Association Between Food Desert Residence and Ischemic Stroke Risk: A Population-Based Study.

Background And Objectives: Food deserts (FDs) are low-income areas with poor access to healthy foods. FD residents have higher rates of several cardiovascular risk factors, but the link between FDs and stroke has not been well studied. We evaluated whether FD residence was associated with incident ischemic stroke within the Greater Cincinnati/Northern Kentucky Stroke Study (GCNKSS) and whether this association was due to low income, poor food access, or both.

Methods: All hospitalized stroke cases in the GCNK region were ascertained during calendar year 2015 using ICD-9 and ICD-10 codes for screening and confirmed by physician review. Patient home addresses were geocoded using Decentralized Geomarker Assessment for Multi-Site Studies. FD locations were obtained from the US Department of Agriculture Food Access Research Atlas, defined as census tracts with both poor food access and low income according to established definitions based on proximity to healthy food sources as well as area poverty rates and median household income. Population estimates were obtained from the 2015 5-year American Community Survey. Poisson regression models were used to calculate census tract-level incidence rates by FD status, as well as by food access and income categories, adjusting for age, sex, race, and income-by-access interaction.

Results: A total of 1,802 first-ever ischemic stroke incidents occurred in the region during the study period. Stroke patients had a mean age of 69.7 years, and 53% were female. In unadjusted models, FD residence (vs non-FD) was associated with higher stroke incidence (incidence rate ratio [IRR] 1.23; 95% CI 1.06-1.42; < 0.01). After adjustment for age, sex, and race, this relationship was attenuated and no longer statistically significant (IRR 1.11; 95% CI 0.96-1.30; = 0.17). In a model where FD status was replaced by area income and food access (i.e., the 2 components of the FD definition), low income was associated with greater stroke incidence after full adjustment (IRR 1.21; 95% CI 1.05-1.39; = 0.01) while poor food access was not (IRR 0.91; 95% CI 0.81-1.01; = 0.08).

Discussion: FD residents are at increased stroke risk, and this is primarily due to low area income rather than poor food access. Alternative measures of the food environment may help elucidate the links between income, dietary patterns, and stroke risk.