Pre-diagnosis predictors of risk reduction after diagnosis among MSM receiving care in Tijuana

Background: The development of effective transmission risk reduction strategies among PLWHA requires greater understanding of behavioral determinants. Little is known about pre-diagnosis predictors of behavior change after diagnosis for any population. Methods: A secondary analysis was conducted using data collected in Tijuana for a study of determinants of late HIV testing. One hundred and eight men who have sex with men were classified as 0=no sexual risk behavior change or 1=reduction from moderate/high to no/low risk after diagnosis. Potential predictors were selected from pre-diagnosis data. Results: Over 60% decreased risk behavior from moderate/high risk to no/low risk; 38.9% remained at moderate/high risk. A logistic regression model significantly explained an estimated 28.4% (Nagelkerke R2) of the variance in reported sexual risk behavior reduction, x2(9, N=106)=24.93, p<0.01. One variable and two interaction terms were significant independent predictors of reported decreased sexual risk behavior post-diagnosis (p<0.05) after accounting for the other variables in the model: having a close relationship with a sexual partner pre-diagnosis (AOR 2.18 [95% CI .890-5.356] p=0.008), the interaction between peer risk behavior and having a close relationship with a sexual partner pre-diagnosis (AOR 0.85 [95% CI 0.850—0.852] p=.024); and the interaction between the number of years living in a high HIV prevalence environment and education level (AOR 1.03 [95% CI 1.029—1.030] p=.023). Age at HIV diagnosis, asymptomatic at diagnosis, and peer risk index were near-significant (p<0.10). Conclusions: Pre-existing social environmental factors appear important to determining risk reduction post-diagnosis. Additional studies are needed to understand these dynamics.