Climate Impact on Bacterial Meningitis Occurrence over West Africa

Abstract

Bacterial meningitis remains a significant public health concern in West Africa, where seasonal outbreaks are closely linked to environmental factors such as low humidity, high temperatures and exposure to Saharan dust. While the influence of local climatic conditions on meningitis is well documented, the role of large-scale climate variability, particularly that of the El Niño–Southern Oscillation (ENSO), influencing global climates through complex atmospheric and oceanic mechanisms is still not fully understood. These mechanisms may affect Saharan dust activities and local climate conditions which are associated with meningitis outbreaks during the dry season. This study provides a comprehensive assessment of the relationship between climate and meningitis in West Africa by exploring the interactions between ENSO variability, Saharan dust dynamics and bacterial meningitis incidence in two climatically distinct regions: the Sahel (SAH) and the Gulf of Guinea (GG). This thesis addressed four specific objectives such as: 1) to characterise the seasonal meteorological and dust-related conditions that shape the meningitis cycle; 2) to examine the interannual and spatial variability of these environmental drivers; 3) to identify the large-scale atmospheric and oceanic mechanisms influencing dust variability; and 4) to develop machine learning models for estimating meningitis incidence, using environmental variables and vaccination coverage as inputs. Weekly environmental and meningitis cases data from 2006 to 2020 were analyzed at seasonal and interannual scale and predict meningitis incidence. At large-scale, monthly Saharan dust, Sea Surface temperature, Sea level pressure and winds components from 1980 to 2020 were employed. Empirical orthogonal function (EOF), lag Spearman correlation, regression map and composite analyses were applied to reveal a local and large-scale relationship between Saharan dust patterns influencing meningitis dynamics and its ENSO impact. At seasonal and interannual scale, high concentration of particulate matter (PM10) and aerosol optical depth (AOD) in January reliably precede meningitis outbreaks. The persistence and intensity of the leading mode of dust variability (70.5% in the SAH with peaks in March–April and 70.8% in the GG during January–March) associated with relative humidity (RH) below 20% in the SAH and 45% in the GG, increase the number of cases. At large-scale, SST anomalies in the equatorial Pacific (warm/cool) coincide with elevated dust variability (AOD-PC1) during the JFM–MAM in the GG and FMA–MAM over the SAH. Models like XGBoost in Nigeria (R² = 0.638), CatBoost in Burkina Faso (R² = 0.619), Gradient Boosting in Niger (R² = 0.487) and Random Forest in Mali (R² = 0.355) identified as the best. However,vaccination status, RH and meridional wind consistently emerged as the most influential predictors across all countries. These findings advance our understanding of how large-scale climate variability and dust dynamics influence meningitis outbreaks in West Africa. They also demonstrate the utility of machine learning (ML) approaches for forecasting disease risk and support the integration of climate-informed early warning systems, particularly in settings where data is scarce.