Cash to carrots: Testing food environment mediation of cash transfer effects on children's diet in the Baby's First Years Study.

In the US, low-income children have poorer dietary behaviors than high-income children and suffer disproportionately from nutrition-related chronic diseases. A growing literature examines whether food environments contribute to nutritional inequality, given evidence that low-income families differentially concentrate in neighborhoods characterized by limited access to healthy foods. Here we use a rigorous, randomized controlled trial (RCT) of poverty reduction among 901 low-income families to test whether improvements in the food environment mediate a demonstrated effect of cash transfers on children's diet. We first replicate prior work on the RCT showing a protective effect of treatment-i.e., assignment to a "high-cash" gift group (in which mothers received $333/month), relative to a "low-cash" gift group ($20/month)-on age-2 fruit and vegetable consumption. We then apply inverse odds ratio weighting (IORW) mediation methods to decompose the total effect of cash transfers into direct and indirect effects operating through changes in four food environment measures: 1) food deserts (census tract); 2) number of SNAP outlets (1-mile buffer); 3) Modified Retail Food Environment Index (mRFEI; census tract); and 4) supermarket proximity (straight-line distance). The proportion of the total effect explained by each food environment mediator ranged from 11% (mRFEI: β = 0.019, CI: -0.066, 0.104) to 14% (supermarket proximity: β = 0.025, CI: -0.057, 0.107) with all confidence intervals including zero, providing no evidence of mediation. Results suggest that changes in neighborhood food environments do not explain the effect of a poverty reduction trial on dietary improvements in early childhood.