Abstract
BACKGROUND: Millions of dollars have been spent in fighting malaria in Namibia. However, malaria remains a major public health concern in Namibia, mostly in Kavango West and East, Ohangwena and Zambezi region. The primary goal of this study was to fit a spatio-temporal model that profiles spatial variation in malaria risk areas and investigate possible associations between disease risk and environmental factors at the constituency level in highly risk northern regions of Namibia. METHODS: Malaria data, climatic data, and population data were merged and Global spatial autocorrelation statistics (Moran's I) was used to detect the spatial autocorrelation of malaria cases while malaria occurrence clusters were identified using local Moran statistics. A hierarchical Bayesian CAR model (Besag, York and Mollie's model "BYM") known to be the best model for modelling the spatial and temporal effects was then fitted to examine climatic factors that might explain spatial/temporal variation of malaria infection in Namibia. RESULTS: Average rainfall received on an annual basis and maximum temperature were found to have a significant spatial and temporal variation on malaria infection. Every mm increase in annual rainfall in a specific constituency in each year increases annual mean malaria cases by 0.6%, same to average maximum temperature. The posterior means of the time main effect (year t) showed a visible slightly increase in global trend from 2018 to 2020. CONCLUSION: The study discovered that the spatial temporal model with both random and fixed effects best fit the model, which demonstrated a strong spatial and temporal heterogeneity distribution of malaria cases (spatial pattern) with high risk in most of the Kavango West and East outskirt constituencies, posterior relative risk (RR: 1.57 to 1.78).