A Gradual Leak, not a Single Bottleneck: Where Children Are Lost along the Immunisation Schedule in 38 Sub-Saharan African Countries

Background: Millions of children in sub-Saharan Africa remain incompletely vaccinated, yet coverage estimates inadequately document where along the immunisation schedule they are lost. Entry-point (never starting) and follow-up (not completing) failures call for different policy responses. We reconstructed the cascade, separated the two failure types, and quantified between-country heterogeneity at each transition. 

Methods: We pooled 163 of 210 sub-Saharan African Demographic and Health Surveys (1989–2024) with a complete vaccination module, analysing children aged 12–23 months (GATHER-compliant). We computed survey-weighted marginal coverage and conditional dose-completion probabilities for BCG, DPT/Pentavalent doses 1,2,3, and the first-dose measles-containing vaccine (MCV1) and assigned countries to four quadrants by entry-point (zero-dose ≥10%) and follow-up failure (≥20%) thresholds. A five-transition Markov chain was fitted DerSimonian–Laird random-effects meta-analysis with empirical-Bayes shrinkage on the logit scale; between-country variance (τ²) and Higgins’ I² were estimated. Sensitivity analyses tested card-confirmed restriction, threshold perturbation, and a Bayesian re-implementation. 

Findings: 263,385 children across 38 countries were analysed. Continental coverage fell from BCG 88·1% to DPT/Penta3 74·2% and MCV1 74·7% (14 percentage-point cumulative attrition). Bayes-shrunk transition probabilities ranged from 95·5% (BCG→DPT/Penta1) to 88·5% (DPT/Penta3→MCV1); the largest mid-cascade leak was DPT/Penta2→ DPT/Penta3 (89·2%, 95% CI 85·8–91·9). 21 countries were both-low, six entry-point dominant (Nigeria 54% zero-dose), four follow-up dominant, and seven both-high. Between-country variance fell roughly tenfold along the cascade (τ² 0·96 to 0·11; I²>97% throughout). The predicted continental average was 59·0% completed and 13·7% zero-dose. Findings were robust to all sensitivity analyses.

Interpretation: The cascade leaks gradually rather than at a single bottleneck, and countries differ roughly tenfold more in schedule entry than in completion. Closing the entry-point gap offers the greatest scope for narrowing continental inequity. Each country’s quadrant predicts the most appropriate intervention package — outreach and zero-dose mapping, reminders and defaulter tracing, or integrated EPI strengthening.