Introduction
Produce Prescription programs (PRx) are a type of Food is Medicine intervention offering prescriptions for free or reduced-cost fresh fruits and vegetables (FVs) to patients with diet-related chronic disease and food insecurity. PRx often operate as partnerships between healthcare systems and food system organizations (e.g., farmers markets, food banks). PRx have consistently been shown to improve FV intake & food security status (Hager et al., 2024).
There is less consistency in findings about PRx impacts on biomarkers such as glycosylated hemoglobin (HbA1c) and blood pressure (Hager et al., 2024; Hager et al., 2023a; Hager et al., 2023b). Many PRx operators face challenges in evaluating participants' biomarker data: biomarker data can be expensive for PRx operators to collect in terms of staff time, expertise, equipment, and burden on program participants (Stotz et al., 2025; Long et al., 2024). In contrast, EHR data provides the opportunity to analyze real world data that is already collected by health care providers and used to inform clinical care.
The present study leverages EHR data to evaluate PRx impacts on biomarkers across programs run by different program operators, with the goal of strengthening the generalizability of findings. The aim of this study was to examine changes in HbA1c, systolic blood pressure (SBP), and diastolic blood pressure (DBP) among participants in four distinct PRx programs across nine clinics. These two outcomes were selected by consensus among the four PRx teams as biomarkers of interest because: 1) their PRx programs focus on management of cardiometabolic disease, and changes in HbA1c and blood pressure are indicative of changes in participants' risk for important health complications; 2) HbA1c and blood pressure are responsive to changes in dietary intake, and increased intake of fruits and vegetables is a goal of PRx programs; and 3) HbA1c and blood pressure are often measured in clinical settings and are likely to be included in participants' EHR.
Methods
We invited 12 PRx programs to participate in this study and five shared data (one additional program shared data but is not included here because it focused on a maternal health population rather than cardiometabolic conditions). Of the eight PRx who declined to participate, most reported that they would not have data in time for the analysis due to program timing (e.g., they were early in their program's enrollment phase) or due to lack of staff time for data extraction or executing data sharing requests. Of the four PRx who shared data, most were either already planning their own EHR data analysis and/or were motivated by the support this study provided in analyzing their individual programs' EHR data.
This was a multi-site, single cohort study. Healthcare partners in four PRx shared patient-level, HIPAA-compliant limited datasets which included lab values collected during baseline (up to one year before enrollment) and follow up (between one and nine months after enrollment). All four PRx programs provided HbA1c data, measured at baseline and follow up.
One program measured SBP and DBP at baseline and follow up. Data were pooled across clinics.
The four PRx programs enrolled participants experiencing food insecurity and/or diet-related chronic conditions but were otherwise heterogeneous (Table 1). Participants were enrolled between 2023 and 2024. These programs were in the Northeastern and Western United States United States Department of Agriculture (USDA) National Institute of Food and Agriculture (NIFA) regions.
All four PRx were funded by the USDA Gus Schumacher Nutrition Incentive Program (GusNIP).Table 1Summary of produce prescription program information for the four participating sites in the Northeast and Western United States (2023–2024).Table 1SiteUnited States Department of Agriculture National Institute of Food and Agriculture RegionNumber of Electronic Health Records collectedPopulation of interestProduce Prescription Program modelIntervention duration(A)Northeast100Participants were adults enrolled in Medicare, Medicaid, or Veteran's Affairs insurance and screened positive for food insecurity$30–60 in coupons to spend at farmers markets and grocery stores (varies by household size) plus nutrition education and wrap around services. Food also provided directly at the clinic after a positive screening.6 months(B)Northeast224Participants screened positive for food insecurity and diet-related chronic disease$20 voucher after completing weekly nutrition education lesson12 weeks(C)West2065Participants were adults enrolled in MediCal (i.e., Medicaid), screened positive for household food insecurity in the past 12 months, and had a diagnosis of type 2 diabetes or pre-diabetes$40 per month loaded onto an electronic benefit card redeemable for fresh produce purchases at participating grocery stores6 months(D)West192Participants were enrolled in Medicaid or Supplemental Nutrition Assistance Program, had low income, nutrition insecurity and a diet-related chronic conditionWeekly visit to a farmer's market to pick up $35 worth of tokens for fruits and vegetables plus nutrition education4 months
Baseline HbA1c values were categorized using American Diabetes Association Blood Sugar Control Guidelines (American Diabetes Association, 2020). Baseline blood pressure values were categorized using American Heart Association Guidelines (American Heart Association, 2024); the hypertensive crisis category was excluded from analyses due to small sample size (n = 21). The outcome variables of HbA1c, SBP, and DBP were normally distributed.
We examined patterns in attrition using a 2-sample t-test to compare differences in mean HbA1c, SBP, and DBP among 2218 participants with only baseline data and 1489 participants with matched baseline and follow-up data. We found no differences (all p > 0.05). Paired t-tests compared changes between matched baseline and follow-up values.
Pearson's Chi-Square test examined changes in blood sugar and hypertension categories before and after participation. Analyses were completed using Stata (Release 18, Statacorp LLC), and p < 0.05 was used for statistical significance.
This program evaluation was determined not human subjects research by the University of Nebraska Medical Center Institutional Review Board.
Results
Table 2 presents HbA1c, SBP and DBP values. Mean HbA1c was 8.2% (95% CI, 8.1, 8.3) at baseline and 7.7% (95% CI, 7.6, 7.8) at follow up. For all participants, a clinically and statistically significant decline in HbA1c was observed (mean = −0.5; 95% CI: −0.6, −0.5).
Participants with >9% baseline HbA1c experienced the largest decreases (mean = −1.9 units, 95% CI: −2.06, −1.67).Table 2Distribution of Hemoglobin A1c and blood pressure (systolic and diastolic) for adults in the overall baseline sample (n = 2218), and the matched baseline and follow-up adult samples (n = 1489) participating in four PRx sites in the northeast and western United States (2023–2024). Changes in baseline and follow-up Hemoglobin A1c and blood pressure for the matched adult sample participating in four PRx sites in the northeast and western United States (2023–2024).aTable 2Overall Baseline SampleMatched SampleBaselineFollow- upChangePaired T-testBiomarker Measurenmean (SD)nmean (SD)nmean (SD)Biomarker Measuremean (SD)T statisticcHemoglobin A1c (%)22188.28 (2.26)14898.23 (2.21)14897.68 (1.79)Hemoglobin A1c (%)−0.54 (1.64)−12.72 **Blood Pressure (mmHg)Blood Pressure (mmHg)Systolic1694130.8 (18.6)931130.9 (17.33)931129.2 (16.91)Systolic−1.77 (17.80)−3.03**Diastolic169478.8 (10.49)93178.8 (10.08)93176.77 (9.91)Diastolic−2.03 (10.38)−5.98**Overall Baseline SampleMatched SampleBaselineFollow- upX2 TestChangePaired T-testn%n%n%Test StatisticbBiomarker Measuremean (SD)T statisticcBlood Sugar Category861.1**Less than 7%80236.2%54336.5%63242.4%0.1 (0.7)4.80**7% to 9%73032.9%50333.8%58139.0%−0.1 (1.1)−2.25*Greater than 9%68630.9%44329.8%27618.5%−1.9 (2.1)−18.58**Hypertension Category151.9**Healthy (<120 / <80 mmHg))38923.3%20822.7%25227.5%HealthySystolic10.1 (14.8)9.87**Diastolic3.6 (9.6)5.39**Elevated (120–129/ <80 mmHg)24914.9%13414.6%15016.3%ElevatedSystolic1.1 (13.6)0.35Diastolic1.2 (9.6)1.41Stage 1 Hypertension (130–139 / 80–89 mmHg)50830.4%27530.0%28631.2%Stage 1 HypertensionSystolic−1.6 (15.0)0.08Diastolic−3.5 (9.3)−6.23**Stage 2 Hypertension (140 / 90 mmHg)52731.5%30132.8%23025.1%Stage 2 HypertensionSystolic−10.7 (17.9)−10.46**Diastolic−5.7 (10.0)−9.99**a9 clinic sites contributed data to HbA1c and 3 sites contributed data to blood pressure. P < 0.05 used as the cutoff for statistical significance.bP values from Pearson's Chi square test.cP values from paired t-test *p < 0.05 **p < 0.01.
0.05 used as the cutoff for statistical significance.
At baseline, the mean blood pressure was 130.95 mmHg (95% CI, 129.84, 132.07) /78.80 (95% CI, 78.15, 79.45). Both SBP (mean = −1.77; 95% CI: −2.91, −0.62) and DBP (mean = −2.03; 95% CI: −2.70, −1.37) decreased significantly. Decreases in blood pressure were most substantial for participants in the highest hypertension category at baseline.
Participants with stage one hypertension experienced decreases in SBP (mean = −1.57; 95% CI: −3.35, 0.21) and DBP (mean = −3.47; 95% CI: −4.57, −2.37). Changes experienced by participants in the baseline stage two hypertension category were even more substantial; decreases in SBP (mean = −10.68; 95% CI: −12.69, −8.67) and DBP (mean = −5.70; 95% CI: −6.83, −4.58) were observed.
Discussion
This analysis found improvements in HbA1c and blood pressure following PRx participation, particularly for those with baseline values in poorer control. The magnitude of HbA1c improvement observed is similar to assumptions used in a PRx microsimulation study, which found that if health and diet improvements (including HbA1c decrease of 0.6 units) persisted over time and at scale, they would result in health care savings of approximately $40 billion (Wang et al., 2023). Decreases in SBP of five- and ten-mmHg—like those experienced by the baseline stage two hypertension category in the present study—are associated with 10%–20% reductions in cardiovascular disease events, independent of comorbidities such as cardiovascular disease (Ettehad et al., 2016).
This study is limited by lack of a control group, although participants were compared to themselves at baseline. While this approach controls for inter-individual variability, it does not control for improvements due to the passage of time (regression to the mean) or to other healthcare received by the participants (e.g., medication). Another potential limitation is projects that were able to provide data for this study may differ from those that were unable to participate.
The four participating programs were offered a small stipend to defray the time spent pulling the data needed for this project. An additional limitation is that the majority of the study sample was from a single PRx program in California, which may not be generalizable to other geographic locations across the United States. A further limitation is that dates of HbA1c and blood pressure measurements were masked in the EHR data shared with the study team to avoid sharing protected health information.
Future studies could use perturbation of dates in order to allow more precise summary of the elapsed time between initial and follow-up biomarker measurements (U.S. Department of Health and Human Services, Office for Civil Rights. Guidance on De-identification of Protected Health Information.
July 15, 2010, 2026).
Conclusion
These present study's findings provide needed estimates of biomarker improvement during participation in heterogeneous real-world PRx using EHR data. These findings, and their reliance on EHR data, support the integration of PRx into the current medical system for patient health monitoring and care management. These findings reinforce the generalizable potential for positive health impact from improved access to fresh FVs.
These results indicate that increasing patients' access to fresh fruits and vegetables through PRx programs can lead to reduced chronic disease burden through improvements in blood pressure and hemoglobin A1c. These results add to the evidence base for PRx, which already included consistent positive effects on food security and fruit and vegetable intake. Further real-world research (e.g., a pragmatic randomized controlled trial) is needed to estimate the effects of PRx relative to standard care.
CRediT authorship contribution statement
Victoria A. Zigmont: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualization. James P.
Marriott: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualization. Elise Mitchell: Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualization. Hilary Seligman: Writing – review & editing, Writing – original draft, Methodology, Investigation, Conceptualization.
Amy L. Yaroch: Writing – review & editing, Writing – original draft, Methodology, Investigation, Funding acquisition, Conceptualization. Christina Nadolne: Writing – review & editing, Writing – original draft, Investigation, Funding acquisition, Data curation.
Julia Caldwell: Writing – review & editing, Writing – original draft, Investigation, Funding acquisition, Data curation. Hester Parker: Writing – review & editing, Writing – original draft, Investigation, Funding acquisition, Data curation. Christopher R.
Long: Writing – review & editing, Writing – original draft, Supervision, Methodology, Conceptualization.
Funding
The GusNIP Nutrition Incentive Program Training, Technical Assistance, Evaluation, and Information Center is funded through a cooperative agreement and is supported by GusNIP grant no. 2019–70030-30415/project accession no. 1020863, grant no.
2023–70435-38766/project accession no. 1029638, and grant no. 2023–70414-40461/project accession no.
1031111 from the USDA NIFA.
This project was funded in part by the USDA, NIFA, GusNIP (grant no. 2020–70030-33128) and Gus Schumacher COVID Relief and Response (grant no. 2021–70034-34982) and the USDA NIFA, GusNIP PPR project 2022–70424-38553 awarded to Everyone's Harvest and the USDA NIFA, GusNIP ARPA PPR project 2022–70424-38544 awarded to Cornell Cooperative Extension Suffolk.
Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Hilary Seligman reports a relationship with The Aspen Institute that includes: non-financial support. Hilary Seligman reports a relationship with American Heart Association Inc that includes: funding grants. Hilary Seligman reports a relationship with National Institutes of Health that includes: funding grants.
Hilary Seligmna reports a relationship with Centers for Disease Control and Prevention that includes: funding grants. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.