Background:  Male circumcision reduces HIV acquisition in men. We conducted a trial of male circumcision in HIV⁺ men and assessed HIV transmission to female partners.

Methods: We randomized 1015 HIV⁺ men with CD4 ≥350 to immediate male circumcision (intervention) or male circumcision delayed for 24 months (control). Married men (n = 770) were asked to invite their spouses: 566 wives enrolled, of whom 245 (43%) were HIV^–. Participants provided written consent, were given information on HIV prevention, wound care and abstention from sex post-operatively, and were offered free condoms and couples counseling and testing. Men were seen at 1, 6 , 12, and 24 months; women at 6, 12, and 24; circumcised men were also seen post-operatively. The intent-to-treat analysis is based on 165 HIV-discordant couples (94 male circumcision arm, 71 control arm) in which both partners enrolled concurrently.

Results: Women's 24-month cumulative HIV incidence was 13.8/100 person-years in the male circumcision arm and 9.6/100 person-years in the control (p = 0.42). Women's incidence was highest in both arms in the 0- to 6-month follow-up interval (male circumcision arm 27.3/100 person-years; control 17.8/100 person-years, p = 0.41), declining at 6- to 24-months to 5.7/100 person-years and 4.1/100 person-years, respectively). In the male circumcision arm, the excess transmissions in months 0 to 6 occurred in couples who resumed intercourse >5 days prior to certified wound healing (5 of 18, 27.8% transmitted) compared to couples who delayed to within 5 days of certified healing or longer (6 of 63, 9.5% transmitted); the latter was equivalent to the 6 of 68 (8.8%) control arm transmissions in months 0 to 6. Rates of condom use, bacterial vaginosis, vaginal discharge, dysuria, and genitourinary disease (GUD) were comparable in female spouses in both arms at follow-up. Male circumcision reduced rates of GUD in circumcised vs uncircumcised HIV⁺ men (10.1% and 15.8%, respectively, PRR 0.64, CI 0.49 to 0.85, p= 0.002). The rate of male circumcision-related moderate adverse events in HIV⁺ men was equivalent to that of HIV^– men enrolled in a parallel trial (3.1% and 3.3%, respectively). There were no serious adverse events in HIV⁺ men. Of the HIV⁺ men, 71.0% had complete wound healing by 30 days post- circumcision, compared to 83.2% of HIV^– men (p <0.0001).

Conclusions: Male circumcision was safe and reduced GUD in HIV⁺ men. There were no direct HIV benefits to women but, potentially, an increased risk of transmission with early resumption of sex. This has programmatic implications:  HIV⁺ men might seek male circumcision as services roll out, necessitating appropriate education and follow-up.