Medically Tailored Meals and Medicaid Utilization in Massachusetts

Key Takeaways

• During meal receipt, recipients had fewer hospitalizations and emergency department visits than eligible comparators.
• Total healthcare costs were lower during enrollment, while primary care visit rates showed little difference between groups.
• Results were similar across sensitivity analyses, and larger net cost-saving appeared in selected higher-burden subgroups.

In a Massachusetts MassHealth Section 1115 demonstration, medically tailored meal receipt during active enrollment was associated with 31% fewer hospitalizations, according to a Nature Medicine analysis. Meal receipt was also associated with lower emergency department use and lower total healthcare costs among participating Medicaid members. The analysis examined associations during the period when meals were being delivered rather than after services ended, and the cost pattern was limited to the enrollment period.

Researchers evaluated the nutrition benefit within the MassHealth demonstration using linked Medicaid claims, encounter, ACO administrative, and Community Servings enrollment data. They used a propensity score overlap-weighted difference-in-difference design with a 6-month baseline and program period in a primary sample of 1,866 recipients enrolled for more than 90 days and 1,372 comparators. Eligibility required ACO enrollment, age younger than 65 years, food insecurity, and at least one of five health-needs criteria. Community Servings delivered 10 meals plus snacks once weekly, usually for 6 months, with dietitian tailoring, initial counseling, and optional follow-up counseling. The program ran from January 2020 through March 2023 across 11 health systems, and mean meal receipt was 6.7 months.

In adjusted analyses, hospitalization rates were lower during meal receipt, with an aIRR of 0.69 and a 95% CI of 0.58 to 0.82. Emergency department visits also declined, with an aIRR of 0.80 and a 95% CI of 0.72 to 0.89. Total healthcare costs were $3,433 lower, with a 95% CI from -$5,651 to -$1,215, and absolute reductions reached 443 fewer hospitalizations and 403 fewer ED visits per 1,000 treated. The investigators reported that these reductions offset 98% of program costs during enrollment; mean per-person program cost was $3,512, or $125 weekly. Primary care visits changed little, with an aIRR of 0.99 and a 95% CI of 0.97 to 1.01.

The overall pattern was similar across multiple model specifications, a negative control test, and a secondary comparison group drawn from managed care members. Stratified analyses found net cost-saving among participants with cardiovascular disease, chronic kidney disease, depression, diabetes, or high overall comorbidity. A secondary analysis that included all program durations still showed lower hospitalization and emergency department use, although total cost differences were not statistically significant. Larger cost differences were concentrated among participants with greater medical complexity.

The evaluation was observational and nonrandomized, so the associations do not establish causality and residual selection bias remained possible despite weighting methods. Researchers also could not confirm meal adherence because dietary data were unavailable. Follow-up addressed the active enrollment period only because claims adjudication lag limited follow-up, so post-program effects were not assessed. Generalizability may be limited to this Massachusetts demonstration and the specific program delivered through Community Servings. The reductions were observed during active enrollment in the Massachusetts program.