Quantifying undetected tuberculosis in Ethiopia using a novel geospatial modelling approach

Tuberculosis (TB) is the leading infectious cause of death globally, with approximately three million cases remaining undetected, thereby contributing to community transmission. Understanding the spatial distribution of undetected TB in high-burden settings is critical for designing and implementing geographically targeted interventions for early detection and control. This study presents the first estimates of numbers of undetected TB cases in Ethiopia at national and local levels using novel geospatial method. We employed a Bayesian geostatistical modelling framework, incorporating national TB prevalence survey and TB notification data together with climatic and environmental variables, to estimate the number of undetected TB cases at district and national levels. Spatial clustering of undetected TB cases was assessed using Moran's Index statistic and Local Indicator of Spatial Autocorrelation (LISA). A Bayesian Poisson regression model with conditional autoregressive (CAR) prior structure was developed to identify drivers of the clustering. We estimated a total of 51,041 undetected TB cases (95% CI: 50,599, 51,486) in Ethiopia, with the majority of these cases predicted in the Oromia region (20,440), Amhara region (9614), and South Ethiopia region (6061). Spatial clustering of undetected TB cases was observed in districts near the international borders, including the Ethiopia-Somalia and Ethiopia-Kenya border regions, as well as in several districts of Southern Ethiopia. The number of undetected TB cases was negatively associated with the proportion of the population with good mass media exposure (Incidence rate ratio (IRR): 0.67 95% CI: 0.56, 0.80) and the proportion of the population with high wealth index (IRR: 0.73, 95% CI: 0.60, 0.90). Our findings revealed a high burden of undetected TB in Ethiopia, with spatial clustering in border regions and areas with limited healthcare access. Targeted TB screening interventions to communities with low socioeconomic status along with improving mass media exposure in these regions, could significantly reduce the burden of undetected TB in Ethiopia.