Nutrition Security Among Medicaid Patients With Diabetes or Prediabetes After Completing a Produce Prescription Program

Abstract

PURPOSE

To increase access to fresh produce for Medicaid patients with type 2 diabetes or prediabetes, a local health department partnered with 3 Federally Qualified Health Centers to implement a produce prescription program in Los Angeles County. Program participants received $40 per month for 6 months through an electronic debit card to purchase fresh fruits and vegetables at participating grocery stores. This observational study assessed improvements in participants’ nutrition security, defined as the consistent access, availability, and affordability of foods and beverages that promote well-being and prevent disease.

METHODS

Program participants completed a questionnaire that included a 2-item household nutrition security screening tool and a 6-item household food security module at the program’s beginning (baseline) and end (follow-up). Analyses were based on 1,309 participants who completed the questionnaire at both time points.

RESULTS

After completing the program, the proportion of participants who reported being nutrition secure increased by an absolute 15.5% from baseline (from 23.2% to 38.7%). This improvement in nutrition security was statistically significant after adjusting for sociodemographics, baseline food security status, and fruit and vegetable consumption (adjusted odds ratio = 2.95; 95% CI, 2.11-4.13). Despite the nutrition incentive, most participants (85.6% vs 88.4% at baseline, P = .03) still stated healthy foods were “too expensive”—although more of them reported being able to locate a store that had healthy food options after completing the program.

CONCLUSIONS

This “food as medicine” intervention demonstrated short-term improvements in nutrition security among Medicaid patients with diabetes or prediabetes, suggesting it successfully fulfilled a basic social and disease management need of this population.

INTRODUCTION

In recent years, “food as medicine” interventions have emerged as a key health care strategy for preventing, treating, and managing diet-related diseases. The United States spends roughly $1.1 trillion each year to treat chronic, diet-related conditions such as hypertension and type 2 diabetes.¹ The produce prescription program (PPR), a type of food as medicine intervention, provides patients who often live in households experiencing food insecurity and/or diet-related health conditions with vouchers or subsidies to purchase fruits and vegetables (produce). PPRs can increase consumption of produce, support food security, and improve clinical and health outcomes including glycemic control.^2-5 One study estimated that if $32 per month in nutrition incentives were provided to every patient with diabetes in the United States who is experiencing food insecurity, an estimated 250,000 cardiovascular events could be prevented over a lifetime.¹

Understanding the ways in which PPRs can shape a patient’s experience with access and availability to healthy foods is a priority for the US Department of Agriculture (USDA), as well as for other federal agencies, including the National Institutes of Health and the Centers for Disease Control and Prevention. The USDA, for example, has started to monitor nutrition security, defining it as “consistent access, availability, and affordability of foods and beverages that can promote well-being and prevent disease.”⁶ Yet PPR research, to date, has primarily focused on how these programs can improve another indicator, food security.^2,5 Insufficient food security, also known as food insecurity, is defined at 2 levels of severity. In households with low food security, the hardships experienced are primarily reductions in dietary quality and variety. In households with very low food security, the hardships experienced are reduced food intake and skipped meals.⁷ By focusing primarily on food sufficiency and overall dietary quality, several important elements are missed if one is to fully understand the barriers to fruit and vegetable consumption. For example, measures of low and very low food security do not capture barriers to healthy eating such as the affordability of healthy foods, lack of access to grocery stores, limited transportation, insufficient knowledge of preparing healthy foods, and uncertainty surrounding qualification for food assistance programs.⁶

Some have therefore called attention to the importance and distinct construct of nutrition security and its absence.^8,9 Studies using population-based samples have found the prevalence of nutrition insecurity ranges from 18% to 44% and correlates with diabetes and poor mental health.^10,11 Although there is limited agreement on the best way to measure nutrition security,^10,12 emerging work in this area of research and practice recognizes the importance of measuring and monitoring this indicator, as a way to better promote and optimize health.⁸ Our study addressed this research and practice gap by measuring this new indicator, nutrition security, in the context of a food as medicine intervention.

We undertook an observational before-and-after study to assess changes in nutrition and food security among Medicaid patients participating in a PPR at Federally Qualified Health Center (FQHC) clinics in Los Angeles County. We also assessed several barriers to healthy eating, including participants’ perceptions of the price of healthy foods in their community and their ability to identify healthy foods.

METHODS

Overview

The Los Angeles County Department of Public Health partnered with 3 FQHCs to implement a PPR during 2020-2024, subsidized by funds from the USDA National Institute of Food and Agriculture.¹³ The goal of the program was to increase fresh fruit and vegetable purchases among low-income patients. Participating FQHC clinics were located throughout Los Angeles County. Through the PPR, enrolled patients received $40 per month for 6 months to purchase fresh fruits and vegetables with an electronic debit card at participating large chain supermarkets or grocery stores.

Individuals were eligible to participate in this program if they were patients of a participating FQHC, aged 18 years or older, enrolled in Medicaid, had a diagnosis of type 2 diabetes or prediabetes, and screened positive for household food insecurity during a routine medical visit. Recruitment began in 2021 and was conducted by health and diabetes educators at 9 clinics within the 3 FQHC systems. Clinical staff at participating clinics were encouraged to make referrals to the PPR; enrollment in the program did not require a physician’s referral.

All study protocols and materials were reviewed and approved by the institutional review board at the Los Angeles County Department of Public Health.

Program Procedures: Enrollment and Follow-up

At participating clinics, health and diabetes educators managed the enrollment and 6-month follow-up appointments of each patient. From the electronic health record system, clinic staff identified prospective candidates based on the program’s eligibility criteria (age, Medicaid receipt, diabetes or prediabetes, food insecurity) and those with a glycated hemoglobin (HbA_1c) laboratory value in the past 3 months. Clinical staff called patients to recruit them and begin enrollment procedures.

After providing consent, each patient enrollee was scheduled for an on-site clinic enrollment appointment in which their electronic debit card was activated after they completed the study questionnaire (baseline). The questionnaire, detailed below, was self-administered, took an average of 20 minutes to complete, and was available in English and Spanish. Participants then completed the same questionnaire 6 months later at a follow-up appointment (end of the program), either in person or over the telephone.

Data were collected on a rolling basis across all clinics during 2021-2024. A total of 2,374 patients were recruited and enrolled in the program across the 3 FQHCs (1,128 patients at FQHC 1; 838 patients at FQHC 2; and 408 patients at FQHC 3).

Study Variables

The primary study outcome (dependent variable) was household nutrition security, defined by the USDA as “having consistent access, availability, and affordability of foods and beverages that promote well-being and prevent disease.”⁶ The study used a new 2-item nutrition security screener developed by the Los Angeles County Department of Public Health, Kaiser Permanente, and Tufts University.¹⁰ A short preamble was provided to define healthy foods (Supplemental Appendix, question 22). This was followed by 2 questions.

The first question asked, “Thinking about the last 12 months, how hard was it for you or your household to regularly get and eat healthy foods?” Five response options were provided; participants who responded, “not very hard” or “not hard at all” were classified as nutrition secure, whereas those who responded, “very hard,” “hard,” or “somewhat hard” were classified as nutrition insecure. The second question began with a short preamble, “People have different reasons for not eating healthier foods. Please indicate how often the following reasons, if any, were true for you or your household in the last 12 months.” This was followed by 12 reasons (barriers to healthy eating) listed as statements; participants were asked to indicate whether each statement was “never true,” “sometimes true,” or “often true” as a reason for why their household did not eat healthier foods.

The secondary study outcome was food security, which was captured using the 6-item short form food security survey module.¹⁴ Participants were classified as food secure if they scored 0 or 1 on the module and as food insecure if they scored 2 to 6 on the module.

Study covariates included sex, age, race/ethnicity, educational attainment, Supplemental Nutrition Assistance Program (SNAP) participation, household size, and location/characteristics of the FQHC from which they received PPR benefits. Fruit and vegetable consumption was captured using the 10-item Dietary Screener Questionnaire fruit and vegetable module; responses to the questions about fruit and vegetable intake were converted to daily cup equivalent estimates.¹⁵ For analyses, we used daily cups of fruits and vegetables at baseline and change in daily cups of fruits and vegetables during the 6 months of observation.

Analysis

A total of 2,374 patients were enrolled in the PPR, of whom 1,309 (55%) completed the questionnaire at both baseline and follow-up and were included in main analyses. Those with missing data on a given study variable were excluded from that analysis. Because the nutrition security questions were added to the questionnaire 3 months after the program started, 186 participants had missing data on this variable. The final sample size depended on the study outcome and model.

We used the McNemar test to identify changes in nutrition security and food security from baseline to follow-up. Two separate binary, multivariable logistic regression models were constructed to test whether there were changes in nutrition security and food security at follow-up. Model 1 was unadjusted. Model 2 was adjusted for covariates at baseline and nutrition security status or food insecurity status. Model 3 was adjusted for those same variables plus 2 variables for fruit and vegetable consumption—daily cups of fruits and vegetables at baseline and change in daily cups of fruits and vegetables from baseline to follow-up.

We conducted an additional analysis using the McNemar test to compare the proportion of participants at baseline and follow-up who gave a response of “sometimes true” or “often true” to the 12 statements giving reasons for why their household may not have been able to eat healthier foods.

All analyses were conducted using Stata version 14 (StataCorp LLC). Results were considered statistically significant when the P value was <.05.

RESULTS

Study Participants

A total of 1,309 patient participants completed the questionnaire at both baseline and follow-up (451 at FQHC 1; 524 at FQHC 2; and 334 at FQHC 3). Over the course of the PPR, $575,820 was issued to participants on their electronic debit cards (ie, $40 per month per participant for 6 months). Participants redeemed $404,025 (70.2% of the total disbursements) at participating supermarkets/grocery stores to purchase fresh fruits and vegetables.

The 1,309 participants completing both baseline and follow-up questionnaires were predominantly female (71.1%) and had a mean age of 56.5 years (Table 1). Nearly 80% identified as being Hispanic or Latino, or of Spanish origin, and 12.9% identified as Asian. The large majority of participants (80.4%) had a high school diploma or less education, and nearly one-quarter reported having 5 or more members in their household.

Table 1.

Characteristics of Participants Completing the Produce Prescription Program (N = 1,309)

Characteristic	Value
Sex, No. (%)
Female	915 (71.1)
Male	372 (28.9)
Age, mean (range), y	56.5 (20 to 91)
Race/ethnicity, No. (%)
Asian	165 (12.9)
Black or African American	36 (2.8)
Hispanic, Latino, or Spanish origin	1,009 (78.8)
Multiracial	30 (2.3)
White	34 (2.7)
Other (single category)	7 (0.6)
Educational attainment, No. (%)
Less than a high school diploma	732 (60.5)
High school diploma or GED	241 (19.9)
Greater than high school diploma	238 (19.7)
SNAP participation, No. (%)
No	965 (73.7)
Yes	344 (26.3)
Household size, No. (%)
1-2 people	479 (37.9)
3-4 people	501 (39.6)
≥5 people	284 (22.5)
Clinic system, No. (%)
FQHC 1	451 (34.5)
FQHC 2	524 (40.0)
FQHC 3	334 (25.5)
Fruit and vegetable consumption at baseline, mean (range), cups/day	2.5 (1.2 to 5.5)
Change in fruit and vegetable consumption from baseline to follow-up, mean (range), cups/day	0.2 (−2.6 to 3.7)

FQHC = Federally Qualified Health Center; GED = General Educational Development Test; SNAP = Supplemental Nutrition Assistance Program.

Note: Participants with missing data for a characteristic were excluded from that analysis.

Changes in Nutrition Security and Food Security

During the 6 months from baseline to follow-up, there was an absolute 15.5% increase in the percentage of participants living in households who reported their situation to be nutrition secure, indicating that it was “not very hard” or “not hard at all” for their household to regularly eat healthy food (from 23.2% at baseline to 38.7% at follow-up, P <.001) (Figure 1). In addition, during the same period, there was an absolute 17.7% increase in the percentage of participants living in households that were food secure (from 25.2% at baseline to 42.9% at follow-up, P <.001).

Figure 1.

Bar Graph of Nutrition Security and Food Security Among Program Participants at Baseline and Follow-up

^a Difference between baseline and follow-up is statistically significant (P <.001).

Table 2 shows the results of the multivariable logistic regression models, which examined the odds of household nutrition security and food security before and after program participation. After adjusting for baseline study covariates and fruit and vegetable consumption, there was a significant increase in the odds of reporting nutrition security from baseline to follow-up (adjusted odds ratio = 2.95; 95% CI, 2.11-4.13; Model 3). Similarly, there was a significant increase in the odds of reporting food security from baseline to follow-up (adjusted odds ratio = 4.27; 95% CI, 4.03-6.00; Model 3).

Table 2.

Odds of Nutrition Security and Food Security After Program Participation

Outcome	Model 1a		Model 2b		Model 3c
	OR (95% CI)	P value	AOR (95% CI)	P value	AOR (95% CI)	P value
Nutrition securityd	3.65	<.001	2.98	<.001	2.95	<.001
	(2.73-4.88)		(2.14-4.15)		(2.11-4.13)
Food securitye	4.54	<.001	4.28	<.001	4.27	<.001
	(3.47-5.95)		(3.06-5.98)		(4.03-6.00)

AOR = adjusted odds ratio; OR = odds ratio; SNAP = Supplemental Nutrition Assistance Program.

^aModel 1 was unadjusted.

^bModel 2 was adjusted for sex, age, race/ethnicity, educational attainment, SNAP participation, household size, clinic system, and nutrition security or food insecurity.

^cModel 3 was adjusted for the above characteristics in Model 2 plus fruit and vegetable consumption at baseline and change in fruit and vegetable consumption from baseline to follow-up.

^dThe N value was 1,123 for Model 1, 1,028 for Model 2, and 1,012 for Model 3.

^eThe N value was 1,309 for Model 1, 1,061 for Model 2, and 1,044 for Model 3.

Reasons for Not Eating Healthier Foods

Table 3 shows the percentage of participants who endorsed the 12 barriers to healthy eating, responding “often true” or “sometimes true,” at baseline and at follow-up. At follow-up, 85.6% of participants still indicated healthy foods were too expensive, although the reduction from 88.4% at baseline was significant (P = .03). There were also improvements across the other barriers listed, including increased ability to identify more healthy foods and ability to locate stores that have healthy food options. At both baseline and follow-up, the majority of participants stated it was hard to make some of the foods from their culture healthy, although this percentage decreased (from 62.0% to 55.7%, P =.001).

Table 3.

Changes in Participants’ Endorsement of Reasons for Not Eating Healthier Foods

Reason	Participants endorsing reason,a %		P value
	Baseline	Follow-up
1. Healthy foods are too expensive.	88.4	85.6	.03
2. There are not a lot of healthy food choices at the stores I usually shop.	42.5	34.4	<.001
3. I don’t live near any stores or food pantries that have healthy foods.	31.7	28.7	.09
4. I don’t have a car or transportation to reach stores or food pantries that have healthy foods.	44.4	43.2	.49
5. I don’t have time to shop for groceries.	27.0	25.2	.30
6. I don’t have time to cook.	22.3	22.2	1.00
7. I don’t have adequate cooking equipment or ways to store food.	15.0	13.9	.44
8. I don’t know how to cook.	17.8	16.1	.23
9. I don’t know what foods are considered healthy foods.	37.7	27.5	<.001
10. I don’t like to eat healthy foods.	23.3	16.6	<.001
11. Some of the foods from my culture are hard to make healthy.	62.0	55.7	.001
12. I don’t think I qualify for food assistance programs.	38.5	37.1	.61

^aGave a response of “often true” or “sometimes true.”

DISCUSSION

This study evaluated the impact of a multisite PPR on nutrition security among Medicaid recipients with type 2 diabetes or prediabetes who were predominately working age and Latino. We found a significant improvement in nutrition security over the 6-month observation period among those who participated in the program, from about 23% at baseline to nearly 39% at follow-up. Viewed another way, the proportion of households that were nutrition insecure fell from about 77% to 61%. In a previous analysis comparing 4 other population-based studies that pilot-tested the same nutrition security questions, the prevalence of nutrition insecurity was between 18% and 44%.¹⁰ We also observed an improvement in food security from baseline to follow-up.

Main Takeaways

Two main takeaways can be drawn from this study. First, nutrition security appeared to perform well as a more inclusive indicator of food access and nutritional quality, providing a well-rounded, 360-degree view on how food as medicine interventions such as the PPR can be implemented and be beneficial in the FQHC setting. For example, PPR implementation can be tailored to help shape community-level experiences and increase participant knowledge about the availability of healthy food options in or near their neighborhood. Unlike other, traditional food security metrics, which focus primarily on access to sufficient calories (ie, viewing assessment of and intervention on hunger as a set of minimum standards or requirements to fulfill), the 2-item nutrition security measure includes documentation of specific access barriers (eg, transportation to sources of healthy foods, affordability of healthy foods). This information allows us to better understand and intervene to improve dietary quality for individuals having elevated health risks due to diabetes, prediabetes, and other diet-related chronic diseases. In a way, exploring nutrition security allows us to capture a more robust portrayal of nutritional status and need(s) than simply asking about food insecurity. It provides information on the nuanced aspects of programs or program implementation that can be adjusted to reduce hunger and poor diet—that is, factors that are potential mediators of clinical outcomes downstream.¹⁶

The second study takeaway is that in the analysis of possible barriers to eating a healthy diet, most PPR participants continued to indicate that healthier foods were too expensive at follow-up, suggesting the program may be overlooking the affordability of food as a key element in program design, redesign, and/or implementation. For instance, although the minimum wage in California increased steadily from 2019 to 2023, which roughly paralleled the timeframe of this PPR, inflation also rose substantially, diminishing this wage gain.¹⁷ Specifically, food prices increased by 25%, rising faster than housing and medical care expenses.¹⁸ For households in the lowest income bracket, food expenses represented roughly one-third of after-tax income. Future iterations of the PPR should account for these effects of inflation by increasing the amount of the nutrition incentive and/or basing the amount on household size. Additional considerations may include partnerships with local farms or with farmers’ markets to offer lower-cost produce options. Addressing food affordability may add to the benefits of participating in a PPR—in our study, participants were more knowledgeable about where to obtain healthy foods after completing the program. Other food as medicine interventions elsewhere in the United States are likely facing similar challenges in PPR delivery.

A growing body of research shows that providing produce prescriptions to subsidize the cost of fruits and vegetables, at least in the short term, can impact a wide range of health outcomes. Studies have found positive associations between program participation and increases in fruit and vegetable consumption and food security (ie, at the same time, independent of one another).² Emerging research also suggests that food as medicine interventions can have beneficial effects on clinical/health outcomes, including blood pressure and body mass index,^19-21 although findings for these outcomes have been mixed among patients with diabetes.²

Stable access to nutritious food is imperative for disease prevention and management. Our study builds on this body of evidence by focusing on nutrition security, which is favorably associated with many health outcomes. Studies that assess the intersection between nutrition security and cardiometabolic disease allude to the role of PPRs in reducing nutrition insecurity.^22-24 Other stakeholders in health care and the food system have elevated the need for policies aimed at increasing nutrition security because of these intersections with health equity and quality of care.²⁵ Improving nutrition security can help increase health equity among individuals and at-risk populations because of its central role in mediating nutrition-related conditions and protecting against poor outcomes.²⁴

Limitations

This study has several limitations. First, we did not use a comparison group because of grant limitations; instead, we used a before-and-after design in a single cohort of participants yielding 2 data points, which limited our ability to definitively attribute the observed changes in nutrition security to the PPR intervention. Second, the questions on nutrition security were not fully validated at the time of our study, although other large population studies had pilot-tested these same questions, and efforts are under way to validate them against traditional measures of food security. It should also be noted that other measures of nutrition security are currently being tested.¹² Third, information on grocery store/supermarket redemption of the incentive at the level of the participant was not available; clearly, more data are needed to further inform this area of research and program impact.

Macro-Level Implications

The PPR approach has the potential to create ripple effects beyond individual health. By improving nutrition security, the program may potentially (1) reduce the economic burden of health care costs associated with diet-related chronic diseases, (2) strengthen the local economy that supports local and regional food systems, and (3) raise awareness about the importance of healthy eating, capitalizing on multilevel, holistic programming. Future research and practice may benefit from exploring these and other more macro-level effects. As the PPR, and other food as medicine interventions, spread nationally, their role in addressing these broader determinants of health and nutrition security could take on greater significance, responding to an unprecedented opportunity to foster healthier communities by developing more equitable food systems across the United States.

Conclusions

The PPR in Los Angeles County demonstrated early success in improving nutrition security among diabetic or prediabetic individuals who were predominantly food insecure at baseline. Study results showed that this food as medicine intervention was integrated relatively seamlessly into several clinical settings, helping FQHCs to identify patient nutritional needs, initiate assistance, and address barriers to accessing nutrient-dense foods, particularly fresh fruits and vegetables. Future interventions of this type should incorporate local contexts into their planning and implementation. By maintaining its focus on improving dietary quality, food affordability, and sustainability of the approach, the PPR could apply its program elements to meaningfully address gaps in nutrition and food security. The PPR could serve as a promising model for generating practical solutions to real-world issues, bridging the gap between food access and nutrition and the downstream clinical outcomes it mediates.¹⁶