Randomized trials
Can estimate the effect of a defined intervention versus a comparator in the studied population and period—not automatic transfer to every setting.
Evidence and operating brief · Public revision 2.0
Food as Medicine and the Future of Proactive Healthcare
Why nutrition is a test of whether healthcare can move upstream without outrunning the evidence.
The thesis
Food is not the whole answer—and that is the point.
The category tests whether healthcare can align evidence, eligibility, workflow, payment, access, safeguards, and measurement around upstream intervention.
7selected studies with decision boundaries
9operating layers
8institutional scorecard domains
Read evidence by design
Different evidence answers different questions.
The report keeps randomized, observational, modeled, program-evaluation, and policy evidence separate so that promise is not confused with observed performance.
Microsimulation
Can model outcomes under explicit assumptions about uptake, cost, risk, and time—not observed results or guaranteed savings.
Program and policy evidence
Can show implementation, current requirements, encouragement, funding, or testing—but not clinical effectiveness without the right design.
Material correction: current HHS materials describe a voluntary hospital-food pledge. CMS separately addresses existing nutrition-service obligations. The prior blanket funding-mandate claim was removed.
What the evidence shows
Mixed results are a design signal.
What supports action
- Well-defined intensive lifestyle programs can reduce risk in selected high-risk populations.
- Targeted and culturally aligned interventions can perform differently from subsidy-only designs.
- Nutrition programs can be tested through bounded, clinically integrated pathways.
- Null and negative findings can improve eligibility, intensity, workflow, adherence, and stop rules.
What the evidence does not justify
- Representing Food as Medicine as one uniform intervention.
- Treating burden statistics as proof that a particular program works.
- Presenting modeled savings as observed outcomes.
- Positioning food as an automatic substitute for indicated medication, procedures, emergency care, or individualized clinical judgment.
The operating model
Build the pathway, not the slogan.
Clinical usefulness depends on who receives what, through which workflow, with which supports, safeguards, outcomes, and adaptation or termination criteria.
EvidenceEligible populationCare planAccess and deliveryClinical workflowSupport and adherenceOutcomesLearning
Inside the report
An institutional decision brief.
The report uses current federal sources, selected research, explicit limitations, and a practical release scorecard.
Burden—with scope
Current federal measures and the exact conclusions they do and do not support.
Evidence by design
Five evidence classes and the boundaries of each.
Seven selected studies
Positive, modeled, mixed, and null results with decision boundaries.
Nine operating layers
Evidence, eligibility, ownership, payment, access, support, data, governance, and learning.
Scorecard and policy context
Eight release domains plus current, scoped institutional context.
A 90-day test and closing position
A bounded implementation method with continue, adapt, pause, and stop decisions.
Use the brief
Design one credible institutional test.
Select one population and one intervention, define the evidence and safety boundary, map the workflow, choose a comparator, and decide in advance what would justify continuing, adapting, pausing, or stopping.