More than half of all U.S. adults have at least one preventable chronic condition, most of them tied to what we eat. That number has not budged in decades. Not for lack of good intentions. Not for lack of policy. The federal government currently runs 200 separate diet-related health efforts spread across 21 agencies. Two hundred. And the chronic disease rates keep climbing.

Now the Food Is Medicine movement is getting its turn. The idea is elegant and, honestly, overdue: treat food as a clinical intervention, prescribe produce like you prescribe statins, and watch outcomes improve. The American Heart Association has invested over $9.4 million in 23 pilot trials. The Rockefeller Foundation put in $100 million in 2024 alone. States are running Medicaid waivers, farmers markets are accepting vouchers, and doctors are writing prescriptions for broccoli instead of just metformin.

Good. All of that is good. And it is still not enough.

Because here is the uncomfortable part that keeps showing up in the data: dropping healthy food on someone's doorstep, without any support for the human behaviors that determine what actually happens to that food, produces modest results at best. A systematic review of produce prescription programs found less consistent evidence for improvements in fruit and vegetable intake and cardiovascular risk factors than for the more basic outcome of food security. Which means we are getting better at measuring whether people have food. We are not consistently getting better at changing what people do with it.

The Box That Doesn't Change the Habit

Consider the numbers from a meta-analysis of 13 U.S.-based produce prescription programs. On average, participants increased fruit and vegetable consumption by 0.8 servings per day, dropped BMI by 0.6 points, and among diabetics, reduced HbA1c by 0.8 points. Those are real gains. But a diabetic with an HbA1c of 9.5 still has an HbA1c of 8.7 after the program ends. Still diabetic. Still at serious risk. And those are the programs that measured outcomes at all. One New York State assessment found that a lack of program evaluation due to capacity and staffing issues was one of the most common challenges. We don't even have clean data on half of these programs.

What we do have is a Stanford research finding that cuts straight to the point. In the Recipe4Health program, patients who received weekly produce deliveries plus group health coaching did better than those who received produce alone. The lead researcher said it plainly: "Health coaching is the scaffolding that patients need to take that produce delivery and translate it into healthy lifestyle behaviors." Scaffolding. Not a bonus feature. Not a premium tier. The structural support without which the whole thing doesn't stand up.

The American Heart Association, to its credit, is now explicitly funding pilots that incorporate behavioral science and implementation science into Food Is Medicine programs. The agency acknowledges that the gaps in the existing literature include, most critically, more definitive studies on the impact on clinical outcomes. That's a polite way of saying: we have enthusiasm and early signals, but we don't yet know how much of this actually sticks in real people's bodies over real time.

What Changes When Someone Is Actually in the Room

Here's what the programs that actually work have in common. Delaware's Feeding Families program: weekly produce deliveries plus bi-weekly check-ins with community health workers plus monthly sessions with a registered dietitian. Blue Cross Blue Shield of North Carolina piloted food delivery alongside health coaching and tracked the downstream effects on costs of care. The South Bronx program funded by Elevance Health sent community health workers to provide coaching alongside the medically tailored meals. In each case, the food was the input. The behavior change support was the mechanism that turned the input into an output.

A study from the Functional Medicine Coaching Academy found that people on a structured dietary plan who worked with a health coach showed better dietary compliance and better health outcomes than those who tried to follow the same plan alone. This should surprise nobody who has ever tried to change a habit while stressed, sleep-deprived, working two jobs, or managing chronic pain. Which is to say: it should surprise nobody at all, because that is the exact profile of most people in these programs.

The AHA's own presidential advisory flagged it directly: there are "significant questions about the role of households, incremental effectiveness of including coaching or other behavior change strategies." Significant questions. After decades of research on habit formation and behavioral change, we are still treating the coaching component as a research variable rather than a design requirement.

This is the part that frustrates me. We know how behavior change works. You need a reason, a plan, and a person in your corner who notices when you go off course and helps you get back without shame. That's it. That's the whole model. It doesn't require a wearable. It doesn't require a 14-step morning routine. It requires consistent human contact from someone who gives a damn about your specific situation.

Patients in behavioral pharmacy programs, where they met in person with peers and coaches, showed significant improvements in fruit and vegetable intake, blood pressure, depression, isolation, and physical activity. Multiple outcomes, one intervention. Because when you change the conditions around someone's behavior, you don't just change one behavior.

Food Is Medicine is a genuinely important movement, and the evidence base is real. But the programs pouring money into supply chains and voucher systems, while treating behavioral coaching as an optional add-on, are solving the easier problem. Access matters. So does the conversation that happens after the box arrives. You do not need a protocol. You need consistency. And consistency is a human skill that humans help build.