|
|
 |
| |
|
|
|
Uncovering Spatial Disparities and HIV Risk Factors Among Males in KwaZulu-Natal: A Bayesian Convolutional Approach |
|
|
|
PP: 817-837 |
|
|
|
doi:10.18576/jsap/140510
|
|
|
|
Author(s) |
|
|
|
Exaverio Chireshe,
Retius Chifurira,
Jesca Mercy Batidzirai,
Knowledge Chinhamu,
Ayesha B.M. Kharsany,
|
|
|
|
Abstract |
|
|
| Human immunodeficiency virus (HIV) remains a critical public health concern in South Africa, with KwaZulu-Natal (KZN) reporting the highest prevalence nationally. Despite the burden, men aged 15–49 remain underrepresented in HIV testing and treatment, exacerbating transmission risks. This study applies spatial statistical modelling to identify geographic heterogeneity and associated risk factors of HIV prevalence among this key demographic. Data were obtained from 3547 men participating in the HIV Incidence Provincial Surveillance System (HIPSS) in KZN between June 2014 and July 2015. A Bayesian convolution model was employed to capture spatially structured and unstructured random effects, using the integrated nested Laplace approximation (INLA) for efficient posterior inference. The results reveal significant spatial variation in HIV prevalence. Age emerged as a primary determinant, with the highest odds among men aged 40–44 (OR: 14.4645; 95% CI: 8.9035–23.5022). Lower educational attainment was associated with increased risk, whereas tertiary education was protective (OR: 0.6098; 95% CI: 0.3733–0.9866). Additional covariates positively associated with HIV infection included drug use, lack of circumcision, TB and STI diagnoses, mobility, and reported use of PrEP. A spatial cluster near Pietermaritzburg, with a 0.45 km radius, exhibited a 55.6% prevalence and a relative risk of 1.66, although this was not statistically significant (p = 0.213). In conclusion, the study highlights the utility of Bayesian spatial models in identifying localised risk patterns and informs recommendations for geographically stratified, male-specific interventions, including integrated service delivery for HIV, TB, and STIs within high-prevalence clusters. |
|
|
|
|
|
|
|
