G T p i R F M T M O c S e s t p i s R a C r T t h f n s F M S H M t P R R E T A w F 0 Research www.AJOG.org 4 ENERAL GYNECOLOGY he effects of male circumcision on female artners’ genital tract symptoms and vaginal nfections in a randomized trial in Rakai, Uganda onald H. Gray, MD, MSc; Godfrey Kigozi, MB ChB, MPH; David Serwadda, MB ChB, MMed, MPH; rederick Makumbi, PhD; Fred Nalugoda, MHS; Stephen Watya, MB ChB, MMed; Laurence Moulton, PhD; ichael Z. Chen, MSc; Nelson K. Sewankambo, MB ChB, MSc; Noah Kiwanuka, MB ChB, MPH; Victor Sempijja, BSc; om Lutalo, MSc; Joseph Kagayii, MB ChB, MPH; Fred Wabwire-Mangen, MB ChB, PhD; Renée Ridzon, MD; elanie Bacon, MPH; Maria J. Wawer, MD, MHS BJECTIVE: The objective of the study was to assess effects of male ircumcision on female genital symptoms and vaginal infections. TUDY DESIGN: Human immunodeficiency virus (HIV)-negative men nrolled in a trial were randomized to immediate or delayed circumci- ion (control arm). Genital symptoms, bacterial vaginosis (BV), and richomonas were assessed in HIV-negative wives of married partici- ants. Adjusted prevalence risk ratios (adjPRR) and 95% confidence ntervals (CIs) were assessed by multivariable log-binomial regres- ion, intent-to-treat analyses. ESULTS: A total of 783 wives of control and 825 wives of intervention 002-9378/$36.00 • © 2009 Mosby, Inc. All rights reserved. • doi: 10.1016 2.e1 American Journal of Obstetrics & Gynecology JANUARY 2009 n control (38.3%) than intervention arm spouses (30.5%, P � .001). t 1 year follow-up, intervention arm wives reported lower rates of enital ulceration (adjPRR, 0.78; 95% CI, 0.63-0.97), but there were o differences in vaginal discharge or dysuria. The risk of trichomonas as reduced in intervention arm wives (adjPRR, 0.52; 95% CI, 0.05- .98), as were the risks of any BV (adjPRR, 0.60; 95% CI, 0.38-0.94) nd severe BV (prevalence risk ratios, 0.39; 95% CI, 0.24-0.64). ONCLUSION: Male circumcision reduces the risk of ulceration, richomonas, and BV in female partners. ey words: bacterial vaginosis, female genital ulceration, male rm men were comparable at enrollment. BV at enrollment was higher circumcision, trichomonas, vaginal infections ite this article as: Gray RH, Kigozi G, Serwadda D, et al. The effects of male circumcision on female partners’ genital tract symptoms and vaginal infections in a andomized trial in Rakai, Uganda. Am J Obstet Gynecol 2009;200:42.e1-42.e7. hree randomized trials and multiple observational studies demonstrate hat male circumcision reduces the risk of uman immunodeficiency virus (HIV) in- ection in men,1-3 and World Health Orga- ization has recommended that circumci- ion be promoted for HIV prevention.4 However, the effects of male circumcision on male sexually transmitted infections (STIs) are more equivocal. In observa- tional studies5,6 and 2 randomized trials,1,7 circumcision was associated with reduced symptomatic genital ulcer disease (GUD) in men but had no effects on symptoms of urethral discharge or dysuria in male participants. If circumcision becomes widely adopted for HIV prevention in men, it is possible that there may be derivative benefits for female partners if the proce- dure reduces male carriage of HIV and rom the Departments of Population, Family, and Reproductive Health (Drs Gray and Wawer and Mr Chen) and International Health (Dr oulton), Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD; Rakai Health Sciences Program, Entebbe (Drs erwadda, Wabwire-Mangen, Makumbi, Kigozi, Nalugoda, Sewankambo, Kiwanuka, Sempijja, Lutalo, and Kagayii), the School of Public ealth (Drs Kiwanuka, Wabwire-Mangen, and Serwadda), the Department of Urology (Dr Watya), Mulago Hospital, and the Faculty of edicine (Dr Sewankambo), Makerere University, Kampala, Uganda; the Bill and Melinda Gates Foundation, Seattle, WA (Dr Ridzon); and he Division of AIDS (Ms Bacon), National Institutes of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD. resented at the 15th Conference on Retroviruses and Opportunistic Infections, Boston, MA, Feb. 3-6, 2008. eceived Dec. 12, 2007; revised June 5, 2008; accepted July 31, 2008. eprints: Ronald H. Gray, Robertson Professor of Reproductive Epidemiology, Johns Hopkins University, Bloomberg School of Public Health, 4132, 615 North Wolfe St, Baltimore, MD 21205. rgray@jhsph.edu. his study was supported in part by Grants (U01 AI11171-01-02) from the National Institutes of Allergy and Infectious Diseases (NIAID), Division of IDS, National Institutes of Health (NIH), and Grants 2228 and 220062228 from the Bill and Melinda Gates Foundation. The laboratory component as supported in part by the Division of Intramural Research, NIAID, NIH. This publication was supported, in part, by fellowships/grants from the ogarty International Center/NIH Grant 2 D 43 TW000010-19-AITRP and 5D43TW001508. i A g n w 0 a C t K /j.ajog.2008.07.069 mailto:rgray@jhsph.edu S t s o T m H c s e f a f U M T p H b 1 c s i t l 2 1 s c w l s t p u B c c p t c p t T o e w 1 r w l n l w n 1 A w c t c c f s t s t a a T p i s f b b s v r p a a 8 n w t H p 4 t T t t t n w s u i f d t p d d v g t s 4 a a G www.AJOG.org General Gynecology Research TIs or directly affects male-to-female ransmission. One observational study uggested that there may be a reduction f HIV, bacterial vaginosis (BV), and richomonas vaginalis infections in fe- ale partners of circumcised men.8 owever, 2 US studies observed no asso- iation between a man’s circumcision tatus and female BV.9,10 Therefore, we xamined data on vaginal infections rom female partners of men enrolled in randomized trial of male circumcision or HIV prevention, in Rakai District, ganda. ATERIALS AND METHODS he trial of male circumcision, sup- orted by the National Institutes of ealth has been described elsewhere.1 In rief, 4996 HIV-negative men aged 5-49 years who accepted voluntary ounseling and receipt of their HIV re- ults were enrolled and randomized to mmediate circumcision (the interven- ion arm, n � 2474) or circumcision de- ayed for 2 years (the control arm, n � 522). The trial was closed early on Dec. 2, 2006, because an interim analysis howed evidence of circumcision effi- acy for male HIV acquisition. At enrollment, male trial participants ere asked to identify their wives or FIGURE Trial Profile 4996 HIV-n men enroll 2474 Intervention 1167 currently married 1203 HIV-negative enrolled wives 799 followed at one year 1264 linked enrolled wives61 HIV-infected wives excluded 36 lost to follow up 835 women enrolled concurrently with husband ray. Effects of circumcision on female partners’ genital symp ong-term consensual partners. Con- w enting men who provided this informa- ion were then linked to their female artners who were enrolled and followed p in a separate study supported by the ill and Melinda Gates Foundation. Be- ause the male circumcision trial was losed, we felt it appropriate to assess the otential effects of male circumcision on he health of their female partners be- ause this could be of relevance to future olicy decisions with regard to promo- ion of male circumcision. The trial profile is given in the Figure. here were 1264 wives currently married r in long-term union with 1167 men nrolled in the intervention arm (1.08 ives per married man), and there were 239 female spouses of 1173 men en- olled in the control arm of the trial (1.06 ives per married man). The number of inked, enrolled women exceeded the umber of enrolled men because of po- ygamous relationships. At the time of female enrollment, 1203 ives of intervention arm men were HIV egative (95.2%); and in the control arm 171 were uninfected with HIV (94.5%). mong these HIV-negative married omen, 835 (69.4%) were enrolled con- urrently with their husbands in the in- ervention arm (ie, before the man’s cir- umcision surgery), and 803 (68.6%) ive 2522 Control 1173 currently married 239 linked to enrolled wives 764 followed at one year 1171 HIV-negative enrolled wives 68 HIV-infected wives excluded 39 lost to follow up 803 women enrolled concurrently with husband s and vaginal infections. Am J Obstet Gynecol 2009. ere enrolled concurrently with their o JANUARY 2009 Ameri ontrol arm husbands. These HIV-unin- ected women who were enrolled at the ame time as their husbands constitute he primary analysis sample for this tudy (Figure). A minority of women (368 in the in- ervention arm and 368 in the control rm) were enrolled 6 or more months fter their husband’s enrollment date. hese women were excluded from the rimary intent-to-treat analysis because f circumcision affected female vaginal ymptoms or infections, the baseline in- ormation for these women could have een biased by the interval of exposure etween their male partner’s circumci- ion and the woman’s initial enrollment isit. Among the 835 HIV-negative concur- ently enrolled HIV-negative female artners of HIV-uninfected intervention rm men, 799 (95.7%) were followed up t 1 year after enrollment, and among the 03 concurrently enrolled female part- ers of control arm, men 764 (95.1%) ere followed up at 1 year. At the time of he closure of the National Institutes of ealth male trial, 90% of men had com- leted 12 months of follow-up, but only 4% of male participants had the oppor- unity to complete 24 months follow-up. herefore, the female follow-up was runcated at 1 year. The married women were visited af- er their husbands had enrolled in the rial and were then followed up at an- ual intervals. At each study visit, omen were interviewed to ascertain ociodemographic characteristics, sex- al risk behaviors, and health status, ncluding symptoms of genital tract in- ections (GUD, vaginal discharge, and ysuria). Symptomatic women were reated syndromically. At each visit, women were asked to rovide a self-collected vaginal swab for iagnosis of BV and T vaginalis. BV was etected by Gram-stained slides from aginal swabs using the criteria of Nu- ent et al,11 a quantitative morphology hat has a standardized 0 to 10 point coring system, whereby 0-3 is normal, -6 is intermediate (transitional state), nd 7-10 is BV. Severe BV was defined as score of 9-10. Enrollment information egat ed 1 tom n BV was available for 825 of 835 con- can Journal of Obstetrics & Gynecology 42.e2 c a o w w a w w t s s f I D fi t a t 1 t t f w s p g i y o o l n i a m l p w n e c t w m t l B t ( ( l m g 2 n e s o s t f w l r c a y v s w 3 c t c t t w b f t d G b U m s m I H H m a i o v R T w o h p w S w t i b a a t e . l a f � r t ( p w m t T i v w ( n v w . o l t a t a c i m c o p T s s c o e f e Research General Gynecology www.AJOG.org 4 urrently enrolled wives of intervention rm men (98.8%) and 783 of 803 wives f control arm men (97.5%). Twelve month follow-up data on BV ere available for 785 intervention arm omen (98.7% of women followed up), nd 751 control arm wives (98.4%) of omen followed up. The number of omen with information on BV was less han the number of women seen because ome participants declined to provide elf-collected vaginal swabs. Trichomonas vaginalis was detected rom cultures of vaginal swabs using the nPouch TV culture method (BioMed iagnostics, San Jose, CA). Because of nancial constraints, trichomonas cul- ure was initiated only late in the study, nd there were limited data available at he time of female enrollment. At the 2 month follow-up visit, data on richomonas were available for 408 in- ervention arm wives (51.1% of those ollowed up) and for 402 control arm ives (52.7% of those followed up). As- ays for human papillomavirus and her- es simplex virus type 2 (HSV-2) are on- oing, so this report focuses on vaginal nfections with trichomonas and BV. We conducted an intent-to-treat anal- sis to assess the prevalence of symptoms f genital tract infections and diagnoses f BV and trichomonas during fol- ow-up in concurrently enrolled HIV- egative wives of male participants in the ntervention, compared with the control rm of the randomized trial. STI symptoms reported over the 6 onth period preceding the 1 year fol- ow-up visit were used to estimate the eriod prevalence of symptoms per 100 omen. Diagnoses of BV and trichomo- as at the 12 month visit were used to stimate the point prevalence of these onditions per 100 women. In addition, abulations of BV scores at follow-up ere stratified by BV scores at enroll- ent to determine rates of progression o BV from normal or intermediate base- ine flora scores, and among women with V at enrollment, we determined persis- ence of BV at follow-up. Unadjusted prevalence risk ratios PRRs) and 95% confidence intervals CIs) were estimated. Adjusted preva- ence risk ratios (adjPRRs) were esti- s 2.e3 American Journal of Obstetrics & Gynecolog ated using multiple log-binomial re- ression, with adjustment for age (15-19, 0-24, 25-29, and 30 years or older), mo- ogamous or polygamous marriage at nrollment, self-reported numbers of exual partners (1, 2, 3, or more), and use f condoms (none, consistent, or incon- istent) during the 1 year follow-up in- erval. Because the prevalence of BV dif- ered between study arms at enrollment, e also adjusted the follow-up preva- ence of BV for the presence of BV at en- ollment. The percent efficacy of circum- ision was estimated as follows: (1- djPRR) � 100%. We also conducted an as-treated anal- sis whereby men allocated to the inter- ention arm but who failed to return for urgery within 6 months of enrollment ere classified as uncircumcised (n � 6). Two control participants received ircumcision from other sources during he first year of follow-up and were cir- umcised crossovers. Enrollment of the HIV-negative male rial participants was funded by the Na- ional Institutes of Health, and the trial as registered (ClinicalTrials.Gov num- er NCT00425984). The enrollment and ollow-up of female partners of trial par- icipants was funded by the Gates Foun- ation and is registered (Clinical.Trials. ov number NCT00124878). The study was reviewed and approved y 2 institutional review boards (IRBs) in ganda (the Scientific and Ethics Com- ittee of the Uganda Virus Research In- titute and the Uganda National Com- ittee of Science and Technology) and 2 RBs in the United States (the Johns opkins Bloomberg School of Public ealth IRB and Western IRB). A com- unity advisory board provided guid- nce for the study and 2 data safety mon- toring boards, 1 for the NIH and the ther for the Gates-funded study, pro- ided oversight. ESULTS able 1 shows the characteristics of the omen at enrollment. Randomization f male trial participants produced a igh degree of comparability in female artner enrollment characteristics and exual behaviors between arms. There b y JANUARY 2009 as also comparability with respect to TI symptoms. Only a small minority of omen had enrollment cultures for richomonas because these tests were nitiated late in the study, so the overall aseline prevalence cannot be assessed. BV at enrollment was less frequent mong women married to intervention rm men (30.6%) than women married o control men (38.3%), and this differ- nce was statistically significant (P � 001). Severe BV (scores 9-10) were also ower in the intervention than control rm wives (1.8% vs 2.6%), but this dif- erence was not statistically significant (P .4). As shown in Table 2, the rates of self- eported GUD were significantly lower in he wives of intervention arm men 12.8%) than in women with control arm artners (16.8%, P � .03). However, there ere no differences in the frequency of fe- ale symptoms of discharge or dysuria by heir husband’s arm of randomization. he prevalence of trichomonas was signif- cantly lower among women with inter- ention arm husbands (5.9%), compared ith women with control arm husbands 11.2%, P � .01). BV prevalence was sig- ificantly lower among the wives of inter- ention arm men (40.3%), compared with ives of control arm men (50.6%, P � 00006). The proportion of women reporting 2 r more sex partners during the fol- ow-up interval was higher in the control han intervention arm women (5.6% nd 3.4%, respectively, P � .02), but here were no differences between study rms in self-reported condom use or onsistency of use during follow-up. Because BV prevalence differed signif- cantly between study arms at enroll- ent with a higher prevalence in the ontrol arm (Table 1), the differentials bserved at follow-up could reflect this reexisting differential at enrollment. herefore, we assessed BV at follow-up, tratified by enrollment vaginal flora cores. Our reasoning was that if circum- ision affected BV, the effects should be bserved among women without BV at nrollment, and this should not be af- ected by disparities in BV prevalence at nrollment. The results are shown in Ta- le 3. http://ClinicalTrials.Gov http://Clinical.Trials.Gov http://Clinical.Trials.Gov ( t l c 0 a s i t 0 e m w t 0 B I n s w S d s ( w l 9 A s t u t 0 t f a i p P o i c 9 9 t r 4 5 1 w w C r www.AJOG.org General Gynecology Research Among women with normal flora scores 0-3) at enrollment, progression o BV during follow-up was significantly ower in the wives of intervention than ontrol arm men (PRR, 0.80; 95% CI, .65-0.97; P � .005). Progression to BV mong women with intermediate flora cores (4-6) at enrollment was less in the ntervention than the control arm, but his was not statistically significant (PRR, .83; 95% CI, 0.65-1.06; P � .2). How- ver, in women who had BV at enroll- ent, persistent BV at 1 year follow-up as significantly lower in the interven- ion arm than control arm women (PRR, .83; 95% CI, 0.72-0.96; P � .02.) We also assessed women with severe V (scores 9-10) at follow-up (Table 3). n the control arm wives with initially ormal vaginal flora, 4.0% progressed to evere BV, whereas no intervention arm ives developed severe BV (P � .0002). imilarly, among women with interme- iate flora at enrollment, progression to evere BV was lower in the intervention 0.7%) than in the control arm (5.6%) ives, and this difference was of border- ine statistical significance (PRR, 0.13; 5% CI, 0.02-1.06; Fisher P � .03). mong women with BV at enrollment, evere BV was lower in the wives of in- ervention than control arm at follow- p, although the difference was not sta- istically significant (PRR, 0.61; 95% CI, .33-1.12). After adjustment for enrollment charac- eristics and sexual risk behaviors during ollow-up, the log binomial adjusted prev- lence risk ratio of GUD among wives of ntervention, compared with control arm articipants, was 0.78; 95% CI, 0.61-0.99; � .04, suggesting a circumcision efficacy f 22% (95% CI, 1-39%). For trichomonas nfection in the intervention relative to ontrol arm wives, the adjPRR was 0.55, 5% CI, 0.34-0.89; P � .02 (efficacy 45%; 5% CI, 11-66%). The adjusted risk of richomonas was increased among women eporting 2 (adjPRR, 2.02; 95% CI, 1.05- .33) and 3 or more sex partners (adjPRR, .12; 95% CI, 1.05-25.77), compared with partner. For BV at follow up, the adjPRR as 0.82 (95% CI, 0.74-0.91; P � .0003), ith an efficacy of 18% (95% CI, 9-26%.) ompared with women without BV at en- ollment, the adjusted risk of BV at fol- TABLE 1 Enrollment characteristics of HIV-negative women by their husband’s study arm Control Intervention n % n % All 803 100 835 100 .............................................................................................................................................................................................................................................. Age, y ..................................................................................................................................................................................................................................... 15-17 22 2.7 29 3.5 ..................................................................................................................................................................................................................................... 18-19 95 11.8 109 13.1 ..................................................................................................................................................................................................................................... 20-24 290 36.1 274 32.8 ..................................................................................................................................................................................................................................... 25-29 212 26.4 233 27.9 ..................................................................................................................................................................................................................................... 30-39 155 19.3 167 20.0 ..................................................................................................................................................................................................................................... 40-49 29 3.6 23 2.8 .............................................................................................................................................................................................................................................. Marital status ..................................................................................................................................................................................................................................... Monogamous 687 85.6 690 82.6 ..................................................................................................................................................................................................................................... Polygamous 116 14.4 143 17.1 .............................................................................................................................................................................................................................................. Religion ..................................................................................................................................................................................................................................... Catholic 494 61.5 516 61.8 ..................................................................................................................................................................................................................................... Protestant 218 27.1 230 27.5 ..................................................................................................................................................................................................................................... Saved 68 8.5 66 7.9 ..................................................................................................................................................................................................................................... Muslim 23 2.9 22 2.6 .............................................................................................................................................................................................................................................. Education ..................................................................................................................................................................................................................................... None 108 13.4 118 14.1 ..................................................................................................................................................................................................................................... Primary 581 72.4 604 72.3 ..................................................................................................................................................................................................................................... Secondary 98 12.2 96 11.5 ..................................................................................................................................................................................................................................... Tertiary 16 2.0 16 1.9 .............................................................................................................................................................................................................................................. Sex partners past year ..................................................................................................................................................................................................................................... 0 3 0.4 1 0.1 ..................................................................................................................................................................................................................................... 1 763 95.0 810 97.0 ..................................................................................................................................................................................................................................... 2 34 4.2 20 2.4 ..................................................................................................................................................................................................................................... 3� 3 0.4 4 0.5 .............................................................................................................................................................................................................................................. Condom use ..................................................................................................................................................................................................................................... Consistent 4 0.5 4 0.5 ..................................................................................................................................................................................................................................... Inconsistent 143 17.8 127 15.2 ..................................................................................................................................................................................................................................... No use 656 81.7 704 84.3 .............................................................................................................................................................................................................................................. Drank alcohol with sex ..................................................................................................................................................................................................................................... Never 543 67.6 565 67.7 ..................................................................................................................................................................................................................................... Sometimes 260 32.4 269 32.2 ..................................................................................................................................................................................................................................... Always 0 0.0 1 0.1 .............................................................................................................................................................................................................................................. STD past year ..................................................................................................................................................................................................................................... GUD 105 13.1 113 13.5 ..................................................................................................................................................................................................................................... Discharge 360 44.8 386 46.2 ..................................................................................................................................................................................................................................... Dysuria 164 20.4 160 19.2 .............................................................................................................................................................................................................................................. Gray. Effects of circumcision on female partners’ genital symptoms and vaginal infections. Am J Obstet Gynecol 2009. (continued) JANUARY 2009 American Journal of Obstetrics & Gynecology 42.e4 l i ( r c � a 0 w C T n f r t fi m e i t p w r 0 0 C a f c a w p w n m w f G k c s a s mpt Research General Gynecology www.AJOG.org 4 ow-up was also significantly increased f the woman had BV at enrollment adjPRR, 1.51; 95% CI, 1.35-1.568). The as-treated analyses yielded similar esults to the intent-to-treat analyses be- ause crossovers were few in number (n 36). For trichomonas, the as-treated djusted PRR was 0.55 (95% CI, 0.34- .88; P � 0.14), and for BV the adjPRR as 0.82 (95% CI, 0.74-0.92). TABLE 1 Enrollment characteristics of HIV-n women by their husband’s study ar Control n Vaginal infections .......................................................................................................... Trichomonas 1/1 .......................................................................................................... Any BV 300/783 .......................................................................................................... Severe BV 20/783 ................................................................................................................... a P � .001. ................................................................................................................... Gray. Effects of circumcision on female partners’ genital sy TABLE 2 Vaginal symptoms, trichomonas, a for HIV-negative women, by male p Outcomes at 1 y follow- up and sexual behaviors Control n/N Vaginal symptoms .......................................................................................................... Genital ulceration 128/763 .......................................................................................................... Discharge 323/763 .......................................................................................................... Dysuria 114/763 ................................................................................................................... Vaginal infections .......................................................................................................... Trichomonas 45/402 .......................................................................................................... BV 380/751 .......................................................................................................... Severe BV 49/751 ................................................................................................................... Sexual behaviors .......................................................................................................... Number of sex partners .......................................................................................................... 1 715/760 .......................................................................................................... � 2 45/760 ................................................................................................................... Condom use .......................................................................................................... None 607/763 .......................................................................................................... Inconsistent 148/763 .......................................................................................................... Consistent 8/763 ................................................................................................................... Gray. Effects of circumcision on female partners’ genital sympt 2.e5 American Journal of Obstetrics & Gynecolog OMMENT his trial of HIV uninfected female part- ers of HIV uninfected men found that emale partners of circumcised men had educed risks of GUD (efficacy 22%), richomonas (efficacy 45%), and BV (ef- cacy 18%). This strongly suggests that ale circumcision may have direct ben- fits for prevention of GUD and vaginal nfections in female partners. ative (continued) Intervention % n % .................................................................................................................. 100 0/1 0 .................................................................................................................. 38.3 252/825 30.5a .................................................................................................................. 2.6 15/825 1.8 .................................................................................................................. .................................................................................................................. oms and vaginal infections. Am J Obstet Gynecol 2009. BV during follow-up visits tner’s randomization arm Intervention % n/N % ......................................................................................................................... 16.8 102/798 12.8 ......................................................................................................................... 42.3 336/798 42.1 ......................................................................................................................... 14.9 113/798 14.2 ......................................................................................................................... ......................................................................................................................... 11.2 24/408 5.9 ......................................................................................................................... 50.6 316/785 40.3 ......................................................................................................................... 6.52 16/785 2.04 ......................................................................................................................... ......................................................................................................................... ......................................................................................................................... 94.1 786/795 96.6 ......................................................................................................................... 5.6 27/795 3.4 ......................................................................................................................... ......................................................................................................................... 79.6 655/798 82.1 ......................................................................................................................... 19.4 134/798 16.8 ......................................................................................................................... 1.1 9/798 1.1 ......................................................................................................................... oms and vaginal infections. Am J Obstet Gynecol 2009. y JANUARY 2009 The findings from this randomized rial are consistent with those from a rior observational study in Rakai, hich reported significantly reduced isks of GUD (PRR, 0.6; 95% CI, .4-1.0), BV (PRR, 0.79; 95% CI, 0.69- .91), and trichomonas (PRR, 0.55; 95% I, 0.55-0.77).8 However, our findings re contrary to 2 small US studies that ound no association between male cir- umcision and BV in female partners, but high proportion of men in these studies ere circumcised, and there was limited ower to detect an effect relative to women ith uncircumcised partners.9,10 We are ot aware of other studies examining fe- ale GUD or vaginal infections associated ith male circumcision. The mechanisms for the protective ef- ects of male circumcision on female UD and vaginal infections are un- nown. However, it is known that cir- umcised men are less likely to have ymptomatic genital ulcer disease,12 and metaanalysis suggested that circumci- ion is associated with reduced rates of Unadjusted PRR (intervention/control) 95% CI .................................................................................................................. 0.76 0.60-0.97 .................................................................................................................. 0.99 0.89-1.12 .................................................................................................................. 0.97 0.75-1.21 .................................................................................................................. .................................................................................................................. 0.53 0.33-0.85 .................................................................................................................. 0.80 0.71-0.89 .................................................................................................................. 0.31 0.18-0.54 .................................................................................................................. .................................................................................................................. .................................................................................................................. 1.03 1.00-1.05 .................................................................................................................. 0.57 0.36-0.91 .................................................................................................................. .................................................................................................................. 1.03 0.98-1.08 .................................................................................................................. 0.87 0.70-1.07 .................................................................................................................. 1.08 0.42-2.77 .................................................................................................................. eg m ......... ......... ......... ......... ......... nd ar ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... u l d m a s m s G w c p m w c t a t f h e r s d e i t t f r i B e s r g p p s a t p t i t c a m c n n f m d p c b s o s M p v t t c G t t b m t i f i t m t f c G c r o w t c o m f ( t f w i c i r t mpt www.AJOG.org General Gynecology Research lceration caused by HSV-2, Haemophi- us ducreyi, and syphilis.13 Therefore, re- uced male carriage of these pathogens ay reduce transmission of these ulcer- tive STIs to women. In addition, the ubpreputial space in uncircumcised en is moist,14 and this may enhance urvival of trichomonas and possibly the ram-negative anaerobes associated ith BV, so removal of the foreskin ould reduce female exposures to these athogens. We considered potential biases that ight affect these results. Our findings ith respect to the effects of male cir- umcision on BV are complicated by he fact that BV was more common mong the control than the interven- ion arm women at enrollment, so dif- erentials observed at follow-up could ave been due to preexisting differ- nces between women in the 2 arms ather than a direct effect of circumci- ion on BV per se. We do not know why randomization id not result in comparability of BV at nrollment, and we have been unable to dentify any factor, from randomization o the selection of the final analysis set, hat would have made a systematic dif- erence in the proportions with BV at en- ollment. However, circumcision signif- cantly reduced the risk of progression to V among women with normal flora at nrollment and reduced the risks of per- TABLE 3 Bacterial vaginosis at follow-up by Vaginal flora score at enrollment Control BV at follow-up/n at enrollment 0-3 124/325 ................................................................................................................... 4-6 67/124 ................................................................................................................... 7-10 185/285 ................................................................................................................... Severe BV ................................................................................................................... (9-10) ................................................................................................................... 0-3 13/325 ................................................................................................................... 4-6 7/124 ................................................................................................................... 7-10 29/285 ................................................................................................................... a Unconditional exact interval (StatXact 6.0). ................................................................................................................... Gray. Effects of circumcision on female partners’ genital sy istent BV among women with BV at en- e ollment (Table 3). This strongly sug- ests that male circumcision provides artial protection from BV in female artners. Although there were no differentials in exual risk behaviors reported by women t enrollment (Table 1), the wives of con- rol arm men did report more sexual artners during follow-up (Table 2), and his could have affected their risks of vag- nal infections. This apparent disinhibi- ion among wives of control arm men ould have occurred by chance or might rise if uncircumcised men experienced ore difficulties with intercourse (eg, aused by phimosis), which caused a mi- ority of their wives to seek other part- ers. However, we adjusted for these dif- erentials in behaviors between arms by ultivariate analyses, and there were no ifferentials in sexual risk behaviors re- orted by male trial participants.1 Thus, onfounding because of differential risk ehaviors is unlikely. The questions with regard to STI ymptoms were asked prior to questions n the woman’s partner’s circumcision tatus, so interviewer bias is also unlikely. oreover, if such bias occurred, it would robably have affected questions on all aginal symptoms, whereas the only pro- ective effects were observed with symp- omatic GUD but not with vaginal dis- harge or dysuria. A similar protective ginal flora score at enrollment Intervention BV at follow-up, % BV at follow-up/n at enrollment BV fol 38.2 122/402 30 ......................................................................................................................... 54.0 60/134 44 ......................................................................................................................... 64.9 130/240 54 ......................................................................................................................... Severe BV ......................................................................................................................... (9-10) ......................................................................................................................... 4.0 0/402 0 ......................................................................................................................... 5.6 1/134 0 ......................................................................................................................... 10.2 15/240 6 ......................................................................................................................... ......................................................................................................................... oms and vaginal infections. Am J Obstet Gynecol 2009. ffect of circumcision specifically against s JANUARY 2009 Ameri UD was also observed among men in he randomized trial.1 Laboratory bias in the diagnosis of richomonas or BV is extremely unlikely ecause technicians were blinded to the ale partner’s circumcision status. Re- ention rates were high and comparable n both study arms, so selective loss to ollow-up cannot explain the study find- ngs. Thus, we conclude that the protec- ive effects of male circumcision on fe- ale GUD and vaginal infections is likely o be a valid observation. These findings may have implications or future programs providing male cir- umcision for HIV prevention because UD and vaginal infections are potential ofactors for HIV acquisition,15-19 and eductions in these conditions because f circumcision may potentially protect omen from HIV infection. Observa- ional studies suggest that male circum- ision is associated with decreased risks f HIV in female partners.20,21 Thus, ale circumcision might protect women rom HIV risk by lowering infectivity eg, reduced male HIV shedding from he preputial mucosa), reducing HIV co- actors such as GUD in both men and omen, and reducing vaginal infections n women. In addition, because circum- ision prevents male HIV acquisition, it s also likely to have an indirect effect via educed female exposures to the virus, hus lowering secondary HIV transmis- PRR (intervention/ control) 95% CI-up, % 0.80 0.65-0.97 .................................................................................................................. 0.83 0.65-1.06 .................................................................................................................. 0.83 0.72-0.96 .................................................................................................................. .................................................................................................................. .................................................................................................................. 0.00 0.00-0.24a .................................................................................................................. 0.13 0.02-1.06 .................................................................................................................. 0.61 0.33-1.12 .................................................................................................................. .................................................................................................................. va at low .3 ......... ......... .8 ......... ......... .2 ......... ......... ......... ......... ......... ......... .0 ......... ......... .7 ......... ......... .3 ......... ......... ......... ......... ions to women. can Journal of Obstetrics & Gynecology 42.e6 p n t s m p A T p t e l t G b i o o C k B v F s o R 1 c U 3 2 c K L 3 T c f 1 P 4 P c p o s ( U p 5 a o L 6 c i 2 7 g P A S 8 c s R C I 9 m w S 1 b u 2 1 o a t 1 g u 2 1 R c a 8 1 A n r 1 e t h c D 1 B fl H 1 a h u D 1 H n L 1 c s P 2 M t i 2 2 s a i Research General Gynecology www.AJOG.org 4 We conclude that male circumcision revents genital ulceration, trichomo- as, and BV in female partners and that his benefit to women should be con- idered when planning scale-up of ale circumcision programs for HIV revention. f CKNOWLEDGMENTS he laboratory component was supported in art by the Division of Intramural Research, Na- ional Institutes of Allergy and Infectious Dis- ases, National Institutes of Health. We would ike to acknowledge the members of the Na- ional Institutes of Health and Bill and Melinda ates Foundation data safety monitoring oards, which monitored this trial, as well as the nstitutional review boards, which provided versight (the Scientific and Ethics Committee f the Uganda Virus Research Institute, the ommittee for Human Research at Johns Hop- ins, and the Western Institutional Review oard). We also are grateful for the advice pro- ided by the Rakai Community Advisory Board. inally, we wish to express our gratitude to tudy participants whose commitment and co- peration made the study possible. EFERENCES . Gray RH, Kigozi G, Serwadda D, et al. Male ircumcision for HIV prevention in men in Rakai, ganda: a randomized trial. Lancet 2007; 69:657-66. . Bailey RC, Moses S, Parker CB, et al. Male ircumcision for HIV prevention in young men in isumu, Kenya: a randomized controlled trial. ancet 2007;369:643-56. . Auvert B, Taljaard D, Lagarde E, Sobngwi- ambekou J, Sitta R, Puren A. Randomized, ontrolled intervention trial of male circumcision or reduction of HIV infection risk: the ANRS s 2.e7 American Journal of Obstetrics & Gynecolog 265 Trial PLoS Med 2005;2:e298. Erratum in: LoS Med 2006;3:e298. . World Health Organization/United Nations rogramme on HIV/AIDS. New Data on male ircumcision and HIV prevention: policy and rogramme implications. Conclusions and rec- mmendations. WHO/UNAIDS Technical Con- ultation, Montreux, March 6-8, 2007. Geneva Switzerland): World Health Organization/ nited Nations Programme on HIV/AIDS; 2007. . 1-19. . Reynolds SJ, Shepherd ME, Risbud AR, et l. Male circumcision and risk of HIV-1 and ther sexually transmitted infections in India. ancet 2004;363:1039-40. . Gray RH, Azire J, Serwadda D, et al. Male ircumcision and the risk of sexually transmitted nfections and HIV in Rakai, Uganda. AIDS 004;18:2428-30. . Bailey RC. Scaling up circumcision pro- rammes: The road from evidence to practice. rsented as the Plenary Talk, 4th International IDS Society Conference on HIV Pathogenesis, ydney, Australia, July 24, 2007. . Gray RH, Wawer M, Serwadda D, et al. Male ircumcision and the risks of female HIV and exually transmitted infections acquisition in akai, Uganda. Presented at the 13th CROI onference on Retroviruses and Opportunistic nfection, Denver CO, 2006. . Zenilman JM, Freisia A, Berger B, McCor- ack WM. Bacterial vaginosis is not associated ith circumcision of the current male partner. ex Transm Infect 1999;75:347-8. 0. Schwebke JR, Desmond R. Risk factors for acterial vaginosis in women at high risk for sex- ally transmitted diseases. Sex Transm Dis 005;32:654-8. 1. Nugent RP, Krohn MA, Hillier SL. Reliability f diagnosing bacterial vaginosis is improved by standardized method of gram stain interpre- ation. J Clin Microbiol 1991;29:297-301. 2. Thoma M, Gray RH. Male circumcision and enital ulcer disease: a meta-analysis. Pre- ented at the 17th International Society for Sex- 2 y JANUARY 2009 ally Transmitted Disease Research, Seattle, 007 (Abstract 542). 3. Weiss HA, Thomas SL, Munabi SK, Hayes J. Male circumcision and risk of syphilis, chan- roid, and genital herpes: a systematic review nd meta-analysis. Sex Transm Infect 2006; 2:101-9. 4. O’Farrell N, Morison L, Moodley P, et al. ssociation between HIV and subpreputial pe- ile wetness in uncircumcised men in South Af- ica. J AIDS 2006;43:69-77. 5. Serwadda D, Gray RH, Sewankambo NK, t al. Human immunodeficiency virus acquisi- ion associated with genital ulcer disease and erpes simplex virus type 2 infection: a nested ase-control study in Rakai, Uganda. J Infect is 2003;188:1492-7. 6. Taha TE, Hoover DR, Dalabetta GA, et al. acterial vaginosis and disturbances of vaginal ora: association with increased acquisition of IV. AIDS 1998;12:1699-706. 7. Martin HL, Richardson BA, Nyange PM, et l. Vaginal lactobacilli, microbial flora and risk of uman immunodeficiency virus type 1 and sex- ally transmitted disease acquisition. J Infect is 1999;180:1863-8. 8. Sewankambo N, Gray RH, Wawer MJ, et al. IV-1 infection associated with abnormal vagi- al flora morphology and bacterial vaginosis. ancet 1997; 350:546-50. 9. Gray RH, Kiwanuka N, Quinn TC, et al. Male ircumcision and HIV acquisition and transmis- ion: cohort studies in Rakai, Uganda. Rakai roject Team. AIDS 2000;14:2371-81. 0. Hunter DJ, Maggwa BN, Mati JK, Tukei PM, bugua S. Sexual behavior, sexually transmit- ed diseases, male circumcision and risk of HIV nfection among women in Nairobi, Kenya. AIDS 000;8:93-9. 1. Fonck K, Kidula N, Kirui P, et al. Pattern of exually transmitted diseases and risk factors mong women attending an STD referral clinic n Nairobi, Kenya. Sex Transm Dis 2000; 7:417-23. The effects of male circumcision on female partners' genital tract symptoms and vaginal infections in a randomized trial in Rakai, Uganda MATERIALS AND METHODS RESULTS COMMENT ACKNOWLEDGMENTS REFERENCES