Food-as-Medicine Contracts Should Buy Nutritional Persistence, Not Enrollment

01 / THE ARGUMENT

The next contract needs a narrower promise

Food-as-medicine has reached a point where a broad category narrative can obscure the decision a purchaser actually faces. An employer, health plan or accountable provider needs to know which patients will receive which intervention, whether they can use it, and what outcome the buyer is financing. My proposed unit of purchase is a completed period of appropriate nutritional support with documented follow-up. A signed enrollment form is only the beginning of that period.

The September 2026 MUTTON-HF publication supplies important, specific evidence: among 206 patients with heart failure in rural Navajo Nation, a culturally and medically tailored intervention reduced the 90-day composite of hospitalization or emergency-department use relative to usual dietary advice, with a relative risk of 0.72 and a 95% confidence interval of 0.54–0.96. This supports that intervention in that population; it does not establish the return for every nutrition benefit. [1]

02 / THE ARGUMENT

Evidence heterogeneity should change procurement

Contrasting results sharpen the commercial question. Doyle and colleagues' randomized study of an intensive grocery and care-support program found no significant between-group HbA1c improvement at six months, despite greater preventive-care engagement. The result cautions against crediting a vendor for all improvement observed after enrollment. [2] The 2026 FAME-F trial also found no significant advantage for household rather than individual meal provision on its primary diet-quality measure, or for dedicated drivers rather than commercial shipping on its loneliness measure. Its small size and wide intervals leave meaningful uncertainty. [3]

My procurement inference is to price the intervention's components separately before buying a bundle. Meal quantity, household support, clinical nutrition expertise, delivery logistics and human contact may solve different problems. A buyer should require a reason for each component, then test the components that drive its own expense and patient experience. Neither a positive trial nor a null finding warrants treating all implementations as interchangeable.

03 / THE ARGUMENT

Persistence is an operating pathway

The proposed pathway begins with clinical eligibility and a practical nutrition assessment: storage, preparation, household sharing, preferences and the patient's ability to receive deliveries. A dietitian-approved plan then connects the vendor to the treating team. Delivery confirmation should distinguish receipt from usability; a repeated refusal, refrigeration problem or mismatch with the care plan should trigger a person with authority to change the service. These are operating design choices rather than new scientific claims.

I would track a nutrition episode from authorization through first usable delivery, continued participation, review and transition. The handoff at the end matters commercially: a temporary intervention needs an agreed destination, whether ordinary food access, continued covered support or a different clinical service. Federal Food Is Medicine principles emphasize culturally appropriate design, community partnerships and local capacity. Those principles inform program design but do not by themselves create coverage or a payment entitlement. [4]

04 / THE ARGUMENT

Build economics around incremental value

For an illustrative buyer model, expected net value equals eligible episodes multiplied by participation, incremental outcome benefit and the buyer's realizable financial value, less food, logistics, clinical follow-up, implementation and evaluation costs. Each input must have a documented definition and a sensitivity range. This is a proposed planning equation, not a published savings estimate. Patient food security and quality of life deserve their own measures even when a purchaser cannot monetize them.

The evaluation should begin before outreach. Prefer random allocation or a defensible phased rollout when feasible; otherwise predefine a comparison cohort, eligibility dates, baseline risk and concurrent medication changes. Analyze everyone assigned or offered the intervention as well as participants, and report missing follow-up. Claims-based savings need adequate maturation and an explicit attribution method. A fall in utilization among selected completers is insufficient grounds for a shared-savings invoice.

05 / THE ARGUMENT

Use the contract to preserve learning

A proposed contract can pay separately for validated service delivery and for demonstrated outcomes, with a reconciliation process that both parties can reproduce. Define allowable substitutions, clinical escalation, patient complaints, data completeness and continuity after a delivery failure. Pay community partners on time rather than asking them to carry a health plan's working capital. Where outcomes remain uncertain, a limited initial cohort with transparent evaluation is a more defensible commitment than an enterprise-wide savings guarantee.

The executive decision is therefore whether a particular nutritional pathway merits expansion, redesign or termination. An early win may be reliable access and lower food insecurity; a later finding may establish a clinical benefit. Keep those findings separate. Scale when the intervention fits the population, patients continue using it, the care team can act on exceptions and the economics survive conservative assumptions. That standard protects both the scientific proposition and the credibility of the business case.

FROM EVIDENCE TO ALLOCATION

The operating and investment case

Proposed design by Azis R. Dabas. These decisions and evaluation criteria are not outcomes established by the cited studies.

Decision: Commission a bounded, clinically defined nutrition episode with an evaluation plan before expanding a food benefit across a covered population.
Accountable owner: Health-plan population-health executive with a medical director, registered dietitian lead, finance partner and community-provider representative.

The delivery sequence

Define eligibility, covered services and comparator before outreach
Assess clinical and practical nutrition requirements
Confirm first usable delivery and dietitian-approved fit
Monitor continued participation and route exceptions to an accountable person
Review clinical and patient-reported outcomes
Reconcile service payments separately from evaluated outcome value

The economics

Illustrative planning model: eligible episodes × participation × incremental outcome benefit × realizable buyer value, minus the complete food, delivery, clinical, implementation and evaluation cost. Do not use an external trial's relative risk as the local savings assumption without validation.

The measures that govern expansion

Time to first usable delivery
Usable deliveries as a share of scheduled deliveries
Participation persistence and reasons for withdrawal
Food-security and patient-experience change
Predefined clinical outcome relative to comparator
Total episode cost and conservatively attributed net value

Stop or redesign when

Pause expansion if food is regularly unusable, clinical exceptions remain unresolved, follow-up data cannot support evaluation, or the business case depends on attributing all pre/post improvement to the intervention.

THE EVIDENCE LEDGER

What supports the argument

Study findings, policy requirements and market signals support different claims. Their boundaries remain visible.

[1] peer-reviewed · September 1, 2026

An Indigenous Food Is Medicine Intervention: The MUTTON-HF Randomized Clinical Trial

Design or status: Pragmatic open-label randomized trial at two Indian Health Service sites; 206 participants; 90-day primary endpoint.
Verified finding: The intervention reduced the hospitalization/ED composite relative to usual advice (RR 0.72, 95% CI 0.54–0.96).
Boundary: Short follow-up and a specific rural Indigenous heart-failure population limit generalization and do not establish local payer savings.

[2] peer-reviewed · December 26, 2023

Effect of an Intensive Food-as-Medicine Program on Health and Health Care Use: A Randomized Clinical Trial

Design or status: Randomized wait-list trial of patients with diabetes and food insecurity in one rural and one urban health-system clinic.
Verified finding: The adjusted between-group six-month HbA1c difference was −0.10 percentage points (95% CI −0.46 to 0.25); preventive-care engagement increased.
Boundary: A particular intensive grocery and support program; missing laboratory follow-up and its setting limit extrapolation to other nutrition interventions.

[3] peer-reviewed · April 1, 2026

Designing Medically Tailored Meal Interventions: The Food as Medicine for Families (FAME-F) Factorial Randomized Clinical Trial

Design or status: 2×2 factorial randomized comparative-effectiveness trial; 93 New England participants receiving meals for 12 weeks.
Verified finding: No significant differences were detected for the primary diet-quality comparison or loneliness comparison between intervention designs.
Boundary: Small sample, wide confidence intervals and no no-meal control; this is not evidence that meals have no benefit or that delivery methods are equivalent.

[4] policy · September 14, 2024

Our Principles for Food Is Medicine

Design or status: Federal program principles; date is the page's last-reviewed date.
Verified finding: The framework emphasizes culturally appropriate nutrition, cross-sector partnerships and investment in under-resourced community capacity.
Boundary: Program principles are neither coverage authorization nor an estimate of causal clinical or financial impact.

FOLLOW THE SOURCE

Sources and editorial method

Selected evidence was reviewed through October 7, 2026. Numbered references connect claims to their underlying records. Economic mechanisms and business cases are the author’s analysis unless a source is cited. The review is selective; publication dates retain the precision available in the source.

An Indigenous Food Is Medicine Intervention: The MUTTON-HF Randomized Clinical Trial
Lauren A. Eberly, Carmen George, Sharon Sandman, Denee Bex, Karianne Jones, Asia Yazzie, Leah Gray, Larissa Morgan, Ada Tennison, Conor Williams, Matt Chandra, Rebecca Wickre, Bennett Wickre, Mackenzie Bolas, Remi Welbel, Delaney Ignace, Benjamin Feliciano, Stacy Hammer, DezBaa Damon-Mallette, Erica Lindsey, Paula Mora, Aijun Ye, Maricruz Merino, Enrique F. Schisterman, Sonya S. Shin. JAMA Internal Medicine. September 1, 2026. peer-reviewed.
DOI: 10.1001/jamainternmed.2026.2879 · Primary verification record · Consensus paper record
Consensus citation count at retrieval: 2. This dated index count is not a measure of study quality.
Effect of an Intensive Food-as-Medicine Program on Health and Health Care Use: A Randomized Clinical Trial
Joseph Doyle, Marcella Alsan, Nicholas Skelley, Yutong Lu, John Cawley. JAMA Internal Medicine. December 26, 2023. peer-reviewed.
DOI: 10.1001/jamainternmed.2023.6670 · Primary verification record · Consensus paper record
Consensus citation count at retrieval: 51. This dated index count is not a measure of study quality.
Designing Medically Tailored Meal Interventions: The Food as Medicine for Families (FAME-F) Factorial Randomized Clinical Trial
Seth A. Berkowitz, Jacqueline Nguyen, Colleen Dagley, Salma Ali, Myklynn LaPoint, Patricia Knoepp, Jean Terranova. Circulation: Population Health and Outcomes. April 1, 2026. peer-reviewed.
DOI: 10.1161/CIRCOUTCOMES.125.012668 · Primary verification record · Consensus paper record
Consensus citation count at retrieval: 2. This dated index count is not a measure of study quality.
Our Principles for Food Is Medicine
U.S. Department of Health and Human Services, Office of Disease Prevention and Health Promotion. HHS Food Is Medicine initiative. September 14, 2024. policy.

Study authors retain credit for their work. Researcher affiliations and publisher names do not imply affiliation with or endorsement of this analysis.