Zonal variations in childhood stunting in Ethiopia: Bayesian geo-additive analysis.

INTRODUCTION: Childhood stunting remains a major public health challenge in Ethiopia, affecting more than one-third of children under five years of age. Achieving the Seqota Declaration target of ending stunting by 2030 requires understanding where the burden is highest and the factors driving geographic disparities. This study provides the first zonal-level analysis of childhood stunting in Ethiopia, identifying key determinants and spatial patterns to inform targeted interventions. METHODS: Data from 4916 children under five years of age from the 2019 Ethiopian Mini Demographic and Health Survey were analyzed. A Bayesian semiparametric geo-additive model was fitted using Markov Chain Monte Carlo methods to estimate linear, nonlinear, and spatial effects while separating structured and unstructured spatial variation across zones. RESULTS: Substantial spatial variation in childhood stunting persisted after adjusting for socioeconomic, demographic, and maternal factors. High-risk zones were concentrated in northern Ethiopia, particularly Wag Hemra, Southern Tigray, and Afar Zone 2. Low maternal education, younger maternal age, early age at first birth, higher birth order, multiple births, rural residence, and household poverty were significantly associated with stunting. Residual spatial clustering indicated the presence of additional unmeasured local factors. CONCLUSIONS: Childhood stunting remains unevenly distributed across Ethiopian zones, with distinct geographic hotspots persisting beyond measured risk factors. Targeting high-burden zones and incorporating spatial evidence into nutrition and maternal-child health programs will be essential for accelerating progress toward Ethiopia's goal of ending childhood stunting by 2030.