Pregnancy loss in East Africa: Pooled prevalence and subnational geographic patterns.

BACKGROUND Pregnancy loss remains a major public health concern in East Africa despite global efforts to improve maternal health. This study aimed to estimate the pooled prevalence and subnational spatial patterns of pregnancy loss among reproductive-aged women in nine East African countries using nationally representative DHS data (2015-2023). METHODS Pooled prevalence was estimated using a Bayesian random-effects model. Spatial relationships across 169 subnational regions were defined using the queen contiguity method. Global spatial autocorrelation was assessed with Moran's I, and local clusters were identified using Anselin's local Moran's I and SaTScan's Bernoulli model. Ordinary Kriging was applied to predict spatial distribution across the study area. RESULTS The pooled prevalence of pregnancy loss was 14% (95% credible interval: 10-21%). Significant spatial clustering was observed (Moran's I = 0.545, p < 0.001), with high-risk areas concentrated in southern Kenya, surrounding regions, and southern Mozambique. SaTScan identified five significant clusters, and Kriging interpolation confirmed elevated risk in these areas. CONCLUSION There are substantial subnational disparities in pregnancy loss across East Africa. Spatial mapping highlights priority regions for targeted maternal health interventions, providing a practical tool to guide policymakers and reduce reproductive health inequalities.