Per-partner condom effectiveness against HIV for men who have sex with men.

OBJECTIVE: Few studies have examined condom effectiveness for HIV prevention among MSM. We estimated condom effectiveness per partner in four cohorts of MSM during 1993-2003 (JumpStart, Vaccine Preparedness Study, VAX004 and Project Explore).
METHODS: We used logistic regression to estimate the increase in odds of new HIV infection per HIV-positive partner for condom-protected receptive anal intercourse (PRAI; partners with whom condoms were always used) and condomless (unprotected) receptive anal intercourse (URAI; partners with whom condoms were sometimes or never used). To estimate condom effectiveness for preventing HIV transmission, we applied the concept of excess odds, the odds ratio minus 1. The condom failure rate was estimated as the excess odds per PRAI partner divided by the excess odds per URAI partner. Condom effectiveness was then 1 minus the failure rate.
RESULTS: The excess odds of HIV infection per HIV-positive partner were 83% for URAI and 7% for PRAI. The resulting failure rate (9%) indicated per-partner condom effectiveness of 91% (95% confidence interval 69-101).
CONCLUSION: The increase in odds of new HIV infection per HIV-positive partner for receptive anal intercourse was reduced by 91% for each partner with whom condoms were always used.