Prevalence and Predictors of Household Food Insecurity and Supplemental Nutrition Assistance Program Use in Youth and Young Adults With Diabetes: The SEARCH for Diabetes in Youth Study

The SEARCH for Diabetes in Youth (SEARCH) study is a multicenter observational study of provider-diagnosed diabetes mellitus in youth diagnosed before the age of 20 years. Initially designed as a multicenter surveillance effort aiming to assess incidence and prevalence of youth-onset diabetes, SEARCH has expanded over the years to a longitudinal cohort study (19). Study sites were located in South Carolina, Ohio, Colorado (including the Navajo Nation in Arizona and New Mexico), Washington, and California.

In brief, in phase 1 (funded 2000–2005), researchers identified cases that were prevalent in 2001 and incident cases occurring between 2002 and 2005. In SEARCH phase 2 (funded 2005–2010), researchers ascertained incident case participants between 2006 and 2010 and also initiated follow-up of enrolled phase 1 incident case participants 12, 24, and 60 months after their initial visits. In SEARCH phase 3 (funded 2010–2015), researchers ascertained incident case participants between 2010 and 2015. The SEARCH cohort was created by recruiting participants from phases 1 and 2 to another in-person visit. The inclusion criteria for this subset included 1) a diagnosis date in 2002–2005, 2006, or 2008; 2) having completed a baseline in-person visit; and 3) having at least 5 years of diabetes duration at the time of the visit. During phase 3, incident case participants who were identified in 2012 were invited for a baseline in-person visit. In phase 4 (funded 2015–2020), the SEARCH cohort was operationally split into the group invited to another study visit and those who were invited to complete surveys but not to an in-person visit. Those invited to the in-person visit included a subset of participants who had completed a cohort or baseline in-person visit in phase 3 (including incident cases from 2002–2006, 2008, and 2012), which comprised all participants with type 2 diabetes, all participants with type 1 diabetes from racial or ethnic minority groups, and a random sample of eligible non-Hispanic Whites with type 1 diabetes. Additional eligibility requirements included at least a 5-year diabetes history, at least 3 years since the participant’s phase 3 in-person visit, and a requirement that the participant be ≥10 years at the time of the phase 4 visit.

The study was approved by and followed procedures in accordance with the ethical standards of the respective local institutional review boards. Before data collection commenced, parents or guardians of participants younger than 18 years provided written informed consent, and participants aged 8–17 years provided assent; all participants aged ≥18 years provided written informed consent. We used cross-sectional data from the SEARCH phase 4 cohort in the present study. This was the first phase of the SEARCH study where the U.S. Department of Agriculture’s Household Food Security Survey Module (HFSSM) was administered to all SEARCH participants.

HFI was assessed using the 18-item HFSSM, which queries about experiences with food security over the previous 12 months (1). The HFSSM was completed by parents or guardians of minor participants and by participants ≥18 years of age. The first 10 questions pertain to all households (with or without children) and the last 8 questions are only asked of households with children aged 0–17 years.

Household food security (possible responses ranging from 0–10 affirmations for households without children and 0–18 affirmations for households with children) was classified into four categories: high food security (zero affirmations), marginal (one to two affirmations), low (three to seven affirmations among households with children; three to five among households without children), and very low food security (eight or more affirmations in households with children; six or more affirmations in households without children). A dichotomous version was created by combining high and marginal (none to two) affirmations into the food-secure group and low or very low (three or more) into the food-insecure group (1).

Participation in SNAP was determined by participants’ response to the question, “In the last 12 months, did you or any members of your household receive food stamps, also known as the Supplemental Nutrition Assistance Program or SNAP?”

Age at most recent phase 4 study visit or survey completion and sex were confirmed from medical record data. Race and ethnicity were obtained through self-report using standard U.S. Census questions on the initial participant survey (20). Combined race and ethnicity data were categorized as non-Hispanic White, non-Hispanic Black, Hispanic of any race, and other (American Indian or Alaska Native, Asian Pacific Islander, multiracial, or unknown).

Participants ≥18 years of age reported their parent’s highest educational degree or level of schooling completed. For those <18 years of age, parents or guardians reported their highest educational degree or level of schooling completed, as well as that of their partner or spouse. Responses were dichotomized as parents having a bachelor’s degree or higher versus less than a college degree.

To assess household income, participants were presented with annual income ranges. Household income was categorized as <$25,000, $25,000–$49,999, and ≥$50,000, with the lower income cutoff closely approximating the U.S. federal poverty line for a family of four in 2019. The young adult SEARCH participants were asked these same questions about their parents, following the rationale that in early adulthood, the socioeconomic characteristics of the parental household influence the socioeconomic position of the young adult.

Current health insurance was assessed by asking about the type of health insurance or health care plan, offering eight answer choices that were subsequently grouped into private health insurance (i.e., insurance through an employer or purchased independently or from the military), Medicaid or Medicare, other, or none. If multiple types of insurance were selected, participants were allocated to the more comprehensive type of insurance. Individuals who gave no indication of having any type of health insurance were put into the no-insurance group.

The prevalence of HFI was estimated as a simple proportion with 95% CIs and was stratified by diabetes type. We used χ² tests to assess whether the prevalence of HFI was associated with diabetes type and if SNAP participation was associated with participant characteristics. Multivariable logistic regression models, stratified by diabetes type, were used to examine differences in HFI by age, sex, race or ethnicity, parental education, household income, and health insurance. Models were also adjusted for clinical site. All analyses were completed using SAS, version 9.4 (SAS Institute, Cary, NC).

The study sample included 2,177 YYA with type 1 diabetes and 384 YYA with type 2 diabetes, for a total of 2,561 participants (Table 1). Compared with participants with type 1 diabetes, those with type 2 were older (mean age, 24.7 vs. 21.0 years) and more likely to be female (66.9% vs. 52.2%) and from racial or ethnic minority groups (non-Hispanic Black, Hispanic, and non-Hispanic other) (81.2% vs. 28.2%). A higher percentage of participants with type 1 diabetes had parents with a bachelor’s degree or higher educational attainment (54.5% vs. 17.4%) and had private health insurance (76.0% vs. 45.3%) than did participants with type 2 diabetes.

The overall prevalence of HFI was 19.7%, with 8.6% of participants reporting living in households with very low food security (Table 2). Prevalence of HFI was higher in older YYA (aged 10–17 years: 16.6%; 18–25 years: 19.2%; >25 years: 25.0%), female participants (21.8% vs. 17.1% male participants), and non-Hispanic Blacks (30.0% vs. 16.8 non-Hispanic White; Supplementary Table 1). Participants whose parents had a lower level of education, with lower household income, and either having Medicaid, Medicare, or no insurance also had a higher prevalence of HFI. Whereas 17.7% of YYA with type 1 diabetes reported experiencing HFI, almost a third (30.7%) of YYA with type 2 diabetes reported HFI (Supplemental Figure 1). In addition, 11.7% of YYA with type 1 diabetes and 20.1% of YYA with type 2 diabetes were identified as marginally food secure. In the multivariable regression model, participants with type 2 diabetes did not have higher odds of experiencing HFI than type 1 diabetes participants (odds ratio [OR] 1.07; 95% CI 0.78–1.47).

In the overall sample, participants with public insurance or no insurance had, respectively, 1.56 (95% CI 1.18, 2.01) and 1.74 (95% CI 1.11, 2.73) times the odds of reporting HFI than did YYA with private insurance (Table 3). The odds of HFI were higher among YYA with household income between $25,000 and $49,999 (OR 4.66; 95% CI 3.33–6.52) and <$25,000 (OR 3.56; 95% CI 2.59–4.89) than among YYA living in households with income ≥$50,000. In addition, participants with parents with less than a bachelor’s degree had higher odds (OR 1.78; 95% CI 1.39–2.27) of experiencing HFI than did YYA with parents with a bachelor’s degree or higher. In YYA with type 1 diabetes, participants receiving Medicaid or Medicare, whose parents had a lower level of education, and with lower household income had greater odds of HFI. In YYA with type 2 diabetes, the odds of HFI were higher among YYA with household income between $25,000 and $49,999 (OR 19.0; 95% CI 3.94–91.8) and <$25,000 [OR 24.4; 95% CI: 4.94–120.2) compared with YYA living in households with income ≥$50,000.

Approximately one in seven households participated in SNAP (14.3%), with a higher percentage of SNAP participation observed among participants with type 2 diabetes than among those with type 1 diabetes (34.8% vs. 10.7%; P < 0.001). Prevalence of SNAP participation differed by all participant characteristics for the overall sample (Table 4). For participants with type 1 diabetes, prevalence of SNAP participation was higher among girls or young women (P = 0.008), non-Hispanic Blacks (P <  0.001), those who reported lower parental education (P <  0.001), lower household income (P <  0.001), and those who reported receiving Medicaid or Medicare or having no insurance (P <  0.001). Among those with type 2 diabetes, prevalence of SNAP participation differed by race/ethnicity (P = 0.001), parental education (P = 0.016), household income (P < 0.0001), and health insurance (P < 0.0001).

In this population-based sample of YYA with diabetes, nearly one in five (19.7%) experienced HFI in the past year, a prevalence that is almost twice that of the general U.S. population in 2019 (2). The overall prevalence of HFI in YYA with diabetes is of comparable magnitude with that of U.S. adults with diabetes (20.3%), the majority of whom are older adults with type 2 diabetes (18). That >30% of YYA with type 2 diabetes experience HFI highlights that, compared with their older adult counterparts, YYA with type 2 diabetes are more vulnerable to HFI and, thus, are more at risk for its associated negative consequences.

The HFI prevalence of 16.6% for youth between 10 and 17 years of age in this cohort was lower than the reported estimates of HFI prevalence among Canadian youth with type 1 diabetes or type 2 diabetes requiring insulin (21.9%) (11), although the difference is partly due to the lower cut point in the scale score to indicate HFI used in Canada. HFI prevalence, using the standard US Department of Agriculture definition (3), was almost one in five (19.2%) for participants aged 18–25 years and one in four (25.0%) for participants between 26–35 years of age. The period of emerging and young adulthood represents a high-risk period for HFI for individuals with diabetes, likely due to the multiple life transitions that young adults experience, which can contribute to income instability (21).

It is worth highlighting that an additional 13% of households of YYA with diabetes (11.7% among type 1 diabetes, 20.1% among type 2 diabetes) were classified as marginally food secure (i.e., affirming one or two food insecurity items), which increases the overall proportion of YYA affected by any level of food insecurity severity to 32.6% overall (29.4% among type 1, 50.8% among type 2). The high HFI prevalence in YYA with diabetes is strong evidence of the need to follow the American Diabetes Association’s recommended HFI screening guidelines for all people with diabetes (5). Without awareness of their patients’ food security status, health care providers are unable to consider this when making decisions about treatment or referrals and may even offer medical advice that would be viewed as entirely unrealistic given the patient’s financial circumstances (22). HFI, for example, is associated with increased risk of hypoglycemia in individuals with diabetes (23) and can occur as a result of inadequate or erratic carbohydrate consumption after the administration of insulin or sulfonylureas. HbA_1c goals should be individualized and reassessed over time (5). Provider awareness of patients experiencing low or very low food security can help guide discussions by the diabetes care team, including targeting a higher HbA_1c goal for the patient if advisable, and reviewing strategies to prevent hypoglycemia, as well as possible ketosis, in the setting of intermittent or prolonged fasting.

HFI has been associated with worse glycemic control and a higher rate of acute diabetes complications (24). In a study examining 228 YYA with type 1 diabetes, Mendoza et al. (25) found that participants living in food-insecure households had 2.64 higher odds (95% CI 1.32, 5.25) of having an HbA_1c value >9.0% as compared with those living in food-secure households. Thus, the awareness of patients’ HFI status may also help providers better understand barriers to appropriate diabetes self-care and potential competing demands for their resources. YYA with diabetes may find themselves caught between competing priorities, such as procuring food and paying for prescribed medications and supplies for diabetes while managing other living expenses (26). In a recent national sample of privately insured children and adults with type 1 diabetes, mean out-of-pocket costs for medical care was nearly $2,500 annually (27). Our finding that HFI disproportionally affects YYA with diabetes who have lower socioeconomic status, along with those with public health insurance or without health insurance, suggests that YYA with type 1 diabetes are particularly vulnerable to balancing out-of-pocket expenses and access to safe, nutritious foods. Importantly, many participants with household incomes above the poverty line reported experiencing HFI. Policies that address the underlying causes of HFI in YYA with diabetes are required and should include innovative interventions to enhance the ability to access healthy food, perhaps by increasing benefits in existing programs for those with chronic conditions. Efforts to improve average wages, subsidized housing, and pharmacy benefits to reduce cost-related medication underuse, including insulin rationing in the setting of rising insulin costs, could be explored as means to help alleviate HFI (28). By reducing competing demands, policies such as these can redirect financial resources toward purchase of nutritious foods and, thus, indirectly reduce HFI and promote better diabetes self-management.

Our findings demonstrate that no subgroup of YYA with diabetes is free from experiencing HFI, reinforcing the need to screen all YYA with diabetes. Consistent with previous studies, racial or ethnic inequities in the prevalence of HFI among YYA with diabetes exist. Overall HFI rates for non-Hispanic Black participants with type 1 diabetes were nearly twice that of non-Hispanic White households, which is consistent with the inequities in HFI in the general U.S. population. HFI prevalence among non-Hispanic Black participants with type 1 diabetes and type 2 diabetes were similar, suggesting that sociodemographic and environmental factors are highly influential for HFI rather than diabetes type. Since 2001, HFI prevalence in the U.S. among non-Hispanic Blacks has been at least twice that of non-Hispanic Whites (2).

Strikingly, although the prevalence of HFI among YYA with type 2 diabetes was 1.7 times as high compared with YYA with type 1 diabetes, we found that participants with type 2 diabetes did not have higher odds of experiencing HFI than did YYA with type 1 diabetes in our multivariable regression analysis. Given that participants of lower socioeconomic status and who identify as non-Hispanic Black or Hispanic composed the majority of our sample of YYA with type 2 diabetes, resolving disparities in HFI also carries major implications for addressing inequities experienced by underserved racial and ethnic minorities. Thus, solutions to address inequities in HFI must not only focus on addressing socioeconomic challenges but also work to combat broader inequities, discrimination, and structural racism if equity in food security is to be achieved (29).

From a diabetes care team standpoint, screening for HFI by diabetes providers is a necessary first step. Screening must be coupled with effective interventions. The most direct way to ameliorate the health consequences associated with HFI is to reduce food insecurity. Diabetes care team members can play an important role by connecting patients to food and nutrition resources as well as other resources to reduce competing demands, including determining whether patients are participating or are eligible to participate in available federal nutrition programs. Approximately 1 in 10 YYA with type 1 diabetes and more than 1 in 3 YYA with type 2 diabetes in our sample reported participating in SNAP, with the type 2 diabetes participation rate representing a almost threefold higher participation rate in SNAP than that of the general population (2). Despite the availability of food-specific safety net programs, 8.5% of YYA with diabetes reported experiencing very low food security, which indicates that eating patterns of one or more household members were disrupted and food intake reduced because the household lacked money and other resources for food. Given that SNAP only covers up to 70% of food expenditures (4), diabetes care teams also should be adept at connecting patients to local resources, including school and summer lunch programs, food pantries, food pharmacies, community gardens, medically tailored meals, vegetable prescription programs, or grocery distribution programs. Studies examining effective intervention strategies, including linkage to community resources or financial assistance, are needed (30).

Strengths of the present study include its large sample size; a study population drawn from what is the largest, multiethnic, population-based registry of individuals with youth-onset diabetes in the U.S. with demographic and clinical characteristics similar to those of the overall U.S. population; and the use of the full 18-item HFSSM, the U.S. reference measure for HFI. However, this study has limitations. First, the small sample size of participants with type 2 diabetes may limit the precision of the HFI prevalence estimate, as well as the ability to detect predictors of HFI in the group with type 2 diabetes. Second, although the survey response rate for SEARCH 4 participants was ∼75%, nonresponse bias may have affected HFI prevalence estimates. Third, we were unable to comment on how state-level factors and policies influenced HFI and the use of food assistance programs (31). Some state policies, for example, can make it difficult to stay enrolled in SNAP, resulting in “churning,” which is seen when households exit and then re-enter the program within 4 months due to burdensome recertification requirements and benefit criteria that are sensitive to changes in household composition (32). Finally, additional work is needed to better understand how HFI is related to key clinical indices, such as HbA_1c. Given the availability of this large sample of SEARCH participants, we plan to evaluate the impact of different levels of HFI on glycemic control in those with youth-onset type 1 and type 2 diabetes.

Data from this large population-based multicenter study confirm that the prevalence of HFI is higher for YYA living with diabetes than in the general U.S. population average of 1 in 10 and may be contributing to diminished health in this population. That one in five YYA with diabetes overall and almost one in three with type 2 diabetes live in households that do not have enough food for every family member to lead a healthy life illustrates the urgent need to identify and address HFI in this vulnerable population.

Funding. Phase 4, SEARCH 4, of the SEARCH for Diabetes in Youth cohort study is funded by grants from the National Institutes of Health (NIH), National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (1R01DK127208-01, 1UC4DK108173), and supported by the Centers for Disease Control and Prevention (CDC). The Population-Based Registry of Diabetes in Youth study is supported by grants from the NIH NIDDK (1U18DP006131, U18DP006133, U18DP006134, U18DP006136, U18DP006138, U18DP006139) and is funded by the CDC (grant DP-15-002). For SE 1–3, the SEARCH for Diabetes in Youth cohort study is funded by the CDC (PA numbers 00097, DP-05-069, and DP-10-001) and supported by grants from the NIDDK and from Kaiser Permanente Southern California (U48/CCU919219, U01 DP000246, and U18DP002714), University of Colorado Denver (U48/CCU819241-3, U01 DP000247, and U18DP000247-06A1), Cincinnati’s Children’s Hospital Medical Center (U48/CCU519239, U01 DP000248, and 1U18DP002709), University of North Carolina at Chapel Hill (U48/CCU419249, U01 DP000254, and U18DP002708), Seattle Children’s Hospital (U58/CCU019235-4, U01 DP000244, and U18DP002710-01), and Wake Forest University School of Medicine (U48/CCU919219, U01 DP000250, and 200-2010-35171). Funding for the SEARCH Food Security cohort study, “Impact of disparities in food security on glycemic control and health care utilization among youth and young adults with diabetes,” is from the NIH NIDDK (1R01DK11746). In addition, a K23 Career Development Award from the NIH NIDDK (DK119465) supported the time of author F.S.M.

Duality of Interest. No potential conflicts of interest relevant to this article.

Author Contributions. F.S.M. contributed to the study design and interpretation of the data, and drafted the manuscript. A.D.L. and J.A.M. contributed to the study design and interpretation of the data. B.A.R. contributed to the study design and conducted the analyses. E.M-D., D.D., J.M.L., B.L., A.B., E.J., and C.T. contributed to interpretation of the data. All authors critically reviewed and edited revisions before approving the final version. J.A.M. is the guarantor of this work and, as such, had full access to the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of the article were presented at the 45th Annual Conference of the International Society for Pediatric and Adolescent Diabetes, Boston, MA, 30 October–2 November 2019.