Changes of sociobehavioural characteristics and HIV in 29 Sub-Saharan African countries, 2000-2018

IntroductionHIV epidemics and associated sociobehavioural characteristics vary widely across Sub-Saharan African (SSA) countries. In this study, we investigated how national changes on sociobehavioural characteristics may be associated with trajectories in HIV epidemics. MethodsWe analyzed aggregated data from 80 DHS surveys of 29 SSA countries between 2000 and 2018. We described each national survey by 46 pre-selected demographic, socio- economic, behavioural and HIV-related variables. Using Principal Component Analysis, we reduced the dimensionality of surveys and visualized them on a 2-dimensional space. We used consensus clustering to identify groups of countries with similar sociobehavioural characteristics over time, and identified the main drivers of change. We used k-means clustering with dynamic time warping to group countries based on the evolution of their national HIV effective contact rates since 1990 (computed based on UNAIDS estimates of yearly HIV incidence and prevalence). ResultsWe identified three groups of countries with similar sociobehavioural and HIV-related characteristics that persisted from 2000 to 2018. While countries within groups remained similar since 2000, differences between the groups increased over time. The most important characteristics that explained the change in the sociobehavioural space since 2000 were: ART coverage, HIV testing, increase in accepting attitudes towards People Living with HIV/AIDS and increasing knowledge about AIDS. We also identified two groups of countries with characteristic evolutions of HIV effective contact rates since 1990. The first group of countries was characterized by a rapid decrease of effective contact rate since the early 1990. The second group experienced a slow decrease of the effective contact rate in the early 1990, followed by a faster decrease after 1995. While the three clusters based on sociobehavioural characteristics were associated with different peak levels of HIV incidence, the two clusters based on the progression of the effective contact rate were associated with the timing when peak HIV incidence levels were reached. ConclusionsOur findings suggest that the initial conditions of nascent epidemics, likely determined by sociobehavioural characteristics, are crucial in determining the long-term progression of HIV epidemics. Our methods can help to predict the effects of behavioural change on the HIV epidemic and design targeted interventions to reach epidemic control.