Modeling The Association Between Lagged Climatic Factors And Malaria Incidence In Ghana

Abstract

Background: Significant advancements in malaria control have been made globally; however, the disease remains a major public health concern in impoverished tropical and subtropical regions where climate heavily influences transmission. This study applied a Generalized Linear Model (GLM) framework to examine lagged climatic effects on malaria incidence, testing model assumptions and addressing overdispersion. Methods: Monthly confirmed malaria case data (2020–2024) were obtained from the Ghana Health Service’s District Health Information Management System II (DHIMS-II). Climatic vari ables — rainfall, temperature, relative humidity, dew point, apparent temperature, sea level pressure, solar energy, solar radiation, and Ultraviolet (UV) index — were obtained from Visual Crossing Weather Data, and population data from the Ghana Statistical Service StatsBank. The outcome was monthly malaria incidence. Optimal lags (0–3 months) were identified via maximum correlation analysis, overdispersion was assessed using the Pearson Chi-square test, and Negative Binomial (NB2) regression was applied within a GLM framework including seasonal effects. Results: Severe overdispersion was detected (Pearson statistic = 9227.55), justifying NB2 over Poisson (Akaike Information Criterion (AIC): 20,877 vs. 8,228,441). This held under a population-adjusted specification with a log(population) offset (AIC: 20,537 vs. 7,058,695). Rainfall was consistently positively associated with incidence across all four lags (Incidence rate ratios (IRR) range: 1.0007–1.0015, all 𝑝 ≤ 0.0005), and solar energy at lag 0 was negatively associated (IRR = 0.963, 𝑝 = 0.013). Temperature, humidity, dew point, UV index, and seasonal indicators were significant only in the unadjusted model, suggesting these associations may partly reflect regional population differences rather than within-region climatic effects. A sustained 100 mm rainfall increase corresponded to a 16.2% incidence increase at a one-month lag alone, rising to approximately 55% when sustained across all four lags. Conclusion: Malaria transmission in Ghana appears to be associated with a wider range of climatic factors than previously recognized, with rainfall and solar energy showing the most consistent associations after accounting for regional population differences. The observed lagged effects suggest that climatic influences operate over different temporal scales, providing a foundation for further investigation and the development of context-specific malaria prevention and control strategies.