A “produce prescription” is a promising Food‑is‑Medicine approach that provides direct resources to help families afford healthy foods. These programs are growing in popularity among health systems, payers, and public health agencies, yet there is still limited evidence on how well they improve outcomes for people with type 2 diabetes and other diet‑related diseases.
Duke researchers set out to help fill that gap by studying whether a 12‑month, $80‑per‑month produce subsidy could improve health ‑ related outcomes for patients with type 2 diabetes who are at risk for food insecurity. Although they found no improvements in hemoglobin A1c, body mass index, blood pressure, or other diabetes‑related markers, the team says the findings offer valuable guidance for designing future produce prescription programs that are more targeted and effective.
Examining food access and diabetes
Published in JAMA Internal Medicine in February, the study examined whether the produce subsidy could improve health outcomes for patients with type 2 diabetes who may struggle to afford nutritious food. Food insecurity is associated with a higher risk of high glucose and a higher risk of low glucose reactions. Produce prescriptions are seen as a promising way to improve outcomes.
“Not being able to access sufficient nutritious food can really shape diabetes outcomes,” said Connor Drake, PhD, MPA, co‑principal investigator and assistant professor in population health sciences at Duke. “There’s an enormous amount of literature on how food insecurity and diabetes are a dangerous and toxic combination.”
While the study results were disappointing, Drake and co‑principal investigator Susan Spratt, MD, professor of medicine and professor in family medicine and community health, said the findings provide important lessons about how Food‑is‑Medicine programs can be designed more effectively. These include produce prescriptions and medically tailored meals, which can be valuable complements to traditional, clinical diabetes management.
Because there has been limited randomized controlled trial data on produce prescriptions, Drake and Spratt designed a large pragmatic trial to rigorously test how such programs — specifically, the Eat Well program created by the Durham nonprofit organization, Reinvestment Partners — perform in real clinical settings. With 2,155 participants, it was the largest produce prescription randomized clinical trial of its kind.
As part of the Eat Well program, participants received a preloaded $80 debit card for up to 12 months to purchase eligible items, including fruits, vegetables, and legumes. Unlike some programs, this trial did not include additional coaching or nutrition education. Participants were randomized either to receive the debit card or to continue with usual care. Ultimately, the team found no meaningful difference in A1c or in health care use between the two groups. Only about 30% of participants used at least 80% of the monthly funds.
Why didn’t the program work?
The results raised a key question for the research team: Why didn’t the program lead to better outcomes? Drake said the findings highlight the value of null trials, which can reveal important program design and implementation issues that need to be addressed in future studies.
One challenge involved how the team identified patients who might benefit. They used simple indicators such as Medicaid coverage and living in areas with fewer resources to identify people likely to be food insecure. But these indicators didn’t always reflect actual food insecurity.
“The reality is that many people on Medicaid, despite on average having lower incomes, are not necessarily food insecure,” Drake said. One lesson he and the other researchers learned is that it is important to identify the households that face cost-related barriers to healthy eating to maximize the potential benefit of a produce prescription subsidy like the one tested.
Geography was another consideration. Durham is changing rapidly, and neighborhoods that were once economically stressed may no longer fit that profile, Drake said. That means geographic indicators, like the “Area Deprivation Index” used in the study, alone may not reliably identify households facing food insecurity.
Another factor may have been the lack of a behavioral component. Some produce prescription programs include nutrition education, coaching, or visits with a registered dietitian. While not all successful programs include these components, the researchers said the trial raises new questions about when additional support may be necessary.
“If I had it to do over, I would collect more exact data on changes in diabetes and anti-hypertension medication doses,” Spratt said. “For all we know, patients who weren’t randomized to the produce prescription benefit had additional medications added or doses raised. Conversely, patients who did have access to the prescription benefit may have medications stopped or doses lowered.”
In future studies, Spratt said she would like to either randomize or provide diabetes self-management education. “We are planning to look at what types of produce participants bought,” she said. “If more fruit was bought than green vegetables that may have been a problem.”
The team also considered program engagement. Although some people did not use the full benefit, the researchers still found no health improvements even among the most engaged participants.
Drake noted that further reflection is needed on whether health systems are the best place for produce prescriptions, or whether these efforts might be more effective if paired with or embedded into existing nutrition assistance programs such as the Supplemental Nutrition Assistance Program (SNAP). “Hopefully this is going to prompt a lot of reflection on how to best design and implement these programs and where they should exist,” he said.
What comes next
Going forward, Drake and his colleagues want to better identify which patients are most likely to benefit from different types of food‑based support — whether that’s a debit card, medically tailored groceries, or more intensive meal delivery services.
The team is interviewing study participants to better understand what they found helpful and what barriers they faced. “What we're hearing is people love these types of programs and find them to complement existing clinical care,” Drake said. “Finding the right way to implement them for the patients and families that stand to benefit the most is an important part of the future of this work.”
Drake is also leading two additional randomized clinical trials informed by lessons from this study. One focuses on congestive heart failure patients and includes produce prescriptions along with coaching, stronger behavioral supports, and screening for food insecurity before enrollment. The other is a large trial for veterans in the Veterans Health Administration that evaluates produce prescriptions accompanied with referrals to dietitian and other educational and diabetes management services that already exist within the health care system. These trials incorporate the specific lessons from Eat Well, including direct screening for food insecurity risk as a requirement for enrollment and more robust education and behavior change support.
Drake stressed that the study does not change the strong evidence showing that nutrition is essential to managing type 2 diabetes. Instead, it clarifies how produce prescription programs should be structured to truly support patients.
“This does not change what you've heard from your doctor — that nutrition is very important,” he said. “The challenge for researchers and practitioners alike will be to identify how to best implement produce prescriptions as a complement to existing diabetes care activities.”