Correlation Analysis of Age Structure Metrics and Cause-Specific Mortality: A Global and WHO Regional Comparison using the Top 5 Causes of Death in 2021

Abstract

Global mortality models often depend on simple aging metrics that ignore the structural diversity in populations. To tackle this issue, we evaluated three metrics: median age, percentage aged 65+, and the new PoPDivergence, against mor tality from the top five global causes of death in 2021 across 178 countries. PoPDivergence, which is based on Kullback-Leibler divergence, identifies differences in population pyramids from optimized references. Using Spearman correlations, Levene’s test, and Kruskal-Wallis analysis, we found that PoPDivergence predicts COVID-19 mortality more accurately in younger regions, such as the East ern Mediterranean. In contrast, traditional metrics better capture chronic disease mortality in aging regions like Europe and in younger areas like Africa, where non-communicable diseases are increasing. No metric significantly predicted lower respiratory in fection mortality. Our study presents a new, statis tically validated framework for selecting age struc ture metrics suited to specific diseases and regional contexts, which offers important improvements for global health modeling and policy development.