Re-examining the meningitis belt: associations between environmental factors and epidemic meningitis risk across Africa

In 2002 Molesworth et al created a logistic regression model to investigate the spatial distribution of meningitis epidemics in Africa and determine geographical risk. We aim to understand if the meningitis belt geography and environmental risk factors have changed in the last 20 years by updating this analysis. Epidemic bacterial meningitis data from 2003-2022 were provided by WHO-AFRO. Districts (ADMN2) across Africa were coded as 1 if they had experienced a meningitis outbreak and 0 if they hadn’t. Environmental data were obtained from open sources. To preserve seasonal climate variation, monthly means of wind speed, rainfall, dust, and humidity were processed into climatic profiles using k-means clustering. Logistic regression was carried out with meningitis epidemic history as the dependent variable and k-means clusters of rainfall, dust, humidity, and windspeed, alongside average land-cover type as the independent variables. A sensitivity analysis was conducted excluding the Democratic Republic of Congo (DRC), due to limited laboratory confirmation of suspected cases. Rainfall, dust, and humidity demonstrated the strongest statistical association with meningitis outbreaks and were included in our final model. With a probability cutoff value of >0.4, our model had a specificity and sensitivity of 81.0% and 84.3% respectively in identifying districts having experienced a meningitis epidemic. With an extended dry season, the Sahelian region had the highest risk of meningitis outbreaks (probability >0.8). The inclusion/exclusion of the DRC had a significant impact on our model. Whilst Republic of the Congo, Gabon, Liberia, and Angola had a moderate risk of meningitis (probability >0.4) within our full model, no countries surrounding the meningitis belt were at risk for outbreaks within the sensitivity analysis. Although our research may suggest a south-westerly expansion of the meningitis belt this is difficult to verify. Surveillance of suspected meningitis cases should be accompanied by laboratory confirmation to improve outbreak definition.