oa HIV sero-discordancy can be predictive of HIV infectiousness in sub-Saharan Africa

Introduction: The majority of the population in reproductive age across sub-Saharan Africa is engaged in stable couples. At least half of the couples affected by HIV are sero-discordant (SDCs), that is couples where one of the partners is HIV infected while the other is not, with potential for HIV within-couple transmission. Countries of sub-Saharan Africa are characterized by varying levels of HIV sero-discordancy. HIV infectiousness also appears to vary across the continent. Understanding the link between the levels of HIV sero-discordancy and HIV infectiousness is important for understanding HIV epidemiology and informing HIV prevention efforts. Methods: We constructed a mathematical model that describes HIV transmission dynamics among a nationally-representative cohort of the population in reproductive age in 23 countries in sub-Saharan Africa. The purpose of the model is to understand analytically the link between HIV sero-discordancy and HIV infectiousness. HIV sero-discordancy was defined as the proportion of SDCs out of all stable couples affected by HIV (Pdiscord) in the population. The model was parameterized using nationally-representative empirical epidemiological and demographic measures derived from the Demographic and Health Survey data for each of the analyzed countries, along with state-of-the art empirical data for HIV natural history. Sensitivity analyses were also incorporated. Results: Our results revealed a strong functional dependence of HIV infectiousness on HIV sero-discordancy. The risk of HIV transmission among SDCs decreased following a non-linear pattern as Pdiscord increased. The non-linear relationship was characterized by a sharp decrease in HIV infectiousness as Pdiscord increased from zero up to nearly 50%. A transitional region followed where the dependence of HIV infectiousness on Pdiscord stabilized. Eventually, HIV infectiousness was slowly varying, with values of Pdiscord exceeding 50% not leading to major declines in infectiousness. Countries of sub-Saharan Africa clustered either in the transitional region around Pdiscord of 50% or in the asymptotic region well above Pdiscord of 50%. We did not identify any country in the highly non-linear region of Pdiscord well below 50%. Our sensitivity analyses with respect to other parameters in the model affirmed these findings. Conclusions: Our findings suggest that the observed levels of HIV sero-discordancy across the African continent can be predictive of HIV infectiousness among stable couples, and suggest that countries fall into two HIV infectiousness domains of high and low infectiousness. Mapping HIV sero-discordancy can inform HIV prevention programs on "hotspot" areas of potentially higher HIV infectiousness.