Abstract WE503: Concordance Between Household and Area-Level Food Security Among People with Pre-Clinical Heart Failure

Abstract

Introduction: Diet is a key component of cardiovascular risk profiling in pre-clinical heart failure. Food insecurity, defined as limited or uncertain access to nutritionally adequate food, is a barrier to a healthy diet and warrants evaluation for intervention. However, self-reported food security is rarely measured, with researchers/clinicians often relying on area-level measurements. This study evaluates concordance between household and area-level food security to assess whether county-level food security is an appropriate proxy for household-level food security. Hypothesis: County-level food security is a poor proxy for household-level food security. Methods: In September 2024, we enrolled 400 participants via the YouGov online survey platform from the Southern Atlantic Region, with a design to include equal proportions of male vs female, Black vs non-Black, and high vs low-income individuals. Household food insecurity was assessed by the 6-item USDA questionnaire. County-level food security was measured by Feeding America’s Mind the Meal Gap study and the CDC’s Social Vulnerability Index (SVI), linking participants’ ZIP codes to county data. Concordance was assessed between household food security categories (high/marginal vs low vs very low) and county food security tertiles (1 = most secure; 3 = least secure), as well as county SVI tertiles. Results: The final sample included 398 participants from nine states (mean age 61.6 ± 11.6). Half lived in counties in food insecurity tertile 1, while 7.5% lived in tertile 2 counties. There was no concordance between county tertiles and household food insecurity categories (weighted kappa = 0.05, 95% CI: -0.03 to 0.12). The concordance between ZIP code SVI tertiles and household food insecurity was weak (weighted kappa = 0.05, 95% CI: 0.002 to 0.09). A weak correlation existed between household food insecurity scores and ZIP code-level SVI (Spearman’s rho=0.15), driven mainly by SVI sub-measures of socioeconomic status (rho=0.20) and housing/transportation (rho=0.19). Conclusion: In this study, county-level food security classifications were not valid proxies for household-level food security. Using them interchangeably for risk stratification may lead to misclassification. Clinicians and researchers should directly assess food security at the household level when evaluating risk.