County-Level Socio-Environmental Factors Associated With Stroke Mortality in the United States: A Cross-Sectional Study

Abstract

We used machine learning methods to explore sociodemographic and environmental determinants of health (SEDH) associated with county-level stroke mortality in the USA. We conducted a cross-sectional analysis of individuals aged ≥15 years who died from all stroke subtypes between 2016 and 2020. We analyzed 54 county-level SEDH possibly associated with age-adjusted stroke mortality rates/100,000 people. Classification and Regression Tree (CART) was used to identify specific county-level clusters associated with stroke mortality. Variable importance was assessed using Random Forest analysis. A total of 501,391 decedents from 2397 counties were included. CART identified 10 clusters, with 77.5% relative increase in stroke mortality rates across the spectrum (28.5 vs 50.7 per 100,000 persons). CART identified 8 SEDH to guide the classification of the county clusters. Including, annual Median Household Income ($), live births with Low Birthweight (%), current adult Smokers (%), adults reporting Severe Housing Problems (%), adequate Access to Exercise (%), adults reporting Physical Inactivity (%), adults with diagnosed Diabetes (%), and adults reporting Excessive Drinking (%). In conclusion, SEDH exposures have a complex relationship with stroke. Machine learning approaches can help deconstruct this relationship and demonstrate associations that allow improved understanding of the socio-environmental drivers of stroke and development of targeted interventions.

Keywords: epidemiology; health policy; machine learning; sociodemographic and environmental determinants of health; stroke.

Figure 1.

Classification and regression tree (CART) analysis to predict county-level stroke age-adjusted mortality rate. Notes: Each path down to a terminal node represents a county SEDH cluster. Box plots in the terminal nodes represent the median age-adjusted stroke mortality (per 100,000 people). The minimum number of counties in a terminal node was set to 150. Clusters were labeled alphabetically based on median age-adjusted all stroke mortality rate, from lowest to highest Table 1. Dot chart of random forest analysis showing variable importance for predicting county-level age-adjusted cardio-oncology mortality. Notes: the most important variable is at the top and scaled to 100%. The importance of the rest of the variables is shown relative to the top one. Full descriptions of the variables available in Table 1. Abbreviations: RMP, Risk Management Plan; NPL, National Priorities List.

Figure 2.

United States County Maps of (A) age-adjusted stroke mortality (per 100,000 people), divided by natural breaks into 10 groups. (B) County clusters, A-J, from smallest to largest age-adjusted stroke mortality.

Figure 3.

Dot chart of random forest analysis showing variable importance for predicting county-level age-adjusted cardio-oncology mortality. Notes: the most important variable is at the top and scaled to 100%. The importance of the rest of the variables is shown relative to the top one. Full descriptions of the variables available in Table 1. Abbreviations: RMP, Risk Management Plan; NPL, National Priorities List.