Conflict patterns in Somalia vary by region, causing uneven disruptions to routine immunization and outbreak control. National summaries often conceal district-level risks. Using conflict-aware geospatial diagnostics can support more targeted subnational planning by identifying areas where insecurity may limit immunization delivery and monitoring. Armed Conflict Location and Event Data (ACLED) conflict events (battles, explosions, or remote violence, and violence against civilians) were aggregated to district-years. Global Moran’s I with queen contiguity and permutation-based p-values and Local Getis-Ord Gi* statistics were applied to identify spatial clustering and hotspots (z > 1.96). Annual hotspot proportion was defined as the percentage of districts classified as hotspots. A static set of 74 district boundaries was used to ensure temporal comparability across the study period. Conflict events showed significant positive spatial autocorrelation each year, with Moran’s I = 0.27 (z = 4.61, p < 0.001) in 2015 and Moran’s I = 0.33 (z = 5.50, p < 0.001) in 2023. Districts classified as hotspots comprised 5.4% in 2015, peaked at 10.8% in 2019, and remained at 10.8% in 2023. Multi-year maps showed persistent concentration of conflict in a limited subset of districts rather than dispersion. Descriptive spatial diagnostics show recurrent, non-random clustering of violent conflict across Somali districts. The analysis offers an operational, conflict-focused framework to inform subnational immunization microplanning, delivery modality selection, and monitoring when combined with relevant programmatic data. The workflow is reusable for conflict-sensitive planning and surveillance in fragile settings.
John Kwame Duah (Tue,) studied this question.