Spatial variation and determinants of household food insecurity in Somalia using spatial and multilevel analysis of nationwide data

Abstract

Background: Food insecurity is a critical challenge in Somalia, hindering SDG 2 progress. Exacerbated by conflict and climate change, understanding its complex drivers at multiple levels is vital for effective interventions. This study investigates the spatial variations and multilevel determinants (individual and community) of household food insecurity nationwide. Methods: Using weighted 2020 Somalia Demographic and Health Survey data (N = 15,826), this cross-sectional study assessed household food insecurity (“lack of food”). Multilevel logistic regression examined individual (e.g., wealth, household head characteristics, assets) and community (region, residence) factors, accounting for data hierarchy. Spatial autocorrelation analyses (Global Moran’s I, Local Moran’s I, Getis-Ord Gi*) identified geographic clustering and hotspots/cold spots. Results: High prevalence (70%) of household food insecurity was observed. Multilevel analysis showed richer wealth status (AOR = 0.21, 95% CI = 0.19–0.25 vs. poor), female household headship (AOR = 0.82, 95% CI = 0.76–0.90), older head age (> 50 years: AOR = 0.85, 95% CI = 0.77–0.93), larger household size (> 7 members: AOR = 0.83, 95% CI = 0.77–0.91), and school attendance by the head (AOR = 0.87, 95% CI = 0.80–0.95) were significantly associated with lower odds of food insecurity. Nomadic residence significantly increased odds compared to urban settings (AOR = 1.99, 95% CI = 1.59–2.51). Significant regional variations persisted. Spatial analyses revealed statistically significant positive spatial autocorrelation (Moran’s I = 0.64, p