Drug Resistance in Sub- Saharan Africa: AConsensus AmongMathematical Models

is a powerful tool for analysis of prob-lems in infectious diseases. However, dif-ferences in key modeling assumptions may lead to disparate predictions [1]. Nevertheless, the principal outcomes of the Botswana model by Supervie et al [2] are qualitatively similar to those of our South Africa model [3]. Despite substan-tial differences in the 2 models ’ structural and parameter assumptions, the analyses we describe below confirm that, in both models, antiretroviral therapy (ART) plus preexposure prophylaxis (PrEP) will increase the prevalence of drug-resistant human immunodeficiency virus (HIV), with the major driver of resistance being ART, not PrEP. The discrepancies noted by Okano and Blower [4] in the predic-tions of the 2 models are primarily due to differences in the articulation of the results and in several optimistic assump-tions made by Supervie et al [2]. Supervie et al [2] use the phrase “prev-alence of transmitted resistance ” instead Figure 1. A, Human immunodeficiency virus (HIV) prevalence trends among men and women in the Botswana model. Data from Supervie et al are shown for comparison. B, Prevalence of HIV drug resistance in the Botswana model. C, Contribution of drug-resistant variants to the drug resistance preva-lence 20 years after preexposure prophylaxis (PrEP) rollout in the Botswana and South Africa models. D, Scatterplot of the percentage of infections prevent-ed versus the percentage reduction in cases with acquired resistance to antiretroviral therapy (ART) from uncertainty analysis of the South Africa model, the reparameterized Botswana model, and the original Botswana model. A random, representative selection of 1000 simulations from a corpus of 50 000 simulations is shown