Geospatial mapping of drug-resistant tuberculosis prevalence in Africa at national and sub-national levels

Objectives To map subnational and local prevalence of drug-resistant tuberculosis (DR-TB) across Africa. Methods We assembled a geolocated dataset from 173 sources across 31 African countries, comprising drug susceptibility test results and covariate data from publicly available databases. We used Bayesian model-based geostatistical framework with multivariate Bayesian logistic regression model to estimate DR-TB prevalence at lower administrative levels. Results We estimated 148,239 DR-TB cases (95% uncertainty interval: 17,499-313,683) in Africa, showing significant variation by country. Eswatini and South Africa had highest case numbers, while Algeria and Egypt had the lowest. The highest DR-TB prevalence was estimated in Eswatini (53.26; 95% uncertainty interval 13.13-66.12), Morocco, Tunisia, and South Africa, while the lowest prevalence was found in Gabon, the Republic of Congo, Sierra Leone, and Mali. Marked subnational variation in DR-TB prevalence was noted, where 91 subnational areas across 12 countries had prevalence rates higher than their respective national averages. Factors such as mean temperature (β = 2.01; 95% CrI: 1.21, 3.42), population density (β = 0.41; 95% CrI: 0.19, 0.95), and fine particulate matter (β = 0.66; 95% CrI: 0.20, 0.80) were positively associated with DR-TB prevalence. Conclusion The study highlights substantial national and subnational variability in DR-TB prevalence across Africa, aiding policymakers in designing localized TB control interventions.