Elimination of HIV in South Africa through expanded access to antiretroviral therapy: Cautions, caveats and the importance of parsimony

TL;DR

This paper critically examines models predicting HIV elimination in South Africa through expanded ART access, highlighting the importance of model simplicity and cautioning against unwarranted assumptions in complex simulations.

Contribution

It compares different HIV transmission models, emphasizing the need for parsimony and scrutinizing assumptions in predicting HIV elimination timelines.

Findings

Simpler models suggest longer timelines for HIV elimination.

Complex models may rely on unsubstantiated assumptions.

Disagreement with claims that current ART scale-up will eliminate HIV in 30 years.

Abstract

In a recent article Hontelez and colleagues investigate the prospects for elimination of HIV in South Africa through expanded access to antiretroviral therapy (ART) using STDSIM, a micro-simulation model. One of the first published models to suggest that expanded access to ART could lead to the elimination of HIV, referred to by the authors as the Granich Model, was developed and implemented by the present author. The notion that expanded access to ART could lead to the end of the AIDS epidemic gave rise to considerable interest and debate and remains contentious. In considering this notion Hontelez et al. start by stripping down STDSIM to a simple model that is equivalent to the model developed by the present author3 but is a stochastic event driven model. Hontelez and colleagues then reintroduce levels of complexity to explore ways in which the model structure affects the results. In contrast to our earlier conclusions Hontelez and colleagues conclude that universal voluntary counselling and testing with immediate ART at 90% coverage should result in the elimination of HIV but would take three times longer than predicted by the model developed by the present author. Hontelez et al. suggest that the current scale-up of ART at CD4 cell counts less than 350 cells/microL will lead to elimination of HIV in 30 years. I disagree with both claims and believe that their more complex models rely on unwarranted and unsubstantiated assumptions.