Level 1: Likely reliable evidence

Three randomized trials in Africa (Uganda, Kenya, South Africa) evaluated the effect of male circumcision on HIV infection rates. In Uganda, 4,996 uncircumcised men aged 15-49 years, living in a rural area, were randomized to circumcision vs. no circumcision (control) and followed for 24 months (Lancet. 2007;369:657-666). The trial was stopped early after interim analysis found that circumcision was associated with an estimated 50% relative risk reduction in acquiring HIV (BMJ. 2007;334:11). There was a 90%-92% follow-up at all time points.

The two-year HIV incidence was 0.66 per 100 person-years with circumcision vs. 1.33 per 100 person-years with control (P=.006, NNT 150 person-years). Moderate-to-severe adverse events occurred in 3.6% of circumcised men, but all resolved with treatment.In another study in Kenya, 2,784 men aged 18-24 years were randomized to circumcision vs. no circumcision (control) and followed for 24 months (Lancet. 2007;369:643-656). The trial was stopped early after interim analysis found that circumcised men had about 50% relative risk reduction in acquiring HIV (BMJ. 2007;334:11). The median follow-up was 24 months; 91.4% had follow-up for HIV status. The two-year incidence of HIV was 2.1% with circumcision vs. 4.2% with control (P =.0065, NNT 48). Adverse events occurred at a rate of 1.5% but resolved quickly.

In South Africa, 3,274 uncircumcised men aged 18-24 years were randomized to circumcision vs. no circumcision andfollowed at three, 12, and 21 months (PLoS Med. 2005;2:e298). The trial was prematurely stopped following interim analysis; mean follow-up was 18 months. Among 3,128 men who were HIV-negative at baseline, HIV infections occurred in 20 circumcised men (1.3%, or incidence 0.8 per 100 person-years) vs. 49 uncircumcised men (3.1%, or incidence 2.1 per 100 person-years) (NNT 56). Results were similar after controlling for sexual behavior, condom use, and health-seeking behavior.