P36.03
Background: In many African countries, women account for more than half of all new HIV-1 infections; however, not all women are at equal risk of acquiring HIV-1. A risk prediction tool that can identify women at highest risk for HIV-1 acquisition could improve prevention research efficiency and inform HIV-1 prevention activities in policy and clinical settings.
Methods: Using baseline data from VOICE (MTN-003), a randomized, double-blinded, placebo-controlled trial conducted in South Africa, Uganda and ZImbabwe that assessed safety and effectiveness of daily oral and vaginal chemoprophylaxis for HIV-1 prevention, we used standard methods for the development of clinical prediction rules to generate a risk scoring tool to predict HIV-1 acquisition over the course of one year. The predictive ability of the score was assessed by calculating area under the curve (AUC) and the score was internally validated using 10-fold cross-validation.
Results: Among 5,029 women enrolled in VOICE, 4,834 women had complete data for factors of interest and were included in the analysis; of these, 248 acquired HIV-1 within one year after enrollment (HIV incidence=6.05% [248/4,093 person-years]). The final risk score resulting from multivariable modeling included the following baseline factors: participant age, married/living with a partner, financial or material support from partner, partner has other partners, curable STI, HSV-2 status and alcohol use. The maximum possible score was 12; 36% of participants had a score>6 and accounted for 66% of HIV-1 infections. The AUC for the score was 0.72 and mean AUC from 10-fold cross validation was 0.70, indicating good predictive ability.
Conclusions: A discrete set of characteristics which can be easily assessed were highly predictive of HIV-1 acquisition over one year. External validation of the risk score is required to evaluate the tool's performance when applied to different populations of women at risk for HIV-1 infection in Africa.
