valentineghanem-bit/ghana-immunisation-coverage-261-districts

Subnational immunisation coverage gaps and structural determinants in Ghana - 261 districts (DHS 2022 + Census 2021) # Subnational immunisation coverage gaps and structural determinants in Ghana **Ecological equity mapping and an interpretable district risk index across 261 districts**  Reporting: STROBE + RECORD  ·  License: MIT --- ## 1. Overview This repository contains the full, reproducible analysis behind an ecological study of childhood immunisation coverage across **all 261 districts of Ghana**. Using the 2022 Ghana Demographic and Health Survey (DHS; region-level immunisation, 16 regions) linked to the 2021 Ghana Population and Housing Census (district-level socioeconomics, 261 districts), it maps coverage, builds an **Immunisation District Risk Index (IDRI)**, characterises its spatial structure, and tests whether machine learning improves district risk classification over logistic regression. The design is explicitly **ecological** and methodologically candid: immunisation coverage is region-resolved (effective N = 16), so it is mapped at region level while spatial and predictive analyses use a district-varying socioeconomic surface. **Author:** Valentine Golden Ghanem · Ghana COCOBOD Cocoa Clinic, Accra; University of Cape Coast (PhD candidate) · ORCID 0009-0002-8332-0220. ## 2. Key findings - Full-immunisation coverage averaged **75.2%** (median 77.8%); **180 / 261 districts (69.0%)** fell below the 80% target. - A marked **Northern–Southern gradient**: Northern-belt districts (n=55) averaged 67.5% full coverage versus 77.2% in the Southern belt (n=206), with higher zero-dose prevalence (3.8% vs 1.1%) and DPT1–DPT3 dropout (10.2 vs 7.7 points). Coverage ranged from 55.8% (Northern) to 87.7% (Western North); Western Region (62.5%) was a southern low-coverage exception. - District vulnerability (IDRI) was strongly spatially clustered (**Global Moran's I = 0.72, z = 17.5, p **Tip:** both files are fully self-contained — ECharts and the map geometry are inlined — so they render **offline** with no server or network. The poster is print-ready at A …