Background Venues (eg, bars, the Internet) are believed to structure STI risk networks, however it is difficult to separate “risky places” from “risky people”, as venues are comprised of the people who attend them. To determine how venues may structure HIV/STI risk, we analysed data collected within social meeting places with “causal inference” statistical approaches, which aim to mimic randomising individuals to venues.

Methods Between 11/2007 and 3/2009, we visited 14 venues in San Diego, California, previously identified as places where men meet male sexual partners. 660 male patrons were interviewed anonymously by computer assisted self interview; 609 reported having ever had sex with a man. We analysed associations between the interview venue and participants' demographics, self-reported HIV and STI status, risk behaviours, and drug use. We used a marginal structural model to test associations between HIV/STI and venue, weighting by age, ethnicity, education, number of sexual partners in the past year, and number of HIV+ individuals known, in order to control for collinearity between venue and these patron characteristics.

Results The mean age of MSM was 34 yrs; 55% reported white, 24% Hispanic, and 8% African American ethnicity; 89% reported at least some college education. 35% reported history of STI and 13% reported that they were HIV-infected. The median number of lifetime male partners reported was 30, with 15% reporting unprotected receptive anal sex with at least five partners in the past 12 months. 39% reported ever meeting a sexual partner at the venue of interview. In univariate analyses, the 14 venues differed significantly by participants' reported age, ethnicity, number of lifetime male partners, past STI infection, HIV+ status, number of HIV+ individuals known, and finding partners at that venue. In multivariable analyses using marginal structural models, after controlling for participants characteristics, venues demonstrated significant structuring of HIV+ status, past STI, and methamphetamine (see Abstract P1-S2.58 figure 1).

Conclusions We demonstrated significant population structure in demographics, social networks, and HIV and STI status among MSM, which may assist in targeted surveillance and prevention. Interview venue may be associated with HIV, STIs and risk behaviours, separate from commonly collected individual-level data. In order to ensure successful in'terventions, a more thorough mechanistic understanding of how risky venues “emerge is needed.”