Correlations of Socioeconomic and Clinical Determinants with United States County-Level Stroke Prevalence

 ABSOLUTELY USELESS! Nothing here gets survivors recovered! Don't you understand what survivors want? 100% recovery is the only goal in stroke! I'd have everyone here fired for not solving stroke!

Correlations of Socioeconomic and Clinical Determinants with United States County-Level Stroke Prevalence

Eric L. Stulberg MD, MPH, Lynda Lisabeth PhD, Andrea L. C. Schneider MD, PhD, Lesli Skolarus MD, MS, Kiarri N. Kershaw PhD, Alexander R. Zheutlin MD, MS, Benjamin R. E. Harris MD … See all authors

First published: 26 July 2024

https://doi.org/10.1002/ana.27039

Sections

PDF

Tools

Share

Abstract

Socioeconomic status (SES) is a multi-faceted theoretical construct associated with stroke risk and outcomes. Knowing which SES measures best correlate with population stroke metrics would improve its accounting in observational research and inform interventions. Using the Centers for Disease Control and Prevention's (CDC) Population Level Analysis and Community Estimates (PLACES) and other publicly available databases, we conducted an ecological study comparing correlations of different United States county-level SES, health care access and clinical risk factor measures with age-adjusted stroke prevalence. The prevalence of adults living below 150% of the federal poverty level most strongly correlated with stroke prevalence compared to other SES and non-SES measures (correlation coefficient = 0.908, R² = 0.825; adjusted partial correlation coefficient: 0.589, R² = 0.347). ANN NEUROL 2024

Socioeconomic status (SES) is a theoretical construct of social standing based off multiple dimensions such as financial wealth, education, social capital, occupation and geographic location, and can be represented by a variety of measurements. Stroke carries the largest disability burden among neurological illnesses in the United States (US), and is substantially influenced by SES at the individual, census-tract, county and national level.^1-3 Few studies have compared associations of different SES measures with key population stroke metrics, such as county-wide stroke prevalence.² Identifying how different SES measures correlate with population stroke metrics, relative to clinical and health care access determinants, could help inform best-practices for adequately capturing the influence of SES in stroke research and reducing SES-driven stroke disparities.⁴ For example, one study compared how a variety of SES indicators such as educational attainment, income and the need for medical assistance correlated with multiple health outcomes, finding that medical assistance reflects a large proportion of the influence of SES.⁵

Similar studies identifying what aspects of SES are key drivers of stroke prevalence could inform how best to represent SES in statistical models and help inform stroke-prevention and survivorship interventions and resource allocation. Using data from multiple nationally representative datasets, we conducted a cross-sectional ecological study comparing correlations of different measures of county-level SES, health care access and clinical risk factor measures with age-adjusted county-level stroke prevalence. We hypothesize county-level measures of SES correlate strongly with stroke prevalence relative to health care access and clinical variables.

Methods

County-Level Stroke Prevalence

Data on annual age-adjusted county-level stroke prevalence estimates were derived from the Center for Disease Control's (CDC) publicly available Population Level Analysis and Community Estimates (PLACES) database (https://www.cdc.gov/places).⁶ PLACES includes 2020 US county-level data. Estimates were derived from responses in the Behavioral Risk Factor Surveillance System (BRFSS) and census data,⁶ allowing for representation of almost all US counties and the variation of stroke prevalence throughout the United States.⁷ Using the county-level as the unit of analyses allows for the comparison of a wide variety of SES measures not often available at the individual-level.

County-Level SES, Clinical Risk Factors and Health Care Access Measures

We compared how the following county-level measures of SES correlate with stroke prevalence: social vulnerability index (SVI),⁸ Gini index,⁹ social deprivation index (SDI),¹⁰ community resilience estimate (CRE),¹¹ prevalence of unemployed adults, prevalence of adults without a high school degree, prevalence of poverty (defined as under 150% of the federal poverty level), median home value and rurality (defined by the rural–urban continuum).¹² Data was derived from 2019 to 2023 with additional details found in Table S1.

We included the following 2020 PLACES self-reported clinical risk factor prevalence estimates: hypertension, diabetes, hyperlipidemia, active smoking and obesity. The number of primary care physicians and hospitals with emergency rooms per 1,000 individuals were derived from the Agency for Health care Research and Quality social determinants of health database's 2020 estimates (https://www.ahrq.gov/sdoh/data-analytics/sdoh-data.html).

All data were deidentified, publicly available and did not require Institutional Review Board approval for use.

Statistical Analyses

We calculated descriptive statistics for each county-level variable and the Spearman correlation between each of the exposures. We then graphed scatterplots with locally estimated scatterplot smoothing (LOESS) curves to demonstrate the association between each exposure and stroke prevalence without imposing any functional form, and calculated Spearman correlation coefficients and R² values for each bivariate association. As poverty had the strongest correlation with stroke prevalence, we calculated its partial correlation coefficient by controlling for all other variables in the study. Last, we created heat maps depicting county-level quintiles of stroke and poverty prevalence and evaluated for congruence of the two variables. Sensitivity analyses excluding values above and below two standard deviations of the exposure distributions were performed to evaluate the impact of outliers.

Results

Our analytic sample included all 3,143 counties in the United States, with <1% of counties excluded for missing data in each bivariate analysis. Table S2 displays the descriptive statistics for each variable. Poverty was strongly correlated with known stroke clinical risk factors (Fig S1). Figures 1 and 2 display the derived scatterplots with LOESS-fitted curves and 95% confidence intervals (CI) for measures of SES, clinical risk factors and health care access measures with county-level stroke prevalence. Among SES and non-SES measures, the prevalence of poverty had the strongest correlation with county-level stroke prevalence (correlation coefficient: 0.908 [95% CI: 0.901–0.915], R² = 0.82).