Geographic and social disparities in sanitation: a Bayesian hierarchical approach using Somalia's 2020 DHS data

ABSTRACT Background: In fragile states like Somalia, where decades of instability have created a severe sanitation crisis, access to basic facilities remains critically low. Effective interventions require a clear understanding of who is being left behind and why. Methods: We analyzed 31,930 households from the 2020 Somalia Demographic nd Health Survey (DHS) using a Bayesian two-level logistic regression model. This method assessed the odds of having any toilet facility (vs. open defecation) by nesting households (Level 1) within 68 administrative districts (Level 2), controlling for individual and contextual factors. A staggering 42.6% of households practice open defecation. Geography was the overwhelming determinant of access; district-level variation was immense (Median Odds Ratio [MOR] = 7.47). Compared to rural households, urban residents had vastly higher odds of access (AOR: 6.47; 95% CrI [6.00, 7.05]), while nomadic households had near-zero odds (AOR: 0.06; 95% CrI [0.05, 0.08]). Paradoxically, after accounting for geography, the expected positive effect of wealth disappeared, and male-headed households had significantly lower odds of access than female-headed ones (AOR: 0.83; 95% CrI [0.78, 0.89]). Conclusion: In Somalia, access to sanitation is fundamentally a function of geography, not just poverty. The immense disparities between districts and lifestyles render uniform national policies ineffective.