR**(95%CI)MainpartnereverrefusedtouseacondomwhenaskedNo1.
01.
0Yes1.
51(1.
17-1.
95)1.
34(1.
03-1.
73)MainpartnereverbecameangryorarguedwhencondomusesuggestedNo1.
01.
0Yes0.
94(0.
74-1.
19)0.
81(0.
62-1.
05)AmountofcontrolovertheuseofcondomsduringsexwithmainpartnerAlotofcontrol1.
01.
0Littleornocontrol1.
41(1.
07-1.
87)1.
86(1.
21-2.
85)EverrefusedtohavesexwithmainpartnerbecausehewouldnotuseacondomYes1.
01.
0No1.
05(0.
83-1.
32)1.
03(0.
87-1.
22)MainpartnereverbecameviolentwhenaskedtouseacondomNo1.
01.
0Yes0.
84(0.
55-1.
30)0.
85(0.
55-1.
34)WomansuggestedcondomusewithmainpartneratlastsexNo1.
01.
0Yes1.
01(0.
80-1.
27)0.
99(0.
84-1.
16)**Eachperceivedcondomcontrolvariablewasinaseparatemodeladjustedforage,education,maritalstatus,assets,studysite,studyarmPettiforetal.
BMCWomen'sHealth2010,10:4http://www.
biomedcentral.
com/1472-6874/10/4Page4of7morethanhalfreportedneverorrarelyusingcondomswiththeirmainpartner(eventhoughthevastmajority(70.
8%)reportedthattheirpartnerhadothersexpart-ners).
Atbaseline,fewerwomenreportedsuggestingcondomusetotheirmainpartnerorrefusingtohavesexwithamainpartnerifhewouldnotuseacondomcomparedtowithclients.
Lackofcondomusewithreg-ularpartnershasbeenwelldocumentedamongsexworkersandisattributedtoadesirebothforgreaterintimacywithregularpartnersandasameanstodiffer-entiateworklifefrompersonalrelationships[11,18,19].
Notusingcondomswithmainpartnersmayalsorepre-sentalowerperceivedriskofdiseaseposedbymainpartnerships[18].
Limitedcondomusewithinmainpartnershipsiswelldocumentedinnon-sexworkerpopulationsaswell[20,21].
Asmentionedabove,wedidfindthatcondomusewithclientsvariedbyclienttype(regularornon-regular);womenreportedsignificantlyfewerunprotectedactswithnon-regularclientsthanwithregularclients.
Itshouldbenotedthoughthatwhiletheoverallproportionofactswherenocondomwasusedwasgreateramongmainpartnersthanwithclients,thetotalnumberofcondomlesssexactswassubstantiallygreateramongclientsbecauseofwomen'sgreaternumberofsexactswithclients.
Therefore,pre-ventionprogramsshouldavoidfocusingtoonarrowlyonsexactswithmainpartners.
Programsshouldaimtoincreasetheproportionofsexactscovered,whethertheseactsarewithmainpartnersorclients.
Wefoundthatwomenwhoperceivedthemselvesashavinglittlecontrolovercondomusewiththeirmainpartnersatbaselineweresignificantlymorelikelytohaveunprotectedsexwiththeirmainpartner;however,theassociationforclientswasweakerandnon-signifi-cant.
Itmaybethatperceptionsofcontrolaremorestronglyassociatedwithmainpartnerswhereindividualchoicemayplayagreaterroleindeterminingwhethercondomsareusedornot[18,22].
Perceptionsofindivi-dualcontrolmaynotexertasgreataninfluenceoncon-domusewithclientsaswithmainpartners,becausecondomusewithclientsmaybedictatedbyfactorsout-sideofthewoman'sindividualperceivedcontrol(forexample,price,clienttype,numberofotherwomencompetingforclients,venue,etc.
)[9,11,12].
ItisforthisreasonthatmuchoftherecentliteratureonriskfactorsforHIVamongsexworkerpopulationsandinterven-tionsaimingtoaddressthisriskhavefocusedontheimportanceofenvironmental-structuralfactors[9,11,23].
LimitationsThepresentanalysishasanumberoflimitations.
First,measuresofcondomusecontrolwereonlycapturedatbaselineandnotateachfollow-upvisit.
Tothedegreethatcondomusecontrolvariesbypartner(particularlyclients),itmaybethatperceptionsofcontroldonotalignwithactualbehaviorsduetochangesinpartnersovertime.
Thisislesslikelytobeaproblemwithana-lysesofbehaviorwithmainpartners,becausemostwomenwillhavethesamemainpartnerthroughouttheTable3UnadjustedandAdjustedRateRatiosand95%ConfidenceIntervalsforassociationsbetweenmeasuresofperceivedcondomusecontrolandthenumberofsexactsunprotectedbyamalecondomwithclientsduringthepastweekamongMalagasysexworkers,2005.
MeasureofperceivedcondomusecontrolUnadjustedRRAdjustedRR*ClienteverrefusedtouseacondomwhenaskedNo1.
01.
0Yes1.
09(0.
81-1.
46)0.
92(0.
73-1.
17)ClienteverbecameangryorarguedwhencondomusesuggestedNo1.
01.
0Yes0.
98(0.
92-1.
04)0.
99(0.
94-1.
03)AmountofcontrolovertheuseofcondomsduringsexwithclientsAlot1.
01.
0Alittle/nocontrol1.
07(1.
03-1.
11)1.
03(0.
99-1.
06)EverrefusedtohavesexwithclientbecausehewouldnotwearacondomYes1.
01.
0No1.
05(1.
02-1.
09)1.
03(1.
00-1.
07)ClienteverbecameviolentwhenaskedtouseacondomNo1.
01.
0Yes1.
00(.
96-1.
04)0.
98(0.
95-1.
02)WomansuggestedcondomusewithclientatlastsexNo1.
01.
0Yes0.
99(0.
95-1.
03)1.
00(0.
97-1.
03)*Eachperceivedcondomcontrolvariablewasinaseparatemodeladjustedforage,education,maritalstatus,assets,studysite,studyarmandtypeofclientPettiforetal.
BMCWomen'sHealth2010,10:4http://www.
biomedcentral.
com/1472-6874/10/4Page5of7short,four-weekfollow-upperiod.
Inaddition,timeframesforthecondomusecontrolquestionsvariedbetweenmainpartnersandclients.
Condomusecontrolquestionsforclientswereaskedaboutthelastsexactwhileformainpartnersthetimeframewas"ever".
Thiswasdonetoimproverecallforclientsaswomenmayhavehadmanyclientsandrememberingsomeoftheseeventsacrosspartnerswouldbemorecomplicatedthanfocusingonthelastclient.
Second,thispopulationwascomposedofwomenwhoreportedthattheydidnotconsistentlyusecondomsatbaseline.
Therefore,thegeneralizabilityofthefindingstowomenwhoareabletousecondomsconsistentlybutwerenoteligibleforthisstudyislimited.
However,thefindingsareveryrele-vanttopopulationssimilartoours-womenwhohavelittlecontrolovertheuseofbarriermethodswithpart-ners-forwhomalternativemethodsofpreventionaremostneeded.
Inaddition,themeasuresusedinthisstudyonlyexamineoneelementofgenderpowerandwerebynomeansacomprehensivemeasureofgenderpowerinthispopulation.
Last,sexualbehaviorandcon-domusedatawereself-reportedandmaysufferfromsocial-desirabilitybias.
ConclusionsWefoundthatperceivedcondomuseandamainpart-nersrefusaltousecondomswhenaskedwereimportantdeterminantsofreportedcondomusewithmainpart-nersamongsexworkersinMadagascar.
Importantly,wefoundthatmeasuresofcondomusecontrolweremorestronglyassociatedwithcondomusewithmainpartnersthanwithclients.
Thesefindingsunderscoretheimpor-tanceofmainpartnersindeterminingtheuseofmalecondoms.
Ongoingpreventioneffortswithsexworkersshouldofferprogramsthatteachwomencondomnego-tiationskillsandconsiderinterveningwithmalepart-nerstochangenormswithregardtocondomuse.
AcknowledgementsThisstudywasfundedbytheUnitedStatesCentersforDiseaseControlandPreventionthroughanInter-AgencyAgreementwiththeUnitedStatesAgencyforInternationalDevelopmentandCONRAD.
Authordetails1DepartmentofEpidemiology,UniversityofNorthCarolina,ChapelHill,USA.
2UNC-MAD,Antananarivo,Madagascar.
3DivisionofReproductiveHealth,CentersforDiseaseControlandPrevention,Atlanta,Georgia,USA.
Authors'contributionsAPwasinvolvedinthedesignofthestudyandintheanalysis,writeupandinterpretationofthestudyfindings.
ANTandMKwereinvolvedinthedesignofthestudyandinthewriteupandinterpretationofthestudyfindings.
MR,BRandKVDwereinvolvedinthedesignofthestudy,studyimplementationanddatacollectionandininterpretationofstudyfindings.
Allauthorsreadandapprovedthefinalmanuscript.
CompetinginterestsTheauthorsdeclarethattheyhavenocompetinginterests.
Received:26February2009Accepted:28January2010Published:28January2010References1.
WarnerL,StoneKM,MacalusoM,BuehlerJW,AustinHD:CondomuseandriskofgonorrheaandChlamydia:asystematicreviewofdesignandmeasurementfactorsassessedinepidemiologicstudies.
SexTransmDis2006,33:36-51.
2.
WaldA,LangenbergAG,KrantzE,DouglasJMJr,HandsfieldHH,etal:Therelationshipbetweencondomuseandherpessimplexvirusacquisition.
AnnInternMed2005,143:707-713.
3.
WellerS,DavisK:CondomeffectivenessinreducingheterosexualHIVtransmission.
CochraneDatabaseSystRev2002,CD003255.
4.
MacPhailC,CampbellC:'Ithinkcondomsaregoodbut,aai,Ihatethosethings':condomuseamongadolescentsandyoungpeopleinaSouthernAfricantownship.
SocialScience&Medicine2001,52:1613-1627.
5.
BlancAK:TheEffectofPowerinSexualRelationshipsonSexualandReproductiveHealth:AnExaminationoftheEvidence.
StudiesinFamilyPlanning2001,32:189-213.
6.
PulerwitzJ,AmaroH,DeJongW,GortmakerSL,RuddR:Relationshippower,condomuseandHIVriskamongwomenintheUSA.
AIDSCare2002,14:789-800.
7.
PettiforA,MeashamD,ReesH,PadianN:SexualpowerandHIVrisk,SouthAfrica.
EmergingInfectiousDiseases2004,10:1996-2004.
8.
AlbarracinD,KumkaleGT,JohnsonBT:Influencesofsocialpowerandnormativesupportoncondomusedecisions:aresearchsynthesis.
AIDSCare2004,16:700-723.
9.
BlankenshipKM,WestBS,KershawTS,BiradavoluMR:Power,communitymobilization,andcondomusepracticesamongfemalesexworkersinAndhraPradesh,India.
AIDS2008,22(Suppl5):S109-116.
10.
SneedC,MoriskyD:Applyingthetheoryofreasonedactiontocondomuseamongsexworkers.
SocialBehaviorandPersonality1998,26:317-328.
11.
KerriganD,EllenJM,MorenoL,RosarioS,KatzJ,etal:Environmental-structuralfactorssignificantlyassociatedwithconsistentcondomuseamongfemalesexworkersintheDominicanRepublic.
Aids2003,17:415-423.
12.
RekartML:Sex-workharmreduction.
Lancet2005,366:2123-2134.
13.
TassiopoulosK,KapigaS,SamN,AoTT,HughesM,etal:Acase-crossoveranalysisofpredictorsofcondomusebyfemalebarandhotelworkersinMoshi,Tanzania.
IntJEpidemiol2009,38:552-560.
14.
BehetsFM,TurnerAN,VanDammeK,RabenjaNL,RavelomananaN,etal:Vaginalmicrobicideanddiaphragmuseforsexuallytransmittedinfectionprevention:arandomizedacceptabilityandfeasibilitystudyamonghigh-riskwomeninMadagascar.
SexTransmDis2008,35:818-826.
15.
ZegerSL,LiangKY:Longitudinaldataanalysisfordiscreteandcontinuousoutcomes.
Biometrics1986,42:121-130.
16.
StoebenauK,HindinMJ,NathansonCA,RakotoarisonPG,RazafintsalamaV:".
.
.
Butthenhebecamemysipa":theimplicationsofrelationshipfluidityforcondomuseamongwomensexworkersinAntananarivo,Madagascar.
AmJPublicHealth2009,99:811-819.
17.
StoebenauK:Symboliccapitalandhealth:thecaseofwomen'ssexworkinAntananarivo,Madagascar.
SocSciMed2009,68:2045-2052.
18.
WarrD,PyettP:Difficultrelations:sexwork,loveandintimacy.
SociologyofHealth&Illness1999,21:290-309.
19.
MurrayL,MorenoL,RosarioS,EllenJ,SweatM,etal:TheroleofrelationshipintimacyinconsistentcondomuseamongfemalesexworkersandtheirregularpayingpartnersintheDominicanRepublic.
AIDSBehav2007,11:463-470.
20.
LagardeE,AuvertB,ChegeJ,SukwaT,GlynnJR,etal:CondomuseanditsassociationwithHIV/sexuallytransmitteddiseasesinfoururbancommunitiesofsub-SaharanAfrica.
Aids2001,4(15Suppl):S71-78.
21.
LescanoCM,VazquezEA,BrownLK,LitvinEB,PugatchD:Condomusewith"casual"and"main"partners:what'sinaname.
JAdolescHealth2006,39:e441-447.
22.
WaddellC:HIVandthesocialworldoffemalecommercialsexworkers.
MedicalAnthropologyQuarterly1996,10:75-82.
23.
ShahmaneshM,PatelV,MabeyD,CowanF:EffectivenessofinterventionsforthepreventionofHIVandothersexuallytransmittedinfectionsinfemalesexworkersinresourcepoorsetting:asystematicreview.
TropMedIntHealth2008,13:659-679.
Pettiforetal.
BMCWomen'sHealth2010,10:4http://www.
biomedcentral.
com/1472-6874/10/4Page6of7Pre-publicationhistoryThepre-publicationhistoryforthispapercanbeaccessedhere:http://www.
biomedcentral.
com/1472-6874/10/4/prepubdoi:10.
1186/1472-6874-10-4Citethisarticleas:Pettiforetal.
:PerceivedcontrolovercondomuseamongsexworkersinMadagascar:acohortstudy.
BMCWomen'sHealth201010:4.
SubmityournextmanuscripttoBioMedCentralandtakefulladvantageof:ConvenientonlinesubmissionThoroughpeerreviewNospaceconstraintsorcolorgurechargesImmediatepublicationonacceptanceInclusioninPubMed,CAS,ScopusandGoogleScholarResearchwhichisfreelyavailableforredistributionSubmityourmanuscriptatwww.
biomedcentral.
com/submitPettiforetal.
BMCWomen'sHealth2010,10:4http://www.
biomedcentral.
com/1472-6874/10/4Page7of7