Modelling geographical variation in the determinants of childhood malaria in Malawi using geographically weighted regression.
TL;DR
Geographically Weighted Regression was applied to data from the recent 2024 Malawi Demographic and Health Survey to estimate district-specific associations between childhood malaria and selected risk factors, namely mosquito net use, place of residence, household wealth, maternal age, and maternal education.
Abstract
Malaria remains a leading cause of morbidity and mortality among children in Malawi with transmission exhibiting marked geographic disparities. While standard regression models estimate the association between childhood malaria and risk factors across the entire study area, they assume spatial stationarity, thereby masking local-area variations in estimated effects. We applied Geographically Weighted Regression, a spatial non-stationarity modeling approach to data from the recent 2024 Malawi Demographic and Health Survey to estimate district-specific associations between childhood malaria and selected risk factors, namely mosquito net use, place of residence, household wealth, maternal age, and maternal education. The overall malaria prevalence among children aged five or under was 36.3%. By applying Geographically Weighted Regression, the study demonstrated substantial spatial heterogeneity in the associations between childhood malaria and the selected predictors across Malawi, revealing locally varying effects that would have been obscured under conventional global modelling approaches. These findings highlight the value of locally estimated relationships for understanding contextual differences in the predictors of childhood malaria and provide district-specific evidence to support geographically targeted intervention strategies and more efficient allocation of prevention and treatment resources within Malawi's National Malaria Control Programme.