Spatial and multilevel determinants of home delivery in Somalia: A spatiotemporal analysis from 2006 to 2020

BackgroundMaternal mortality remains a major public health challenge in fragile and conflict-affected settings, where access to skilled birth attendance is uneven. In Somalia, home delivery continues to predominate despite efforts to expand maternal health services. National averages, however, obscure substantial geographic and contextual disparities, and limited evidence integrates spatial and multilevel determinants of delivery location over time.ObjectivesThis study aims to answer the following research question: What are the individual, community, and regional determinants of home delivery in Somalia, and how have the spatial clustering and inequalities of this practice evolved between 2006 and 2020?DesignA retrospective, population-based cross-sectional design.MethodsWe conducted a retrospective, population-based cross-sectional analysis using nationally representative data from the 2006 Multiple Indicator Cluster Survey and the 2020 Somalia Health and Demographic Survey. Place of delivery was dichotomized as home versus health-facility delivery. Exploratory spatial data analysis was performed using Empirical Bayes smoothing, Global Moran's I, Geary's C, Local Indicators of Spatial Association, and Getis-Ord Gi* statistics. Ordinary kriging was applied to generate continuous risk surfaces, and spatiotemporal changes were assessed between 2006 and 2020. Multilevel logistic regression models were fitted to identify individual- and regional-level determinants of home delivery.ResultsHome delivery declined nationally over the study period (from 90.15% in 2006 to 79.40% in 2020) but remained highly prevalent, with pronounced regional heterogeneity. Significant positive spatial autocorrelation indicated non-random clustering of home delivery, particularly in southern and central regions, and spatial clustering intensified over time despite overall improvement (Global Moran's I increased from 0.34 in 2006 to 0.54 in 2020). At the individual level, higher household wealth (AOR = 0.12, 95% CI: 0.09-0.15) for highest quintile and adequate antenatal care utilization (AOR = 0.30, 95% CI: 0.25-0.35) for $\ge$ 4 visits were associated with lower odds of home delivery, whereas no school attendance (AOR = 1.95, 95% CI: 1.70-2.24) and higher parity (AOR = 1.28, 95% CI: 1.10-1.50) were associated with higher odds. Mudug and Gedo had markedly higher adjusted odds than Awdal. The fully adjusted intraclass correlation coefficient was 0.118, indicating substantively important residual regional variation.ConclusionHome delivery in Somalia is shaped by intersecting socioeconomic and geographic factors, with persistent and intensifying spatial inequalities. Region-specific strategies addressing structural barriers, such as deploying mobile midwifery outreach in southern hotspots like Mudug and Gedo, are essential to improve equitable access to facility-based childbirth.