Comparison of associations of food security instruments and mediators with premature all-cause and cardiovascular disease death in US adults

BACKGROUND: Food insecurity is associated with high morbidity and mortality and is typically measured with the 10-item US Adult Food Security Survey Module. Shorter instruments may capture similar information, but this has not been validated against mortality in general populations. 

METHODS: A nationally representative sample of individuals aged 20 to 74 years from the US National Health Interview Survey 2011 to 2018 was included, with deaths linked to the National Death Index through 2019. Cardiovascular disease deaths were ascertained by International Classification of Diseases-Tenth Revision codes for heart disease or stroke. Standard 10-, 6-, and 2-item food security instruments were compared for associations with premature all-cause and cardiovascular disease deaths occurring before age 75 years using Cox regression adjusted for demographics and social determinants of health and C statistics. Findings were replicated in the National Health and Nutrition Examination Survey 2004 to 2018, and differences were explored using mediation analysis. 

RESULTS: We included 218 136 National Health Interview Survey participants (mean age, 45.3 years; 50.8% women). Over a mean 5.0-year follow-up, 7025 premature deaths were observed (1711 from cardiovascular disease). In multivariable-adjusted models, hazard ratios (95% CIs) for all-cause death were similar among food security instruments (10-item, 1.22 [1.13, 1.32]; 6-item, 1.23 [1.13, 1.34]; and 2-item, 1.23 [1.14, 1.32]), and C statistics were identical (0.823). Hazard ratios (95% CIs) for cardiovascular disease deaths were also similar among food security instruments (10-item, 1.38 [1.17, 1.62]; 6-item, 1.27 [1.07, 1.51]; and 2-item, 1.41 [1.20, 1.66]), and C statistics ranged from 0.852 to 0.853. In the National Health and Nutrition Examination Survey replication (n=37 027, mean 7.8-year follow-up), associations were attenuated and became not statistically significant after adjustment for several cardiometabolic intermediates, particularly enrollment in food assistance programs, diabetes, low diet quality, inadequate or excessive sleep, and depression. 

CONCLUSIONS: A 2-item food security instrument captures similar mortality risk information compared with 10- and 6-item instruments. Furthermore, potential intermediate cardiometabolic factors may explain associations between food insecurity and mortality.