From Global to Local: A Multiscale Geographically Weighted Regression Analysis of Bovine Brucellosis Risk Factors.

Bovine brucellosis remains a major zoonotic threat despite ongoing control measures. Conventional strategies often target broad administrative units, potentially overlooking local dynamics relevant to elimination in low-prevalence settings. We conducted a township-level spatial epidemiological study in southwestern Hubei Province, China, analyzing serological data from 63,222 cattle in 6335 herds collected in April 2024. Spatial clustering was assessed using Moran's I, Getis-Ord General G, and Local Moran's I (LMi), while multiscale geographically weighted regression (MGWR) evaluated associations with six township-level covariates: terrain flatness (plain-to-hill ratio [PHR]), road network density (RND), cattle density (Cden), goat density (Gden), large-scale rearing ratio (LSR), and incoming cattle flow (ICF). A distinct high-risk belt was identified in the southeast-to-east-central region, with positive townships forming high-high clusters. PHR was a significant positive predictor, while RND was negative; MGWR highlighted localized positive effects of LSR. These findings demonstrate the importance of fine-scale, geographically tailored interventions for brucellosis elimination.