Exploring HIV hotspots in Zambia and Zimbabwe: an analysis using the INLA-SPDE approach

Abstract

In 2021, 25.6 million people living with HIV/AIDS resided in Sub-Saharan Africa (1). HIV is still a relevant topic, which explains why 10 of the 17 SDGs include actions and targets which can aid the most affected populations. The literature shows that geographical factors facilitate transmission, and living in high prevalence communities significantly increases the probability of individuals to become HIV positive as well (2). The aim of this preliminary study is to investigate geographical patterns of the prevalence of HIV in Zambia and Zimbabwe, using data from the Demographic and Health Survey. Indeed, this survey provides both the HIV status of respondents – determined by ELISA-type tests – and the geolocalisation of its clusters (of households). The data were used in a Bayesian hierarchical model within the framework of INLA-SPDE (3). These models allow to model HIV prevalence at the cluster level – which is assumed to be continuous in an underlying Gaussian field and with a Beta distribution. One model was estimated for each country, and post- estimation results were used to investigate geographical patterns of the prevalence of HIV. These values are then represented in the maps, which show several hotspots along the borders particularly for Zambia and Zimbabwe. Relevant hotspots are found in both countries, especially along the borders. The results may highlight the effect of seasonal migration across borders on HIV diffusion. Moreover, using these methods could allow policy-makers to create programs that are measured to the people living in those areas, thus taking ethnicities, religious backgrounds, and cultural practices of their residents into consideration.