Background:  The potential effect of ART chemoprophylaxis on heterosexual spread of HIV-1 infection in resource-limited settings is uncertain.

Methods:  A complex mathematical model that includes age, gender, heterogeneity in sexual activity, variable infectiousness, and different sexual mixing patterns was used to simulate an HIV-1 epidemic in southern Sub-Saharan Africa and the potential effect of ART pre-exposure prophylaxis (PrEP). Epidemiological outcomes were determined under different scenarios of PrEP. Sensitivity analyses were performed to examine the effect of uncertainty in input parameters including the effectiveness of PrEP, level of coverage, PrEP discontinuation rate, HIV drug resistance and sexual disinhibition. Partial correlation between rank transformed input and output variables was performed to calculate partial rank correlation coefficients (PRCC).

Results:  Under the scenario of no sexual disinhibition occurring after PrEP introduction, the key parameters influencing the reduction in new HIV-1 infections after 5 years were the effectiveness of PrEP (PRCC = 0.94), PrEP discontinuation rate (PRCC = –0.94), level of risk-group coverage (PRCC = 0.92), and time to achieve target coverage (PRCC = –0.82). The risk of acquisition or transmission and persistence of drug resistance were not significant. In the scenario with sexual disinhibition, the effectiveness of PrEP and extent of sexual disinhibition had the greatest effect on prevention. An optimistic scenario of PrEP with 90% effectiveness and 75% coverage, predicted a 74% decline in cumulative new HIV-1 infections after 10 years with PrEP administered to the general population, and a 28.8% decline administered to groups with highest sexual activity (16% of the initial model population). Roughly 2.7 million to 3.2 million new HIV-1 infections could be averted in southern Sub-Saharan Africa using the latter strategy. Even with a 100% increase in at-risk behavior from sexual disinhibition, a beneficial effect was seen with 90% effective PrEP across a broad range of coverage (25 to 75%). However, with lower effectiveness of PrEP (≤50%), a rise in infections was observed with similar sexual disinhibition.

Conclusions:  PrEP is predicted by complex modeling to have significant public health benefit. This benefit can be lost, however, by sexual disinhibition of the population on PrEP or a high PrEP discontinuation rate, especially with marginal efficacy of PrEP (≤50% or less).