Mechanisms of change in an online randomized-controlled trial for HIV-negative and HIV-positive men who have sex with men
D. Smolenski, J.M. Wilkerson, B.R.S. Rosser
University of Minnesota, Epidemiology and Community Health, Minneapolis, United States
Background: Online HIV prevention interventions for men who have sex with men (MSM) show promising results. Still unknown, however, are the intermediate factors, or, mechanisms that contribute to behavioral change.
Methods: In 2006, 650 MSM were randomized to complete an online intervention, Sexpulse, or a standard questionnaire. After completing the randomized condition, participants completed a post-test survey that included a measure of intentions to practice safer sex. Follow-up occurred at 3-, 6-, 9-, and 12-months to monitor changes in the number of male unprotected anal intercourse partners (UAIMP). We measured intentions to practice safer sex using a three-item composite variable (7-point Likert, α=.87 intervention, α=.88 control). We fitted a structural equation model to isolate the direct effect of the intervention on UAIMP as well as the indirect effect through change in intentions to practice safer sex. A secondary model stratified participants on baseline HIV status to explore model moderation. We used a generalized linear model with a negative binomial link function to estimate the structural models.
Results: Overall, intervention participants reduced the number of UAIMP (IRR=0.69, 95% CI=0.54, 0.88) at three months. From the omnibus mediation model, change in intentions accounted for 73% of the change in UAIMP at three months. Among HIV negative participants, the IRR of UAIMP at three months was 0.69 (95% CI=0.51, 0.91); 80% of this effect was explained by change in intentions. Among HIV-positive participants, we did not detect a statistically-significant overall (IRR=0.79, 95%CI=0.52, 1.18) or mediated effect.
Conclusions: The observed behavior change at three months occurred, in part, because of changes in intentions. The different associations identified between HIV-negative and HIV-positive participants suggest a need for further tailoring of intervention content, and further study of mechanisms specific to each group.