Summary of key model parameters.

1In the Bernoulli model, we assumed HET and IDU females engaged in vaginal receptive sex, while HET and IDU males engaged in vaginal insertive sex. We also considered transmission via contaminated needle sharing for IDU. For MSM, we assumed 50% of their sex acts were insertive anal and 50% were receptive anal.2We assumed every individual in a particular transmission category had the same number of annual sex acts. The annual number of sex acts for HET was reported in the National Survey of Family Growth (NSFG) and the National Survey of Sexual Health and Behavior (NSSHB) for HET. We assumed IDU and MSM had the same annual number of sex acts as HET.3The effect sizes of risk reduction for HIV-infected and uninfected at-risk individuals were estimated by the percent reduction in unprotected sex acts (unprotected vaginal sex or anal sex). We included behavioral studies that reported the reduction in number (or percent) of unprotected sex acts. We took the median values of the reviewed studies in which the reported reduction in unprotected sex acts between intervention and control groups was statistically significant.4The proportion of protected sex acts among HIV-positive aware persons who do not receive risk reduction was calculated from the proportion of protected sex for unaware HIV positive persons and the reduction in unprotected sex for aware HIV-infected persons. That is, the proportion of protected sex acts for aware HIV-infected persons = 1-(1- proportion of protected sex acts for unaware HIV-positive persons)×(1- reduction in unprotected sex acts for aware HIV-infected persons).