AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Comparing Vulnerability Profiles of Premature and Older-Age Cardiovascular Mortality Across U.S. Counties
By Huang J, Kim U, Li S et al.
Background: Cardiovascular disease (CVD) remains the leading cause of mortality in the United States, yet the contextual, environmental, and behavioral drivers may vary across life stages. We compared county-level vulnerability profiles associated with premature CVD mortality (ages 15–64) and older-age CVD mortality (ages ≥65) across the contiguous United States. Identifying these heterogeneous risk patterns may inform age-specific prevention strategies and healthcare resource allocation. Methods: We integrated county-level age-adjusted mortality rates for disease of the heart (2020–2024) from the SEER Program (derived from National Center for Health Statistics mortality data) with 60 pre-selected multidimensional predictors obtained from the 2023 AHRQ Community-Level Health Database and CDC PLACES. Predictors represented socioeconomic conditions, healthcare access, environmental exposures, demographics, and behavioral and cardiometabolic risk factors. Separate XGBoost models predicted premature and older-age mortality across 2,844 counties. Model interpretability was assessed using Shapley Additive Explanations (SHAP). Geographic patterns were mapped, and SHAP-based variable importance rankings were compared between age groups. Results: Geographic patterns differed substantially between age groups ( Figure 1 ). Premature CVD mortality exhibited a fragmented spatial pattern with localized high-mortality clusters across the West and Midwest, whereas older-age mortality demonstrated broader regional clustering, particularly across the Midwest. Both age groups experienced persistently elevated mortality throughout the Southern US. The machine learning models demonstrated good predictive performance for premature mortality (R ² =0.70) and moderate performance for older-age mortality (R ² =0.49). Premature mortality was most strongly associated with high blood pressure, socioeconomic conditions (income), and climate-related factors (extreme heat days and winter temperature), whereas older-age mortality was driven primarily by smoking, high blood pressure, physical inactivity, and short sleep ( Figure 2 ). Predictor rankings differed substantially between age groups ( Figure 3 ). Smoking increased from the 5th-ranked predictor of premature mortality to the leading predictor of older-age mortality, while median income and poverty were markedly less influential among older adults. Conclusions: Premature and older-age CVD mortality exhibit distinct geographic distributions and predictor profiles across U.S. counties. Socioeconomic and environmental factors were more prominent for premature mortality, whereas behavioral and cardiometabolic factors contributed more strongly to older-age mortality. These findings support age-specific precision public health strategies for CVD prevention and resource allocation.