Background:  Seropositioning and serosorting have been reported as strategies utilized by some men who have sex with men (MSM) to reduce HIV transmission risk while practicing unprotected anal sex. 

Methods:  The EXPLORE study was a randomized trial of an individualized behavioral HIV prevention intervention in HIV^– MSM in 6 U.S. cities. HIV serostatus and risk behavior ascertained using ACASI were assessed at 6-month intervals for as long as 48 months. Seropositioning was defined as the practice of insertive rather than receptive sex with positive or unknown vs negative partners, and serosorting as preferential use of condoms with positive/unknown partners. These preferences were summarized at both the participant and group level by odds ratios (OR). Data on >3000 men were used to characterize prevalence and predictors of seropositioning and serosorting and their effects on HIV seroconversion.  

Results:  The OR for seropositioning was ≥2 for 31% of participants and ≥3 for 23% of participants. Seropositioning was seen across all demographic categories, with OR ranging from 1.2 to 1.6; differences across study sites were also apparent. Seropositioning was more common in unprotected than protected anal sex (OR 1.43, 95%CI 1.39 to 1.46, p <0.001). However, we could exclude any substantial protective or adverse effect of seropositioning on risk of HIV seroconversion.  The OR for serosorting was ≥2 for 48% and ≥3 for 41% of participants. It was seen in all demographic subgroups, and was more pronounced in receptive than insertive sex (1.35, 95%CI 1.31 to 1.38). Site differences were again seen. We found a 12% decrease in the risk of seroconversion for each log increase in the OR for serosorting (p = 0.0005).

Conclusions:  High levels of seropositioning and serosorting were practiced by a sizable minority of MSM in all demographic categories. Our data provide no evidence that seropositioning has any substantive effect against HIV acquisition. In contrast, serosorting was associated with decreased risk of HIV infection, and was the more common strategy, possibly reflecting the intervention counseling emphasis on condom use. Because previous analyses in this and other datasets clearly point to increased risk associated with unprotected sex with multiple HIV^– partners, additional models will address whether the protective effects of serosorting can be overcome with larger numbers of partners.