The global landscape of peripheral artery disease: an integrated framework from macro-epidemiology to micro-risk-an integrated ecological and cross-sectional study

Background: Given the considerable global disease burden, the shifting epidemiological landscape of peripheral artery disease (PAD) demands urgent and sustained attention. Materials and methods: Using an ecological analysis of Global Burden of Disease 2021, we analyzed global, socio-demographic index (SDI), age-, and sex-specific trends in PAD age-standardized death rate (ASDR), age-standardized disability-adjusted life years rate (ASDALYR), age-standardized incidence rate, and age-standardized prevalence rate (ASPR) from 1990 to 2021, with AutoRegressive Integrated Moving Average projections to 2050. Risk-factor transitions were quantified via population-attributable fractions. To connect population trends with individual determinants, we additionally analyzed National Health and Nutrition Examination Survey (NHANES) data using multivariable logistic regression and interpretable machine learning (ML). Results: Globally, PAD ASDR and ASDALYR declined by 36% and 30% from 1990 to 2021, respectively, while ASPR declined modestly and was projected to increase until 2050. High-SDI regions achieved marked reductions, whereas low- and lower-middle-SDI regions showed rising trends across all major PAD metrics. Age-stratified estimated annual percentage change analysis identified adults aged 55–64 with the slowest decline of disease burden. Dominant risk factors of PAD deaths shifted from smoking (28.31%–20.04%) toward high fasting plasma glucose (21.35%–37.83%), and impaired kidney function (29.37%–31.38%). NHANES analysis identified advanced chronic kidney disease (OR 3.54, 95% CI 2.87–4.37), hyperglycemia (OR 1.17, 95% CI 1.11–1.23), systolic hypertension (OR 2.29, 95% CI 1.79–2.92), and older age (OR 1.08, 95% CI 1.07–1.09) as primary associated factors. ML models integrating metabolic and sociodemographic variables achieved good predictive performance for PAD risk. Conclusion: Our findings delineate a comprehensive landscape of PAD burden across multiple population strata, highlighting the limited improvement in adults aged 55–64 and the ongoing shift in attributable risk from behavioral to metabolic contributors. Integrating ML models incorporating metabolic and socio-demographic factors further enhances predictive value, offering new insights to inform targeted public-health strategies and clinical management.