Abstract

Background: Pre-exposure prophylaxis (PrEP) with tenofovir and emtracitibine effectively reduces new HIV infections. Unfortunately, the optimal scenario for implementing PrEP so that most infections are prevented at the lowest cost is unknown. This is especially relevant in rural Africa where there are challenges regarding access to and availability of care. We used mathematical modeling to predict the impact of different PrEP strategies on averting new infections, prevalence, and cost in Macha, a rural setting in Southern Zambia. Methods: A deterministic mathematical model of HIV transmission was constructed based on the Macha epidemic. The antenatal prevalence in Macha is 7.7%, and 10-20% of the population is tested annually. Treatment is started at CD4<350. The population was stratified into four different groups that had different numbers of new sexual partners per year, obtained by calibrating the model to the Macha epidemic. PrEP has shown to reduce new HIV infections by 44-73%, depending on adherence. In our model we divided our population into four different groups that had different numbers of new sexual partners per year. The exact number of partners and exact proportion for each group was obtained by calibrating the model to the Macha epidemic through Monte Carlo filtering techniques. We then investigated the impact of targeting half of the most sexually active individuals for PrEP (10% of the total population), versus the impact of randomly putting 40-60% of the population on PrEP. PrEP would be considered cost-effective at a cost up to three times the gross national income per capita of Zambia, or $3210, per quality adjusted life year (QALY) gained. Results: Targeting PrEP to the two highest sexually active groups resulted in a similar number of infections averted as compared to a non-targeted strategy over ten years (21% and 27% reduction in new infections respectively), and a similar prevalence (6.4% and 6.2% respectively). However, the cost effectiveness of the targeted strategy was superior to the non-targeted strategy, more than 5 times the cost per QALY at $124 versus $642. Cost effectiveness was partially explained by the effectiveness of PrEP at $61 per QALY with 90% PrEP effectiveness and $218 per QALY with 20% effectiveness, and not strongly explained by the percentage of the high sexual risk groups that could be targeted. If only 25% of the targeted group goes on PrEP, the cost effectiveness still remains about $120. Conclusions: Targeting PrEP to individuals in high-risk groups in rural Africa is a cost effective strategy for reducing new infections. The more people that can be accurately identified as being in a higher sexual risk group, the greater the cost effectiveness will be.