On Bayesian Generalized Linear Mixed Modeling of Cholera Risk Factors: A Case Study of Masvingo Province, Zimbabwe

Abstract

Cholera remains a significant public health threat in Masvingo, Zimbabwe, particularly in districts with inadequate water, sanitation and healthcare infrastructure. This study applies a Bayesian Generalised Linear Mixed Model (BGLMM) to analyse key risk factors associated with cholera incidence across seven districts (Masvingo, Chivi, Zaka, Bikita, Gutu, Mwenezi and Chiredzi) in Masvingo Province, Zimbabwe, during the 2023/ 2024 outbreak. Descriptive statistics show that cholera affects a young population with an average age of 24.6 years. The average distance to a health facility is 5.95km, indicating potential challenges in accessing healthcare. The frequency distributions reveal that 60.9% of the sampled population reported cholera cases. Significant fixed-effect predictors include gender, access to health facilities and seasonal risk, with the wet season posing a substantially higher risk (Estimate = 6.143, 95% CI: 3.649–9.114). Random effects suggest district-level variations, with Masvingo showing the highest risk deviation, followed by Chiredzi and Mwenezi. These findings were supported by a geospatial cholera case density heatmap, visualising concentrated high-risk zones (dark red) in the three districts. Chivi, Zaka, Bikita and Gutu appear as lighter spots, suggesting lower cholera intensity. Model fit statistics, Widely Applicable Information Criterion (WAIC) (110.115) and Leave-One-Out Cross Validation (LOO) (112.700), validate the model`s adequacy for evaluating cholera determinants. The study highlights the urgent need for integrated public health strategies, including infrastructure development, community health education and targeted interventions to curb cholera incidence. Such efforts will enhance resilience and reduce cholera incidence among vulnerable populations in Masvingo Province.