EPD404 - Neighborhood Retail Food Environment and Risk of Early-Onset Colorectal Cancer: An Ecological Analysis

Purpose/Background: The rising incidence of early-onset colorectal cancer (EOCRC) is poorly understood. While certain diets are a known risk factor for colorectal cancer, few studies have examined the role of the neighborhood food environment in EOCRC. We investigated the relationship between food environment and EOCRC incidence across the state of Connecticut.

Methods/Interventions: We conducted a retrospective ecological study to evaluate neighborhood-level characteristics and EOCRC incidence among all adults aged 20-49 years old diagnosed between 01/2013 and 03/2025 within a statewide academic health system. We hypothesized that census tracts with unhealthier food environments would have a higher incidence of EOCRC after controlling for other census-level covariates. Food establishment data for each census tract were obtained from Data Axle, a comprehensive database of businesses, and reclassified according to study-specific definitions. An Unhealthy Food Environment Index ("UFEI") was derived from the Centers for Disease Control and Prevention's modified Retail Food Environment Index (mRFEI) and defined as the ratio of unhealthy food retailers to the total number of unhealthy and healthy food retailers within a half-mile boundary of each census tract (range 0-100). Residential addresses were geocoded to U.S. census tracts, which were categorized into tertiles of deprivation using the CDC's Social Vulnerability Index. EOCRC cumulative incidence was aggregated at the census tract level, with additional tract-level covariates extracted from the CDC's 2022 PLACES survey. An adjusted negative binomial regression, offset by the log of the 2019 census population aged 20-49 years, was used to model the relationship between UFEI and EOCRC incidence.

Results/Outcomes: A total of 642 patients were diagnosed with EOCRC across 632 census tracts between 2013 and 2025. The median age of diagnosis was 44 years (IQR, 39-47). Median UFEI was 62.71 [IQR, 0-100] in low vulnerability tracts (n=222), 82.78 [IQR, 61.35-100] in medium vulnerability tracts (n=216), and 100 [IQR, 81.13-100] in high vulnerability tracts (n=194; Table 1). Adjusting for census tract-level sociodemographic characteristics, a higher UFEI was significantly associated with increased cumulative incidence counts of EOCRC in the highest vulnerability census tracts (cumulative incidence rate ratio [IRR] = 1.058, 95% CI: 1.002,1.118, p = 0.043; Figure 1). In addition, for every one-percentage-point increase in the tract-level prevalence of diabetes, the cumulative incidence rate of EOCRC increased by 12.2%. No significant associations were observed between UFEI and EOCRC case counts in lower vulnerability tracts.

Conclusion/Discussion: Living in an unhealthy food environment was associated with higher EOCRC incidence in the most socioeconomically vulnerable census tracts. These findings highlight the need to examine food quality as a potential causal factor contributing to the rising incidence of EOCRC.

Figure 1. Negative Binomial Regression: Incidence Rate Ratios for Early-Onset Colorectal Cancer in the Highest Vulnerability Census Tracts
Model adjusted for census tract-level sociodemographic characteristics; offset by log population aged 20-49. Abbreviations: IRR, incidence rate ratio; CI, confidence interval; PCP, primary care provider.
Table 1. Sociodemographic Characteristics and Unhealthy Food Environment Index of Connecticut Census Tracts Stratified by Social Vulnerability Index
a Represents results of Negative binomial regression

b Represents results of Kruskal-Wallis H Test