Medically Tailored Meals Lower A1c in FAME-D Diabetes Trial

TL;DR: Six months of home-delivered medically tailored meals plus telephone lifestyle coaching lowered hemoglobin A1c and systolic blood pressure more than a $40-per-month grocery subsidy in adults with type 2 diabetes and food insecurity.

Answer first: meals outperformed a grocery subsidy

In the Food as Medicine for Diabetes (FAME-D) randomized comparative effectiveness trial, participants assigned to medically tailored meals (MTM) had lower hemoglobin A1c at six months than participants assigned to a food subsidy, according to the abstract published in Diabetes as 1249-OR: Medically Tailored Meals Lower Hemoglobin A1c: The Food as Medicine for Diabetes (FAME-D) Randomized Clinical Trial | Diabetes | American Diabetes Association.

The reported six-month means were 7.74% for MTM versus 8.17% for the subsidy group, a mean difference of -0.42 percentage points, with a 95% confidence interval from -0.82 to -0.03 and p=0.037. Systolic blood pressure was also lower with MTM, at 124.3 mmHg versus 130.8 mmHg, a mean difference of -6.5 mmHg, with a 95% confidence interval from -12.0 to -1.1 and p=0.02.

The investigators concluded that six months of MTM lowered HbA1c and systolic pressure compared with a food subsidy and that medically tailored meals can help improve health outcomes for people living with food insecurity and type 2 diabetes. The trial record cited in the abstract is NCT04828785, and the primary time point was six months from randomization.

Why the comparison mattered

The abstract states that food insecurity leads to worse type 2 diabetes outcomes, but which food insecurity intervention best improves those outcomes is unclear. The objective was therefore to compare the effectiveness of two food insecurity interventions: medically tailored meals and food subsidies.

That head-to-head design distinguishes FAME-D from studies that compare a nutrition benefit with usual care alone. Both study arms received an active food-related intervention for six months. The question was not whether food help matters at all, but whether delivered meals tailored to diabetes diet guidance plus coaching produce larger clinical changes than money earmarked for healthy groceries.

How the two study arms were designed

Adults with type 2 diabetes and food insecurity were randomized to one of two six-month strategies.

One group received MTM consisting of home delivery of 10 meals per week tailored to type 2 diabetes diet guidelines by a registered dietitian-nutritionist, plus a telephone lifestyle intervention adapted from Look AHEAD. The other group received a food subsidy of $40 per month for healthy groceries.

The lifestyle component included six main sessions plus six brief check-ins. According to the abstract, the sessions focused on skills for healthy eating, increased activity, and behavior change, with or without a weight-loss goal according to participant preference.

The primary outcome was hemoglobin A1c. Other outcomes were systolic blood pressure, diastolic blood pressure, and weight.

Who took part in FAME-D

The abstract reports that 194 participants were randomized and that 178 participants, or 92%, had six-month follow-up measurements.

Mean age was 56 years, with a standard deviation of 12 years. A total of 117 participants, or 60%, were women. Fifty-five participants, or 28%, self-identified as Black, and 35 participants, or 18%, self-identified as Hispanic.

At baseline, mean HbA1c was 8.4%, with a standard deviation of 1.7 percentage points. Mean systolic pressure was 132.7 mmHg, with a standard deviation of 19.6 mmHg. Mean diastolic pressure was 80.1 mmHg, with a standard deviation of 11.7 mmHg. Mean weight was 103.5 kg, with a standard deviation of 24.5 kg.

The named authors listed for the presentation were Seth Berkowitz, Katharine A. Ricks, Jean Terranova, Colleen Dagley, Barbara Steiner, Gina Kruse, Thomas C. Keyserling, John B. Buse, Deborah Wexler, and Linda M. Delahanty. Funding reported in the indexed record was NIH/NIDDK grant R01DK125831.

Blood sugar: what changed at six months

The primary result favored MTM. At six months, the MTM group averaged 7.74% HbA1c and the subsidy group averaged 8.17%.

The estimated between-group difference was -0.42 percentage points. The 95% confidence interval extended from -0.82 to -0.03, and the p-value was 0.037, meeting the conventional threshold for statistical significance described in the abstract as significantly lower.

Baseline HbA1c for the full randomized population had averaged 8.4%. The abstract does not report separate baseline means by arm in the text supplied, so the published comparison rests on the six-month values and the adjusted mean difference.

For readers managing diabetes, HbA1c reflects average blood glucose over approximately two to three months in routine clinical use. The trial itself defined HbA1c as its primary outcome and did not, in the abstract text, assign a clinical category to the observed 0.42-point difference.

Blood pressure: systolic improved, diastolic did not

Systolic pressure showed a larger absolute separation. The MTM group averaged 124.3 mmHg at six months, compared with 130.8 mmHg in the subsidy group. The mean difference was -6.5 mmHg, with a 95% confidence interval from -12.0 to -1.1 and p=0.02.

Diastolic pressure moved in the same direction but did not meet the significance criterion used by the authors. Mean diastolic values were 76.1 mmHg for MTM versus 78.8 mmHg for the subsidy group. The mean difference was -2.7 mmHg, with a 95% confidence interval from -5.9 to 0.5 and p=0.10.

Baseline means for the study population were 132.7 mmHg systolic and 80.1 mmHg diastolic. The abstract reports the six-month values as group means rather than as within-person changes.

Weight: no significant between-group difference

Weight also did not differ significantly between arms at six months. Mean weight was 101.6 kg with MTM versus 102.5 kg with the subsidy. The mean difference was -1.0 kg, with a 95% confidence interval from -2.9 to 0.9 and p=0.31.

Mean baseline weight was 103.5 kg, with a standard deviation of 24.5 kg. The lifestyle intervention allowed participants to choose whether to pursue a weight-loss goal, and the abstract describes the coaching focus as healthy eating, increased activity, and behavior change.

The pattern across outcomes was therefore mixed: HbA1c and systolic pressure favored delivered meals, while diastolic pressure and weight showed numerical differences favoring MTM that were not statistically significant.

What the meals included beyond food delivery

The MTM intervention combined two elements: prepared food and behavioral support.

The food element was home delivery of 10 meals per week, tailored to type 2 diabetes diet guidelines by a registered dietitian-nutritionist. The abstract does not list menus, calorie levels, sodium limits, carbohydrate targets, or vendors.

The behavioral element was a telephone lifestyle intervention adapted from Look AHEAD. It comprised six main sessions and six brief check-ins centered on healthy-eating skills, physical activity, and behavior-change strategies. Participants could select a weight-loss goal or focus on other behavioral goals.

The comparator was less intensive in contact and food provision: $40 per month designated for healthy groceries, without the delivered meals or described coaching package.

Context: a grocery stipend alone did not move A1c in another 2026 trial

A second randomized trial presented in the same Diabetes supplement underscores why intervention design may matter. The abstract titled 1250-OR: Efficacy of a Food Is Medicine Program in Low-Income, High-Risk Adults with Type 2 Diabetes: The KP ENRICH Study tested a technology-based grocery prescription program.

That study enrolled adults with Medicaid insurance, type 2 diabetes, and A1c at or above 7.5% in Kaiser Permanente Northern and Southern California from October 2023 through December 2025. It excluded people with advanced kidney disease, those who were institutionalized or pregnant, those with no internet access, those requiring an interpreter, or those with no valid credit or debit card.

The intervention used the Instacart online platform and provided a $100-per-month stipend for healthy foods such as fruits, vegetables, legumes, nuts and seeds, with four free deliveries per month for six months. The control arm received four free Instacart deliveries per month for six months without a stipend. The primary outcome was six-month A1c using an intent-to-treat approach.

Among 1,007 subjects, with 510 in intervention and 506 in control as listed, both arms started with similar characteristics: mean age 53.2 versus 53.1 years, women 67.9% versus 66.6%, and mean entry A1c 9.34% versus 9.21%. Registration to use Instacart was 88.6% versus 50.8%, and mean monthly orders were 1.5 versus 0.2. Follow-up A1c was available for 93.4% of subjects.

Both arms reduced A1c: minus 0.63 percentage points in the intervention group and minus 0.53 percentage points in the control group. After adjustment for baseline A1c and stratification factors, six-month A1c was 8.63% versus 8.65%, a difference of -0.06 percentage points, with a 95% confidence interval from -0.26 to 0.14 and p=0.55. The authors concluded that monthly funds for healthy foods through an online platform did not improve six-month glycemic control versus usual care in that low-income, high-risk population.

Taken together, the two abstracts suggest that simply providing grocery money may not shift A1c, while providing prepared diabetes-tailored meals plus coaching did shift A1c relative to a smaller subsidy in FAME-D. The populations, stipend sizes, delivery models, comparators, and baseline A1c levels differed, so the results do not establish a single dose-response for food spending.

What this means for food-as-medicine planning

For health systems and nutrition programs, FAME-D compares a higher-touch model with a lower-cost subsidy. The higher-touch model delivered food directly to the home on a weekly basis and paired it with telephone coaching. The lower-cost model provided $40 monthly for groceries.

The trial retained 92% of randomized participants for six-month measurements, which supports the feasibility of follow-up in a food-insecure diabetes population. The abstract does not report adherence to meal consumption, grocery purchases, coaching attendance, medication changes, hypoglycemia, or costs.

People with diabetes and food insecurity who are considering meal programs can discuss referral options, eligibility rules, diabetes medications, and blood-pressure monitoring with their clinical team. Programs vary in meal frequency, nutrition tailoring, delivery area, and duration, and the FAME-D abstract describes only the specific combination tested: 10 meals weekly plus the defined telephone program for six months.

Limits of what the abstracts establish

The findings come from conference abstracts published June 7, 2026, not from a full peer-reviewed paper excerpt in the source text. The FAME-D abstract reports group means, differences, confidence intervals, and p-values, but does not include subgroup results, adherence data, adverse events, or longer-term follow-up beyond six months.

The conclusion applies to adults with type 2 diabetes and food insecurity enrolled under trial conditions and to the comparison tested. It does not compare MTM with no food assistance, with a larger grocery benefit, or with in-person counseling. It also does not report whether systolic-pressure or A1c differences persisted after meals ended.

The KP ENRICH comparison likewise applies to its own eligibility criteria and to a $100 monthly online grocery stipend versus deliveries without a stipend. Its result of no significant A1c difference does not contradict FAME-D directly because the interventions and control conditions were different.