ABSTRACT
BACKGROUND: Recent studies have shown HIV incidence declines at a population level in several African countries. However, these studies have not directly quantified the extent to which incidence declines are attributable to different HIV programs. METHODS: We calibrated a mathematical model of the South African HIV epidemic to age- and sex-specific data from antenatal surveys, household surveys, and death registration, using a Bayesian approach. The model was also parameterized using data on self-reported condom use, voluntary medical male circumcision (VMMC), HIV testing, and antiretroviral treatment (ART). Model estimates of HIV incidence were compared against the incidence rates that would have been expected had each program not been implemented. RESULTS: The model estimated incidence in 15-49 year olds of 0.84% (95% CI: 0.75% to 0.96%) at the start of 2019. This represents a 62% reduction (95% CI: 55% to 66%) relative to 2000, a 47% reduction (95% CI: 42% to 51%) relative to 2010, and a 73% reduction (95% CI: 68% to 77%) relative to the incidence that would have been expected in 2019 in the absence of any interventions. The reduction in incidence in 2019 because of interventions was greatest for ART and condom promotion, with VMMC and behavior change after HIV testing having relatively modest impacts. HIV program impacts differed significantly by age and sex, with condoms and VMMC having greatest impact in youth, and overall incidence reductions being greater in men than in women. CONCLUSIONS: HIV incidence in South Africa has declined substantially since 2000, with ART and condom promotion contributing most significantly to this decline.
