The consequences of age-specific fertility rates and HIV subfertility on national HIV epidemic estimates from antenatal clinic prevalence

Jeff Eaton, Imperial College London
Milly Marston, London School of Hygiene and Tropical Medicine (LSHTM)

National HIV epidemic estimates are generated by fitting a mathematical model to HIV prevalence among pregnant women attending antenatal care. We created an age-stratified version of the UNAIDS EPP model that calculates prevalence among pregnant women accounting for age-specific fertility rates and the effect of HIV on fertility. We estimated the effect of HIV on fertility by duration of HIV infection using data from five population-based cohort studies in the ALPHA network. The model was fitted to HIV surveillance data from nine countries in eastern and southern Africa. Overall, adjusting for age-specific fertility rates and HIV subfertility resulted in similar estimates of current HIV prevalence, but reduced peak HIV incidence during the 1990s by between 4 and 15% and reduced AIDS deaths between 1985 and 2000 by 8%. Model fit and out-of-sample prediction for household-based prevalence surveys were modestly improved. Previous HIV epidemic estimates based on ANC survey data may overstate levels of HIV incidence and mortality.