Authors: 

Rosenberg NE, Young AM, Zou Y, Justman J, Charurat ME, Moyo S, Zuma K, Bekker LG, Stranix-Chibanda L, Phiri SJ, Sam-Agudu NA, Yotebieng M, Hudgens MG, Chi BH, Shook-Sa BE.

Abstract: 

Background:

Globally, approximately half of new HIV acquisitions occur among African adults. This analysis examines which African men are at highest risk of acquiring HIV-1 and in greatest need of HIV pre-exposure prophylaxis (PrEP).

Setting:

National population-based surveys from Eswatini, Ivory Coast, Kenya, Lesotho, Malawi, Namibia, Nigeria, Rwanda, South Africa, Tanzania, Uganda, Zambia, Zimbabwe.

Methods:

The thirteen surveys were pooled and sampling weights were applied to represent all susceptible men aged 15-59 years-old. HIV-1 incidence was calculated based on a recent HIV-1 testing algorithm. A least absolute shrinkage and selection operator (Lasso) regression model was fit with 28 variables to predict recent HIV-1. Models were trained and internally cross-validated to estimate area under the receiver-operating characteristic (ROC) curve (AUC). Along the ROC curve, at sensitivity levels from 10%-90%, performance tradeoffs were evaluated.

Results:

Of 167,121 participants, 112 had recent HIV-1, representing 256,000 new annual infections among 122 million men. Only two variables were retained—reporting a male sexual partner and living in a subnational area where a high proportion of adults have detectable HIV-1 viremia. Overall AUC was 0.80 (95% Confidence Interval (CI): 0.71-0.89); cross-validated AUC was 0.76 (95% CI: 0.64-0.87). At 10% sensitivity, 25,000 cases could be averted if 357,000 men adhered to PrEP (Number Needed to Treat [NNT]=14). At 90% sensitivity, 229,000 cases could be averted if 50 million men adhered to PrEP (NNT=219).

Conclusion:

This predictive, parsimonious, generalizable risk assessment tool could help policymakers weigh tradeoffs between PrEP reach and efficiency.