A three-arm pilot RCT of 210 low-income patients in a Dallas safety-net healthcare system found that a Food is Medicine (FIM) intervention — comparing usual pantry services, medically tailored groceries (MTG) alone, and MTG plus food resource coaching — achieved 81.4% completion over four months. Notably, Spanish-speaking eligible individuals enrolled at a significantly higher rate (60% vs. 44% English speakers, p=0.004), and food selections became measurably more nutritious across all three groups regardless of intervention arm.
Food insecurity is increasingly recognized as a clinical risk factor, and FIM programs have gained policy momentum — including provisional Medicaid coverage in some U.S. states — yet rigorous adherence data from randomized designs remain scarce. This trial fills a meaningful gap by demonstrating that even a relatively light-touch medically tailored grocery model can sustain engagement in populations historically difficult to retain in dietary interventions.
The finding that nutritional quality improved across all arms, including the control group receiving usual pantry services, complicates clean causal attribution to the MTG components specifically — a classic contamination or attention effect. The pilot scale (210 participants) limits statistical power to detect differential effects between arms. Still, the high retention rate challenges the assumption that food-insecure populations cannot sustain research participation. This is confirmatory and encouraging rather than paradigm-shifting, but it meaningfully strengthens the feasibility case for larger definitive trials linking FIM adherence to hard chronic disease outcomes.