The Zimbabwean HIV epidemic is generalized, and heterogeneous at the district level. Combination HIV prevention (CHP) has been rolled out in Zimbabwe over the past decades, including antiretroviral therapy (ART), voluntary male medical circumcision (VMMC), prevention of mother to child transmission, behavior change programmes, and condom distribution. Evaluating the impact of these programs on the HIV epidemic is important to improve intervention planning.
Together with local policy makers and stakeholders, we developed a multidistrict, individual based HIV transmission model that simulates dynamic interactions between districts to accurately represent transmission dynamics, and quantified it using Zimbabwean demographic, epidemiological, and behavioral data. We used this model to evaluate the impact and cost-effectiveness of CHP in Zimbabwe over the period 2011 – 2015. This period was chosen as it encapsulates the national HIV strategic plan, and because the two large-scale population based surveys were conducted at the end of that period. We also estimate the future impact of alternative strategies.
We simulated the Zimbabwean HIV epidemic over 4 different districts, representative of rural, urban, mining, and commercial farming districts, and were able to reproduce district specific and national census data, sexual behavior in key and general populations, and HIV prevalence and incidence. We show that CHP in Zimbabwe over the period 2011 – 2015 prevented an estimated total of 90 thousand new infections, at 2259 US$ per infection averted (table). Interventions were most cost-effective in urban districts, and least cost-effective in rural districts. Importantly, our model closely reproduced national HIV incidence estimates in 2015 without specifically tuning to these data, serving as an important validation of our unique approach, and shows that we managed to closely reproduce the effects of CHP on incidence.
We have shown that CHP in 2011-2015 in Zimbabwe was highly cost-effective, even over the short period of implementation. Our approach in modeling a geospatially dynamic representation of the Zimbabwean HIV epidemic proved successful, and could be a valuable to further understand underlying transmission dynamics, and in turn optimize location specific resource allocation, allowing for the dynamic spillover effects of these interventions to other areas. Further expanding these tools could help policy makers in Zimbabwe and other countries to develop efficient and effective strategies to end AIDS by 2030.