Motivation for Substance Use, Venue of Sexual Encounter, and Sexual Risk Behavior among Men who Have Sex with Men (MSM) in the Multicenter AIDS Cohort Study (MACS): Factor Analysis, and Global vs. Event-Level Generalized Linear Mixed Model (GLMM) Assessment Approach

We use factor analysis to explore motivations for substance use before and/or during sexual activity in order to identify underlying structure in response types. The sample included 1º, 2º, and 3º motivation responses of 1,012 seropositive and 1,084 seronegative participants enrolled in the Multicenter AIDS Cohort Study (MACS) seen between 2006 and 2010. We found a single common factor for all alcohol and substance use combinations that combines the social and sexual domains. We also identified a second factor with four distinct loading patterns conditional on the alcohol or substance combination used. While there are strong sexual components for each, the underlying structure is complex, and also includes aspects of the social and personal domains. These underlying motivational structures are important in understanding why men engage in substance use in the context of sex, and may help identify men who participate in high-risk sexual behavior.We simultaneously modeled between-subject and within-subject variability using Generalized Mixed Linear Models (GLMMs) to explore the role of key "person variables" (HIV serostatus, sexual sensation seeking, and partner type) specific to the venue of sexual encounter in the association between substance use and sexual risk in the Multicenter AIDS Cohort Study (MACS). GLMMs were fit for each of three venues (Internet, bars, and bathhouses) using data from 1,012 seropositive and 1,084 seronegative participants seen between 2006 and 2010. We were able to show that venue-specific measurements of HIV serostatus, partner type, and sexual sensation seeking (SSS) are important in understanding the relationship between substance use and sexual risk, and may help explain the absence of consistent main effects seen in correlational or experimental studies. These person variables may aid the development of multivariate theoretical models that better fit substance use and sexual risk behavior associational data.We utilize the multiple measurement approaches available in the Multicenter AIDS Cohort Study (MACS) to build Generalized Linear Mixed Models (GLMMs) describing the association of substance use and sexual risk measured at the global level, and compare results to our previously published work using a venue-specific assessment approach. GLMMs were used to simultaneously model between-subject and within-subject variability in sexual risk behaviors (HIV serostatus, sexual sensation seeking, partner type, and venue of sexual encounter) among 1,012 seropositive and 1,084 seronegative participants seen between 2006 and 2010. All alcohol and drug use combinations were associated with having a higher numbers of unprotected anal intercourse (UAI) partners since last visit, regardless of venue of sexual encounter, when measured at the global level. These results reflect those from a venue-specific analyses of substance use and sexual risk conducted in the same cohort. While the global assessment approach used here does not permits causal interpretation of findings, we argue that establishing causality may not be a necessary condition for identifying the underlying person variables that confound the association between substance use and sexual risk, nor the utility of these variables in designing and implementing more tailored interventions