Small area estimation of anemia risk in Ethiopian zones by combining the survey and census dataset with hierarchical Bayes Fay–Herriot modeling

Childhood anemia remains a critical global public health challenge, particularly in lower- and middle-income countries. Although national survey datasets, such as the Demographic and Health Survey (DHS), provide robust estimates of anemia prevalence at national and regional levels, small sample sizes at localized administrative divisions such as zones lead to high sampling variability, rendering direct survey estimates unreliable for local policy planning. To address this subnational data gap, this study applied a hierarchical Bayes small area estimation (SAE) approach to generate reliable zonal-level estimates of childhood anemia prevalence in Ethiopia. By leveraging the strength of area-level auxiliary variables derived from the Population and Housing Census dataset, the SAE model effectively borrowed strength across spatial domains. The model-based anemia prevalence estimates demonstrated substantially smaller coefficients of variation compared to direct survey-based estimates, confirming a marked reduction in sampling error and a significant gain in statistical precision. Ultimately, by providing reliable and precise zonal-level evidence, this research fulfills national demands for disaggregated data at lower administrative tiers. These localized estimates support evidence-based policy interventions, facilitate equitable health resource allocation, and align directly with global commitments toward achieving the Sustainable Development Goals (SDGs) for good health and well-being.