Pre-exposure prophylaxis (PrEP) is highly efficacious for prevention of HIV acquisition, but adherence to PrEP remains a major barrier. HPTN 082 is testing strategies to support PrEP adherence in young African women. A mathematical modelling approach is used to predict HIV incidence in the absence of PrEP among HPTN 082 participants, using the previously validated VOICE risk score and baseline sexual activity data from HPTN 082. This predicted incidence will provide a counterfactual to estimate PrEP effectiveness in this population.
The VOICE risk score (5-10, with 10 the highest risk score) is calculated for each woman based on baseline factors including age, marital status, financial stability, STIs, and alcohol usage. Using these data and self-reported sexual behavior not included in the risk score, we developed a Markov chain model of partnership formation, sexual behavior, and HIV transmission and used it to predict HIV incidence in the absence of PrEP. The model is calibrated using reported sexual activity, incidence data from the VOICE trial, and epidemiological data from the 2012 South African National HIV survey.
HPTN 082 enrolled 451 African women ages 16-25 with a median VOICE risk score of 7. 15% (68) reported anal sex in the last month, 30%(135) reported multiple partners in the last three months, and 49% (221) had a partner with unknown HIV status. Without PrEP, we predict an HIV incidence of 9.9% (95%CI 8.9-10.9), ranging from 6.3% (5.1-7.8) in women with a risk score of 5 to 21.5% (18.2-24.6) in women with a risk score of 10. Increased incidence at higher risk scores could be due to self-reported differences in sexual behavior. The remaining increase in incidence was attributed by the model to higher partner HIV prevalence. For example, women who did not live with their main partner were more likely to have multiple partners (OR=2.3, 95%CI 1.7-3.4) but also more likely to use a condom (OR=2.2, 95%CI 1.4-3.4). The model inferred a greatly increased HIV prevalence among their partners (OR=6.2, 95%CI 3.1-12.6).
HPTN 082 recruited a cohort of young African women who had multiple risk factors and would benefit from PrEP, given the predicted HIV incidence of 9.9%. These predictions will allow us to evaluate the effectiveness of PrEP stratified by VOICE risk score using an objective measure of adherence (tenofovir levels) from HPTN 082.