Retrospective analysis of macroscopic health, socioeconomic, and demographic risk predictors for COVID-19 accumulated mortality ratio

Background : COVID-19 pandemic resulted in stark disparities in mortality outcomes across countries, influenced by a complex interplay of demographic, economic, political, and health-related factors. This study investigates the macroscopic risk factors of COVID-19 mortality using stepwise multiple linear regression models on data from 174 countries. Methods : We carried out multiple regression modeling, with automated predictor selection. Two regression model-ing approaches were employed: one using only main effects (Modeling approach A) and another including pairwise interaction terms (Modeling approach B) to capture conditional effects, and another using for improved interpretability. Ten distinct analyses were conducted, covering comparisons across country development status levels (developed, developing, least developed) and between temporal phases (2020–2023), enabling both structural and dynamic assessments. Model-ing approach B was applied only when data availability was sufficient to avoid overfitting. Predictors considered per country are (1) the prevalences of health conditions (obesity, diabetes, hypertension), (2) socioeconomic indices (Gini index , GDP, democracy index), and demographics (age distribution). Results : The analysis identified age distribution, especially the proportion of individuals aged 65 and older, as the most consistent and statistically significant predictor of COVID-19 mortality. Obesity and income inequality (Gini index) were also found to be significant predictors across pandemic temporal phases, and country development categories. The study does not find any effect of the prevalence of respiratory conditions. The prevalence of diabetes is consistently found as 1 negatively correlated with mortality, while obesity prevalence is a positively correlated predictor of mortality. The effect of socioeconomic factors are significant at the global level, and it is more important in developed countries than in developing and under-developing counties. Conclusions : Aging population seems to be the strongest predictor for bad pandemic outcomes across all analyses carried out. The effect of the prevalence of medical conditions and socioeconomic factors seems conditioned to the stage of development of the countries and the pandemic phases. Negative correlation of diabetes prevalence and the absence of effect of respiratory disease prevalence are unexpected findings.