Background In KwaZulu-Natal, South Africa, young women face an extraordinarily high risk for HIV acquisition, with annual incidence estimates of 6%. ART-based strategies for HIV prevention have the potential to significantly decrease HIV incidence, but the impact of PrEP in addition to ART scale-up is undefined. Modeling studies suggest that PrEP targeted to highest-risk groups could maximise benefits and contain costs.

Methods We developed a deterministic transmission model of HIV that stratifies the population by age, sexual activity, and includes HIV infection stage. The model was parameterized using data from community-based HIV counselling and testing studies in KwaZulu-Natal and validated using independent HIV prevalence and incidence estimates.

We estimated the effectiveness and cost-effectiveness of a ‘test and treat’ scenario, targeted PrEP by age and sexual activity, and general PrEP provision. Each scenario was in addition to anticipated baseline ART scale-up for CD4≤ 350 from 35% in 2013, as observed in KwaZulu-Natal, to 60% in 2018 (following national guidelines). We assumed PrEP efficacy of 70%.

Results ‘Test and treat’ (ART for 80% of all HIV-positive persons) reduced HIV incidence by 58% and averted 25% of cumulative infections by 2025, at an additional $39,900 per infection averted compared to baseline ART scale-up. PrEP targeted to 60% of 20–29-year-olds, in addition to baseline ART scale-up, reduced incidence by 42% and averted 22% of infections at an additional $22,500 per infection averted, whereas PrEP targeted to 80% of high-risk individuals reduced incidence by 33% and averted 13% of infections at an additional $7,400 per infection averted. PrEP coverage of 20% of the general population reduced incidence by 37% and incident infections by 18%, at an additional $26,900 per infection averted.

Conclusion In a generalised HIV epidemic setting PrEP is a cost-effective addition to ART, with targeted PrEP being more cost-effective than generalised PrEP distribution.