Background Empirical evidence suggests that HIV incidence rate within stable discordant sexual partnerships in sub-Saharan Africa (SSA) varies between 1.2 and 19.0 per 100 person-years. Estimating HIV incidence rate within stable discordant partnerships is critical for determining the contribution of HIV sero-conversions among these partnerships to total HIV incidence.

Methods We constructed a mathematical model based on competing-hazards formalism to estimate HIV incidence rate within stable discordant partnerships across 20 countries in SSA. We also used the model to analyse the patterns of HIV discordancy in SSA. The model was parameterised using Demographic Health Survey data and analyses were conducted at endemic equilibrium. Sensitivity analyses were performed to explore the dependence on the dynamical drivers of discordancy.

Results Our model fitted well the empirical epidemiological measures of HIV discordancy and yielded an estimate for HIV incidence rate among discordant partnerships of 11.2 per 100 person-years (95% CI 8.9 to 13.4 per 100 person-years). HIV incidence rate ranged between 5.5 and 17.2 per 100 person-years across the countries. We also identified HIV incidence rate within stable discordant partnerships and HIV incidence rate from sources external to the partnership (or equivalently HIV prevalence) as key determinants of the variability in discordancy measures across SSA.

Conclusions Our estimate for the HIV incidence rate among discordant partnerships agrees well and falls in the mid-range of empirical estimates for this measure. There is however considerable variability across the countries. Biological and behavioural factors including differences in transmission rates such as due to male circumcision, may have contributed to the variability in HIV incidence rates among discordant partnerships across SSA. More research is needed to elucidate the determinants of this variability in incidence rates.