Spatial-temporal heterogeneity and determinants of HIV prevalence in the Mano River Union countries

Abstract

Background: Utilizing population-based survey data in epidemiological research with a spatial perspective can integrate valuable context into the dynamics of HIV prevalence in West Africa. However, the situation in the Mano River Union (MRU) countries is largely unknown. This research aims to perform an ecological study to determine the HIV prevalence patterns in MRU.

Methods: We analyzed Demographic and Health Survey (DHS) and AIDS Indicator Survey (AIS) data on HIV prevalence in MRU from 2005 to 2020. We examined the country-specific, regional-specific and sex-specific ratios of respondents to profile the spatial-temporal heterogeneity of HIV prevalence and determine HIV hot spots. We employed Geodetector to measure the spatial stratified heterogeneity (SSH) of HIV prevalence for adult women and men. We assessed the comprehensive correct knowledge (CCK) about HIV/AIDS and HIV testing uptake by employing the Least Absolute Shrinkage and Selection Operator (LASSO) regression to predict which combinations of CCKs can scale up the ratio of HIV testing uptake with sex-specific needs.

Results: In our analysis, we leveraged data for 158,408 respondents from 11 surveys in the MRU. From 2005-2015, Cote d'Ivoire was the hot spot for HIV prevalence with a Gi_Bin score of 3, Z-Score 8.0-10.1 and P < 0.001. From 2016 to 2020, Guinea and Sierra Leone were hot spots for HIV prevalence with a Gi_Bin score of 2, Z-Score of 3.17 and P < 0.01. The SSH confirmed the significant differences in HIV prevalence at the national level strata, with a higher level for Cote d'Ivoire compared to other countries in both sexes with q-values of 0.61 and 0.40, respectively. Our LASSO model predicted different combinations of CCKs with sex-specific needs to improve HIV testing uptake.

Conclusions: The spatial distribution of HIV prevalence in the MRU is skewed and the CCK about HIV/AIDS and HIV testing uptake are far below the threshold target set by UNAIDS for ending the epidemic in the sub-region. Geodetector detected statistically significant SSH within and between countries in the MRU. Our LASSO model predicted that different emphases should be implemented when popularizing the CCK about HIV/AIDS for adult women and men.

Keywords: Africa; Comprehensive correct knowledge; Geodetector; Least Absolute Shrinkage and Selection Operator; Machine learning; Spatial distribution of HIV prevalence; Spatial stratified heterogeneity.