Neighborhood retail food environment, diet quality and type 2 diabetes incidence in four Dutch cohorts

Current evidence on the associations between the food environment and type 2 diabetes (T2D) is inconsistent and did not investigate the behavioral mediating pathway. To investigate whether accessibility of food retailers in the residential neighborhood is associated with T2D incidence in four Dutch prospective cohorts, and whether this is mediated by diet quality. In this prospective multi-cohort study we included four Dutch cohort studies (ntotal=10,249). Nearest distances from all participants' home to supermarkets, fast-food outlets and green grocers were calculated at baseline (2004-2012). Incidence of T2D during follow-up was assessed with cohort-specific measures. T2D incidence ratios (IR) adjusted for demographics, lifestyle and environmental factors were estimated using Poisson regression in each cohort, and results were pooled across cohorts using a random-effects model. In two cohorts (n=7,549), mediation by adherence to the Dutch Healthy Diet index 2015 (DHD15-index, range 0-13) was investigated using linear and Poisson regression analyses. Over a mean follow-up of 7.5 years, 569 (5.6%) participants developed T2D. Mean(SD) age in the cohorts ranged from 41.1(12.9) to 67.4(6.8) years. No associations were observed between accessibility of different food retailers and T2D incidence (βsupermarket:0.02(-0.01;0.06), βfast-food:-0.01(-0.04;0.03), βgreen grocer:0.01(-0.05;0.07)). Mediation analyses indicated that every 100 meter living further from a supermarket or green grocer was associated with lower adherence to DHD15 (βsupermarket=-0.1 (95%CI:-0.3;0.0), βgreen grocer=-0.1 (95%CI:-0.1;0.0)), whereas living further away from fast-food associated with higher adherence (βfast-food=0.1 (95%CI: 0.0;0.2)). Higher adherence to DHD15 was associated with lower T2D incidence (IR=0.93 (95%CI: 0.88;0.99)). Spatial accessibility of food retailers was not associated with risk of T2D. Nevertheless, consistent associations in hypothesized pathways were observed, such that spatial accessibility to healthier food retailers was associated with higher diet quality and spatial accessibility of unhealthier retailers with lower diet quality. Higher diet quality, in turn, was associated with lower T2D risk.