Abstract:
Despite making significant progress in tackling its HIV epidemic, South Africa, with
7.7 million people living with HIV, still has the biggest HIV epidemic in the world. The Government,
in collaboration with developmental partners and agencies, has been strengthening its responses
to the HIV epidemic to better target the delivery of HIV care, treatment strategies and prevention
services. Population-based household HIV surveys have, over time, contributed to the country’s
efforts in monitoring and understanding the magnitude and heterogeneity of the HIV epidemic.
Local-level monitoring of progress made against HIV and AIDS is increasingly needed for decision
making. Previous studies have provided evidence of substantial subnational variation in the HIV
epidemic. Using HIV prevalence data from the 2016 South African Demographic and Health Survey,
we compare three spatial smoothing models, namely, the intrinsically conditionally autoregressive
normal, Laplace and skew-t (ICAR-normal, ICAR-Laplace and ICAR-skew-t) in the estimation of
the HIV prevalence across 52 districts in South Africa. The parameters of the resulting models are
estimated using Bayesian approaches. The skewness parameter for the ICAR-skew-t model was not
statistically significant, suggesting the absence of skewness in the HIV prevalence data. Based on
the deviance information criterion (DIC) model selection, the ICAR-normal and ICAR-Laplace had
DIC values of 291.3 and 315, respectively, which were lower than that of the ICAR-skewed t (348.1).
However, based on the model adequacy criterion using the conditional predictive ordinates (CPO),
the ICAR-skew-t distribution had the lowest CPO value. Thus, the ICAR-skew-t was the best spatial
smoothing model for the estimation of HIV prevalence in our study.
