Which African Countries are at Risk of Missing SDG 3.2? Bayesian Mapping of Under-Five Mortality Using UNICEF 2024 Data

Technical Summary: Bayesian Mapping of Under-Five Mortality and SDG 3.2 Risk in Africa

Problem Statement
Despite global progress in reducing child mortality, the pace of improvement across Africa remains uneven, raising concerns about the continent's ability to meet Sustainable Development Goal (SDG) 3.2. This target aims to reduce under-five mortality to fewer than 25 deaths per 1,000 live births by 2030. Existing assessments often rely on observed mortality levels or deterministic projections that fail to adequately quantify the uncertainty surrounding future attainment or the relative urgency of interventions. Furthermore, national averages frequently obscure persistent geographical inequalities and spatial clustering of mortality risks. There is a critical need for a framework that integrates uncertainty quantification, distance from the SDG target, and the pace of mortality reduction to identify countries most at risk of missing the 2030 deadline.

Methodology
The study employed a Bayesian spatial ecological forecasting design using 2024 country-level data from 49 African countries obtained from UNICEF. The analytical framework utilized Integrated Nested Laplace Approximation (INLA) to fit structured additive regression models.

• Model Specification: The authors compared Gaussian, Gamma, and Exponential likelihoods. The spatial structure was modeled using an intrinsic Conditional Autoregressive (ICAR) prior to account for spatial dependence between neighboring countries.
• Predictors: Explanatory variables included DTP3 immunization, exclusive breastfeeding, underweight and stunting prevalence, life expectancy, population demographics, health expenditure, maternal mortality ratio, and prevalence of pneumonia and diarrhoea.
• Model Selection: Model fit was evaluated using the Deviance Information Criterion (DIC), Watanabe-Akaike Information Criterion (WAIC), and conditional predictive ordinates.
• New Indices: To move beyond descriptive statistics, the study developed two composite metrics:
  • SDG Failure Index (SFI): Combines the posterior exceedance probability (the probability that a country's mortality rate exceeds the 2.5% threshold) with the magnitude of the SDG gap.
  • Priority Intervention Index (PII): Integrates the SFI with the Annual Rate of Reduction (ARR) to classify countries into high, moderate, or low priority for intervention.
• Forecasting: Bayesian posterior predictive simulations were used to estimate the probability of attaining SDG 3.2 by 2030, assuming the continuation of recent mortality reduction trends.

Key Results

• Spatial Clustering: Significant positive spatial autocorrelation was observed (Moran's I = 0.355, p < 0.001). Hotspot analysis identified a "High-High" cluster in Benin, Nigeria, and Cameroon, indicating a localized region of persistently high mortality. Conversely, Tunisia was identified as a "Low-Low" cluster.
• Model Performance: The Gamma model provided the best fit (DIC = 114.92; WAIC = 111.71). In this model, diarrhoea prevalence was the only significant predictor of under-five mortality (posterior mean = 0.030; 95% CrI: 0.004–0.056) after accounting for spatial dependence and other covariates.
• SDG Attainment Status: Of the 49 countries analyzed, only five (10.2%) had already achieved the SDG 3.2 target, and four (8.2%) were near achievement. Conversely, 23 countries (46.9%) were classified as "High Risk," and 41 countries (83.7%) were deemed unlikely to achieve the target by 2030 under current trajectories.
• Regional Disparities: West Africa recorded the highest mean mortality (7.05%), followed by Central Africa (6.89%). North Africa had the lowest mean (1.64%) and the lowest exceedance probability (0.082).
• Inequality: The Gini coefficient for under-five mortality across the continent was 0.263, indicating moderate inequality. The top 10 worst-performing countries (e.g., Nigeria, Niger, Somalia) were concentrated in West, Central, and East Africa, while the best performers (e.g., Libya, Tunisia, Morocco) were in North Africa.

Key Contributions
The paper introduces a practical analytical framework for SDG monitoring that integrates:

1. Bayesian Exceedance Probability Mapping: Quantifying the probability that a country will remain above the mortality threshold, rather than just estimating point values.
2. Composite Indices: The development of the SDG Failure Index (SFI) and Priority Intervention Index (PII) allows for the simultaneous assessment of mortality deficits and the urgency of required action.
3. Probabilistic Forecasting: Moving beyond deterministic trends to provide country-specific probabilities of target attainment by 2030, explicitly accounting for uncertainty.

Significance and Claims
The authors claim that their study provides a more comprehensive assessment of child survival prospects than previous literature, which has predominantly focused on describing mortality patterns or identifying risk factors without evaluating the probability of missing international targets. By identifying that nearly half of African countries are at high risk of missing SDG 3.2, and that only five are likely to succeed, the study offers an evidence-based framework for prioritizing resource allocation. The authors posit that the integration of exceedance probability, the SFI, and the PII provides a mechanism for policymakers to distinguish between countries that are merely above the threshold and those with a near-certain likelihood of failure, thereby facilitating geographically targeted interventions in West, Central, and parts of East Africa. The study concludes that without accelerated progress and strengthened health systems, particularly in addressing diarrhoeal diseases and regional inequalities, the majority of African countries will miss the 2030 target.