Background:  Although efficacy is unknown, some men who have sex with men (MSM) attempt to reduce their HIV risk by adapting their sexual practices to the perceived HIV serostatus of their partners. We used data from several cohorts to assess the association of seroadaptive practices with HIV seroconversion.

Methods:  Data were pooled from 4 studies (VPS, VAX004, Explore, and Step) with sexual behavior self-reports and HIV testing every 6 months for 18 to 36 months. Anal sex behavior reported at each visit was assigned to 1 of 6 hierarchical categories: 1) no unprotected anal sex (UAS; no anal sex or 100% condom use); 2) monogamy (UAS but only has 1 HIV^– partner); 3) top only (insertive for all anal sex); 4) serosorting (anal sex only with HIV^– partners); 5) seropositioning (receptive anal sex only with negative partners, always insertive with HIV⁺ or unknown partners), and 6) risky sex (any receptive anal sex with HIV⁺ or unknown serostatus partners). We used Cox models with baseline hazard stratified by study to evaluate the independent association of risk category with HIV seroconversion, controlling for number of partners, age, race, and treatment assignment.

Results:  A total of 14,485 HIV^– MSM contributed to 40,101 person-years of follow-up. Proportion of visits per category:  no UAS 48.5%; monogamy 9.4%; top only 10.1%; serosorting 6.9%; seropositioning 3.3%; and risky sex 21.9%. All categories had lower seroconversion rates than risky sex. Compared with no UAS, monogamy (HR 0.55; 95% confidence interval [CI] 0.32 to 0.95) and top only (HR 0.65; 95%CI 0.44 to 0.95) were less risky, while serosorting carried higher risk (HR 2.04; 95%CI 1.53 to 2.73). Seropositioning was similar in risk to UAS (HR 0.86; 95%CI 0.51 to 1.46). African Americans (HR 1.56; 95%CI 1.14 to 2.14) and Latinos (HR 1.29; 95%CI 1.02 to 1.63) were more likely to seroconvert, after accounting for behavioral category. No independent associations were found with age or number of partners.

Conclusions:  All seroadaptive practices considered in this study appear to be protective when compared to receptive anal sex with an HIV⁺ or unknown serostatus partner. MSM reporting monogamy and being top-only have the lowest risk of seroconversion, while those who practice seropositioning have a similar risk to those reporting no UAS. The increased risk associated with serosorting, compared to no UAS, likely results from mistaken perceptions of partner serostatus.