Use of Bayesian Statistical Methods for Disease Mapping of Tuberculosis Incidence in Northern Nigeria

Bayesian disease mapping provides spatially smoothed, uncertainty-aware incidence estimates that are statistically superior to crude rate maps for characterising the geographic distribution of tuberculosis burden in northern Nigeria, where reporting systems are incomplete and small-area estimation is required. This study applies Bayesian spatial models to map tuberculosis incidence across LGAs in the North West and North East geopolitical zones using 2020 to 2022 tuberculosis notification data from the National Tuberculosis and Leprosy Control Programme. Besag-York-Mollie intrinsic conditional autoregressive models implemented in R-INLA provided spatially smoothed incidence estimates with 95 percent credible intervals. Posterior exceedance probabilities identified LGAs with incidence significantly above the zonal median. Spatial clustering was assessed by Moran's I statistic. Crude notification rates ranged from 42 to 328 per 100,000 but were highly unstable in small-population LGAs. Bayesian smoothed estimates substantially reduced extreme value instability. Moran's I confirmed significant spatial clustering (I = 0.34, p = 0.001). High-burden LGA clusters were identified in Borno, Kano, and Zamfara States. Posterior exceedance probabilities identified 18 LGAs with greater than 95 percent probability of incidence exceeding twice the median. The study provides a statistically rigorous tuberculosis burden atlas for northern Nigeria and recommends its use by the National Tuberculosis Programme to guide targeted intervention resource allocation. Keywords: Bayesian disease mapping, tuberculosis, spatial statistics, northern Nigeria, INLA