{
  "abstract": "Background Food insecurity (FI) is linked to mental health outcomes cross-sectionally, but little is known about temporal patterns of FI and changes in mental health. FI can exacerbate the mental health challenges of managing diabetes, creating a vicious cycle that worsens youth and young adults’ (YYAs) mental well-being.Purpose We investigated the association of temporal patterns of FI with symptoms of depression, anxiety, and stress, and changes therein in YYAs with youth-onset type 1 (T1D) and type 2 diabetes (T2D).Methods Longitudinal data (2016–2022) including three time points (t 1, t2, and t3) from 747 YYAs with T1D and 97 YYAs with T2D were analyzed using multivariable linear regression. Ascertained with the Household Food Security Survey Module, food security was classified as persistently food secure, persistently food insecure (PFI), and intermittently food insecure (IFI). Mental health at t3 and changes from t2 to t3 were characterized with the Center for Epidemiologic Studies Depression Scale, the Generalized Anxiety Disorder Scale, and Cohen’s Perceived Stress Scale.Findings Among YYAs with T1D and T2D, 6.6% and 16.5% were PFI, 20.3% and 42.3% were IFI, respectively. In YYA with T1D, PFI and IFI were associated with greater depressive, anxiety, and stress symptoms at t 3, and with increased symptoms over time. In YYA with T2D, PFI was associated with greater depressive symptoms at t3 but not with changes over time.Interpretation The study identified a previously unrecognized link between prolonged exposure to FI and increased incidence of mental health issues. Both persistent and intermittent FI were associated with adverse mental health symptoms in YYA with diabetes, more so for those with PFI. Subsequent research should prioritize interventions that address FI in this population to evaluate their effectiveness in enhancing both physical and psychological well-being. It should be designed to not only address FI, but also comprehensive support, including mental health services and education.",
  "authors": [
    {
      "affiliations": [
        "Department of Epidemiology and Biostatistics, University of South Carolina Arnold School of Public Health, Columbia, South Carolina, USA"
      ],
      "name": "Ali M Alfalki"
    },
    {
      "affiliations": [
        "Department of Epidemiology and Biostatistics, University of South Carolina Arnold School of Public Health, Columbia, South Carolina, USA"
      ],
      "name": "Emmanuel F Julceus"
    },
    {
      "affiliations": [
        "Department of Psychology, University of South Carolina College of Arts and Sciences, Columbia, South Carolina, USA"
      ],
      "name": "Kate Flory"
    },
    {
      "affiliations": [
        "CHBD, University of Washington, Seattle, Washington, USA"
      ],
      "name": "Jason A Mendoza"
    },
    {
      "affiliations": [
        "Seattle Children’s Hospital, Seattle, Washington, USA",
        "University of Washington, Seattle, Washington, USA"
      ],
      "name": "Faisal S Malik"
    },
    {
      "affiliations": [
        "Department of Health Promotion, Education, and Behavior, University of South Carolina Arnold School of Public Health, Columbia, South Carolina, USA"
      ],
      "name": "Edward A Frongillo"
    },
    {
      "affiliations": [
        "Department of Biostatistics and Data Science, Wake Forest University School of Medicine, Winston-Salem, North Carolina, USA"
      ],
      "name": "Beth A Reboussin"
    },
    {
      "affiliations": [
        "Department of Lifecourse Epidemiology of Adiposity and Diabetes, Colorado School of Public Health, Aurora, Colorado, USA"
      ],
      "name": "Anna Bellatorre"
    },
    {
      "affiliations": [
        "Department of Lifecourse Epidemiology of Adiposity and Diabetes, University of Colorado Anschutz Medical Campus, Aurora, Colorado, USA"
      ],
      "name": "Dana Dabelea"
    },
    {
      "affiliations": [
        "Pediatrics, University of Washington, Seattle, Washington, USA",
        "Seattle Children’s Research Institute, Seattle, Washington, USA"
      ],
      "name": "Catherine Pihoker"
    },
    {
      "affiliations": [
        "Department of Epidemiology and Biostatistics, University of South Carolina Arnold School of Public Health, Columbia, South Carolina, USA"
      ],
      "name": "Angela D Liese"
    }
  ],
  "full_text": "WHAT IS ALREADY KNOWN ON THIS TOPIC Available cross-sectional evidence suggests that food insecurity (FI) is associated with adverse mental health among youth and young adults (YYAs) with diabetes. Studies have shown that depressive symptoms, anxiety, and stress are worse and significantly more likely to be reported among YYAs with FI compared with those who are food secure, including low-income populations. While several studies have demonstrated an association between FI and negative physical health consequences in populations with diabetes, there is a paucity of longitudinal studies. This study provides a longitudinal perspective on the specific impact of chronic and sporadic FI on mental health among YYAs with diabetes, an understudied population.WHAT THIS STUDY ADDS This study is based on longitudinal analyses of a cohort of YYAs with diabetes followed over 7 years (2016–2022), thereby offering insights into potential association between FI and mental health issues and a comprehensive overview of trends over time. The results showed that the mental health risks associated with FI, which in turn may affect physical well-being, self-care, and functioning, resulting in considerable impairment across multiple life domains and economic burdens. Including YYAs with type 1 and type 2 diabetes from a diverse multicenter study enhances the generalizability of the findings.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY FI has been and still is a significant public health burden, especially for people with chronic diseases such as diabetes. The association between FI and poor mental health in YYAs underscores the need for integrated policy approaches that address both food security access and mental health services. Future research should prioritize interventions that address FI and evaluate their effectiveness in enhancing physical and psychological well-being in young adults with diabetes. Addressing both food security and mental health requires integrated policy and care that includes diabetes management and engagement of managed care providers, particularly Medicaid programs, to support access to nutritious food, mental health services, and coordinated diabetes care.Introduction Food insecurity (FI) is defined as “limited or uncertain availability of nutritionally adequate and safe foods or limited or uncertain ability to acquire acceptable foods in socially acceptable ways”. 1 FI remains a challenging problem, impacting approximately 12.8% of families in the USA in 2023, approximately 18 million individuals.2 3 During the 2021–2023 period, there was considerable variation at the state level, with the prevalence of FI among households ranging from 7.4% in New Hampshire to a high of 18.9% in Arkansas.1 FI is a particular concern for individuals with diabetes, as nutrition is a key component to diabetes self-management. FI is associated with fear of hypoglycemia, diabetic ketoacidosis, and elevated hemoglobin A1c levels.4 Although type 1 diabetes (T1D) and type 2 diabetes (T2D) differ in etiology and treatment regimens, FI affects both groups.5 6 Youth and young adults (YYAs) with diabetes face unique difficulties in affording diabetes-related supplies and diet, which may contribute to stress, anxiety, and depressive symptoms.7 Recognizing FI as a shared and critical issue across diabetes types underscores the need to examine its impact on mental health and well-being.A growing body of cross-sectional research indicates that in the general population FI is linked to many adverse mental health outcomes, including increased anxiety, depression, and stress.8 Specific studies highlight that FI is associated with severe depressive symptoms in youth and adults living with diabetes.9 For instance, young adults with T1D living in FI households reported an average of 7.4 points higher on depressive symptom scale, while those with T2D reported an average of 4.8 points higher compared with their respective food secure peers.10 Further, an analysis of the National Health and Nutrition Examination Survey revealed that individuals in households with FI reported higher numbers of depressive symptoms over the past 12 months, regardless of underlying health conditions.11 These findings are concerning because the prevalence of diabetes in youth continues to increase, as does the prevalence of mental health issues.12 13 Longitudinal research on temporal patterns of FI is rare but has the potential to shed light on the relations between changes in FI and changes in outcomes.14 This is particularly relevant for YYAs with diabetes, because the stress of managing diabetes can intensify the impact of FI, further exacerbating mental health outcomes in this population.15 We investigated the relation between FI patterns across three time points and depressive symptoms, anxiety, and stress at the third time point, as well as the changes in these symptoms based on the differences between measurements taken at the second and third time points among YYA with diabetes. We hypothesized that, compared with YYA with diabetes who are persistently food secure, those who experience persistent or intermittent FI will have elevated depressive, anxiety, and stress symptoms at the third time point, and corresponding changes in symptoms between the second and third time points. This study is novel in its longitudinal analysis of the patterns of FI over time, which offers fresh perspectives on how changes in FI status influence mental health outcomes in YYAs with diabetes, an area that has not been extensively explored in previous research.16 Research design and methods Study design and population The SEARCH Food Security 1 (SFS 1) Cohort Study is an ancillary study to the SEARCH for Diabetes in Youth Study 4 (SEARCH 4). 17 The SEARCH 4 had five sites across the USA; however, only three (South Carolina, Washington, and Colorado) participated in the SFS 1 Cohort Study’s follow-up 1 (FU1) and (FU2). These sites received ancillary funding to support a study on food security and relevant outcomes (R01DK117461 MPI Liese and Mendoza), whereas the remaining SEARCH 4 sites did not participate due to logistical and funding constraints. Participants were aged 10–38 and had youth-onset diabetes mellitus (T1D or T2D). The present study employed data from three separate periods of time, encompassing the SEARCH 4 (2016–2019) assessment as our baseline time point as well as the subsequent SFS 1 FU1 and FU2 data collections follow‑ups. Between 9 and 27 months (average=18) elapsed between SEARCH 4 and SFS 1 FU1 (data collection was initiated November 2018 and ended February 2021); the interval between FU1 and FU2 was approximately 9 months. FU2 data collection was initiated in October 2019 and ended in March 2022. The mean participant age was in the mid-20s, and thus effects of school vacation timing due to reduced access to school-based meals were unlikely to substantially influence the findings. Prior to data collection, the SEARCH 4 and SFS 1 protocols were reviewed and approved by each site’s institutional review board. Written informed consent was obtained from participants aged 18 years or older and parents or legal caregivers of participants under the age of 18. Mental health measures were self-reported by participants aged 14 years and older, consistent with validated use in adolescent and adult populations. This age-appropriate administration ensured measure validity across developmental stages. All minors gave their assent to participate. Study methods and results are reported following the Strengthening the Reporting of Observational Studies in Epidemiology Statement for cohort studies.18 A total of 2668 individuals participated in the SEARCH 4 study of whom 1840 were eligible for the SFS 1 study because of being enrolled at one of the three SFS 1 data collection centers. Of these, 1161 participated in SFS 1 FU1 and 1011 participated in SFS 1 FU2 and were assessed for FI and mental health symptoms, specifically extent of depression, anxiety, and stress, during both data collection follow‑ups (figure 1). For the current data analyses, 1075 participants at SEARCH 4 were included from the original 1840 eligible for SFS 1 (figure 2), because 569 individuals were lost to follow-up and 196 were missing one or more covariates, 844 participants (747 with T1D, 97 with T2D) were followed at the third time points. Whereas SFS 1 FU1 data collection occurred before the COVID-19 pandemic, SFS 1 FU2 overlapped with the pandemic’s early and mid-phases. The analytic sample included participants with complete data on FI, mental health outcomes, and covariates at the required time points; participants missing key exposure or outcome variables were excluded. Excluded participants were on average younger with a shorter diabetes duration and were more likely to be male, from South Carolina and were less likely to be Hispanic (online supplemental table 1). Analyses were conducted using complete-case data to ensure consistent longitudinal estimates; potential bias was assessed by comparing characteristics of participants retained with those lost to follow-up.SP110.1136/bmjdrc-2025-005511.supp1Supplementary data Figure 1Temporal flowchart of data from SEARCH 4 and SFS 1 follow‑ups. FI, food insecurity; FU1, follow-up 1; FU2, follow-up 2; SEARCH 4, SEARCH for Diabetes in Youth Study 4; SFS 1, SEARCH Food Security 1.Figure 2A comprehensive flowchart of the analytical data process from the initial data collection through data cleaning to the final sample used. FU1, follow-up 1; FU2, follow-up 2; SEARCH 4, SEARCH for Diabetes in Youth Study 4; SFS 1, SEARCH Food Security 1; CA, California; CO, Colorado; OH, Ohio; SC, South Carolina; WA, Washington.Measures Household food insecurity FI was assessed with the Household Food Security Survey Module (HFSSM), and evaluated at three time points: SEARCH 4, SFS 1 FU1, and FU2. FI was measured using the HFSSM with a 12-month reference period to ensure comparability with national surveillance and prior SEARCH data. While the HFSSM is validated for the general population, it has not been validated specifically for individuals with diabetes. 19 Young adult participants aged 18 years and older without minors living in the household were asked to complete the 10-item HFSSM questionnaire,20 including items such as “Did you ever cut the size of your meals or skip meals because there wasn’t enough money for food?” If there were children in their households, participants were asked to complete the 18-item HFSSM questionnaire, which includes items such as “Were the children ever not eating enough because you couldn’t afford enough food?”.20 For participants under the age of 18 years, a parent/guardian completed the 18-item questionnaire.20 We employed the US Department of Agriculture Economic Research Service definition of FI.20 Individuals with three or more affirmative responses to the HFFSM items were classified as FI, while those with two or fewer were considered food secure.20 Among the 844 participants with three food security assessments used for analyses, three distinct groups were identified: persistently food secure (PFS), which comprised individuals who consistently remained food secure at all assessment timepoints; persistently food insecure (PFI), comprising individuals who experienced FI at all observed time points; and intermittently food insecure (IFI), comprising individuals who fluctuated between food security and FI (ie, at least one time point each of food security and FI).Depressive symptoms Depressive symptom scores were obtained from self-reported responses to the Center for Epidemiologic Studies Depression (CES-D) Scale, which includes 20 items. 21 The items were administered following the SEARCH 4 protocol with two items omitted related to suicidality. Respondents rate the frequency of experiencing each symptom, which ranges from 0 (indicating “rarely or none of the time”) to 3 (indicating “all of the time”). The CES-D total score ranges from 0 to 60, with higher scores indicating more and/or more frequent symptoms of depression. Specifically, for adolescents, we applied the three-category, a 24-cut-point stratification to classify scores as minimal (0–15), mild (16–23), and moderate/severe (24–60). For adults, we used a score of ≥16 to indicate clinically significant depressive symptoms.22 Data on depressive symptoms were used from the final two time points, SFS 1 FU1 and FU2.Anxiety symptoms The assessment of anxiety symptoms was evaluated at the final two time points, SFS 1 FU1 and FU2, by using the 7-item Generalized Anxiety Disorder (GAD-7) scale. The GAD-7 has been widely recognized as a reliable assessment instrument, and its utility has been demonstrated in YYA. 23 The rating for each item is conducted using a Likert scale to measure the frequency of anxiety symptoms, ranging from 0 (indicating “not at all”) to 3 (indicating “nearly every day”). The GAD-7 total score ranges from 0 to 21, with higher scores indicating more and/or more frequent symptoms of anxiety.23 24 Stress symptoms The frequency of stress symptoms was evaluated at the final two time points, SFS 1 FU1 and FU2, using the Cohen’s Perceived Stress Scale (PSS). The PSS is a self-reported 14-item questionnaire that has demonstrated adequate reliability and use among YYA. 25 The rating for each item assesses the frequency of stress symptoms, ranging from 0 (indicating “never”) to 4 (indicating “very often”). The PSS total score ranges from 0 to 56, with higher scores indicating more and/or more frequent symptoms of stress.25 Covariates Covariates included age, sex, race and ethnicity, study site, continuous glucose monitor (CGM) use, insulin regimen, duration of diabetes, health insurance status, household income, and parents’ education levels, considering their highest degree level. All data were self-reported, either by the adult participants or by the parents or guardians of those under 18 years old, where the use of parent-reported data was a standard practice in the study’s design to ensure accurate and reliable information. Race and ethnicity were categorized separately consistent with US census standards: race included (white, black, and other (including Asian, Native American, Pacific Islander, and multiracial)) whereas ethnicity included (Hispanic and non-Hispanic). The assessment of education, based on self-reports from both parents and participants, was similarly divided into individuals with less than a high school education, high school graduates, those with an associate degree, and those with a bachelor’s degree or higher. The educational attainment of adult participants was based on their own self-report, teens (14–17 years) on their own self-reported highest level of schooling completed and children (10–13 years) on their parents’ report.We collected household income (which included salaries, alimony, child support, and other financial support) in categories: less than US$25,000, US$25,000 to US$49,999, US$50,000 to US$74,999, and US$75,000 or greater. Health insurance was categorized into private insurance, Medicare/Medicaid, and a combined category of other types or no health insurance. All analyses were stratified by diabetes type to account for clinical and treatment differences. Within each type, models were adjusted for diabetes duration, insulin regimen (insulin pump vs no insulin pump for T1D and insulin pump, no insulin pump, or no insulin for T2D), and CGM use (yes/no). We recognize that CGM use, and insulin regimens differ between T1D and T2D populations due to variations in clinical indication, access, and insurance coverage. Thus, their inclusion aimed to reduce confounding rather than imply causal relationships with mental health. These variables were added to the multivariable model because they were shown to confound the association between FI and mental health symptoms among YYA with diabetes.8 26 Statistical analysis We used SAS V9.4 (SAS Institute, Cary, North Carolina, USA) for all analyses. A multivariable linear regression was applied to examine the relation between FI patterns and the outcomes of depressive symptoms, anxiety symptoms, and stress symptoms at the third time point with PFS serving as reference category. To evaluate the changes in depressive, anxiety, and stress symptoms based on FI patterns, we employed a multivariable linear regression analysis, calculating the differences in the outcome variables by subtracting the values at the second time point (SFS 1 FU1) from the values at the third time point (SFS 1 FU2). A positive difference implied an increase in the symptoms while a negative difference implied a decrease of symptoms. All models were conducted both unadjusted and adjusted for demographic, socioeconomic, and clinical covariates, including age, sex, race/ethnicity, study site, health insurance status, household income, education level, CGM use, insulin regimen, and duration of diabetes and stratified by diabetes type.We performed further analysis by including the outcome at the second time point as an additional covariate to allow the change in mental health outcomes from the second time point to the third time point to vary with (ie, depend on) the value of the outcome at the second time point.27 This inclusion relaxes the often untrue assumption that a one-unit difference at the beginning of an interval is equivalent to a one-unit difference at the end of the interval while likely increasing statistical power.28 Linear regression models were fitted to estimate the mean difference (β) in depressive, anxiety, and stress symptom scores between categories of FI status, adjusting for relevant covariates. We carefully considered the relevance of each covariate, ensuring that variables related to FI groups were included to account for potential confounding effects. Variables that were not expected to influence this relationship were excluded to maintain model parity. The final set of covariates included in the models was selected based on their theoretical relevance and empirical association with FI and mental health outcomes.Results Among the 844 YYA with diabetes (747 T1D, 97 T2D), 69.4% were classified as PFS; in contrast, 7.7% experienced PFI, and 22.9% experienced IFI. Most participants were ≥18 years old, with a small minority (<23%) younger than 18 at the baseline time point. Among YYA with T1D, 73.1% were in the PFS group, 6.6% in the PFI group, and 20.3% in the IFI; the percentage of females in the PFI group was higher than that in the PFS group as was the percentage identifying as non-Hispanic black and those with an annual household income less than US$25,000 ( table 1). Conversely, among the PFI group, a lower proportion of participants and parents of participants had completed a bachelor’s degree or more, and a lower proportion of participants reported having private health insurance and using an insulin pump or CGM compared with those with T1D with PFS.Table 1Socio-demographic and clinical characteristics of YYAs with diabetes according to food insecurity pattern over time. The SEARCH Food Security Cohort StudyCategorySubcategoryType 1 diabetes, n (%)747 (88.5)Type 2 diabetes, n (%)97 (11.5)PFS546 (73.1)IFI152 (20.3)PFI49 (6.6)PFS40 (41.2)IFI41 (42.3)PFI16 (16.5)SexFemale305 (55.9)85 (55.9)36 (73.5)32 (80.0)33 (80.5)13 (81.3)Male241 (44.1)67 (44.1)13 (26.5)8 (20.0)8 (19.5)3 (18.7)Age (years)Mean (SD)21.4 (4.9)21.6 (4.4)22.2 (4.7)25.9 (4.6)26.0 (3.2)26.5 (4.8)Diabetes duration (years)Mean (SD)11.8 (2.9)11.9 (2.9)12.1 (3.1)11.5 (2.8)11.6 (3.1)12.1 (3.3)Age category<18 years146 (26.7)34 (22.4)7 (14.3)3 (7.5)1 (2.4)1 (6.2)≥18 years400 (73.3)118 (77.6)42 (85.7)37 (92.5)40 (97.6)15 (93.8)Study siteSouth Carolina271 (49.6)73 (48.0)24 (48.9)33 (82.5)31 (75.6)10 (62.5)Colorado164 (30.0)51 (33.6)18 (36.7)6 (15.0)5 (12.2)4 (25.0)Washington111 (20.3)28 (18.4)7 (14.3)1 (2.5)5 (12.2)2 (12.5)RaceWhile433 (79.3)21 (13.8)36 (73.5)8 (20.0)9 (22.0)6 (37.5)Black42 (7.7)17 (11.2)8 (16.4)29 (75.5)26 (63.4)6 (37.5)Others71 (13.0)114 (75.0)5 (10.1)3 (7.5)6 (14.6)4 (25.0)EthnicityHispanic58 (10.6)12 (8.0)3 (6.0)2 (5.0)5 (12.2)3 (18.7)non-Hispanic488 (89.4)140 (92.0)46 (94.0)38 (95.0)36 (87.8)13 (81.3)Parent educationLess than high school graduate12 (2.2)5 (3.3)2 (4.1)1 (2.5)5 (12.2)2 (12.5)High school graduate63 (11.5)28 (18.4)11 (22.5)15 (37.5)12 (29.3)4 (25.0)Associate degree or some college120 (21.9)54 (35.5)17 (34.7)13 (32.5)16 (39.0)5 (31.3)Bachelor’s degree or above351 (64.3)65 (42.8)19 (38.8)11 (27.5)8 (19.5)5 (31.2)Annual household incomeLess than US$25,00075 (13.7)41 (26.9)22 (44.9)17 (42.5)22 (53.7)7 (43.7)US$25,000 to US$49,999104 (19.1)44 (28.9)15 (30.6)15 (37.5)16 (39.0)9 (56.3)US$50,000 to US$74,99988 (16.1)25 (16.5)9 (18.4)5 (12.5)0 (0.0)0 (0.0)Greater than US$75,000279 (51.1)42 (27.6)3 (6.1)3 (7.5)3 (7.3)0 (0.0)Health insurancePrivate insurance457 (83.7)108 (71.1)32 (65.3)21 (52.5)19 (46.3)7 (43.7)Medicare and/or Medicaid57 (10.4)30 (19.7)13 (26.5)10 (25.0)15 (36.6)4 (25.0)Other—no health insurance32 (5.9)14 (9.2)4 (8.1)9 (22.5)7 (17.1)5 (31.3)Insulin regimenInsulin pump361 (66.1)87 (57.2)28 (57.1)1 (2.5)0 (0.0)1 (6.3)No insulin pump185 (33.9)65 (42.8)21 (42.9)17 (42.5)25 (61.0)8 (50.0)No insulin (T2D only)–––22 (55.0)16 (39.0)7 (43.7)CGMYes258 (47.3)45 (29.6)9 (18.4)5 (12.5)10 (24.4)0 (0.0)No288 (52.8)107 (70.4)40 (81.6)35 (87.5)31 (75.6)16 (100.0)The p values for sex, age category, race, ethnicity, education, income, study site, health insurance, CGM, and insulin regimen were calculated using a χ2 test. The p values for age and diabetes duration were calculated using analysis of variance. P value for education, income, insurance, and CGM was <0.05 in participants with T1D. All covariates were measured at the baseline.CGM, continuous glucose monitoring; IFI, intermittent food insecurity; PFI, persistent food insecurity; PFS, persistent food security; T1D, type 1 diabetes; T2D, type 2 diabetes; YYAs, youth and young adults.Among YYA with T2D, 41.2% were in the PFS group, 16.5% in the PFI group, and 42.3% in the IFI; the percentage of females in the PFI group was higher than in the PFS group as was the percentage identifying as non-Hispanic white and those with an annual household income less than US$25,000 (table 1). Among both T1D and T2D participants, at the third time point, the PFI and IFI groups had more unfavorable mental health symptom scores for all three outcomes than the PFS group (table 1).At the third time point, the mean depression, anxiety and stress scores were (14.7, 7.1, and 24.4 for T1D) and (16.9, 7.7, and 25.3 for T2D) (table 2). Based on established scale thresholds, these values correspond to mild-to-moderate depressive symptoms22 (CES-D ≥ 16 indicating clinical significance), mild anxiety (GAD-7 cut-offs of 5, 10, and 15 for mild, moderate, and severe), and moderate perceived stress (PSS-10 scores 14–26 indicating moderate stress). The average changes between the second and third time points were small, suggesting relative stability of symptom levels over time. PFI and IFI were associated with frequency of depressive, anxiety, and stress symptoms at the third time point among those with T1D, even after adjustment for socio-demographic and clinical covariates (table 3). The PFI group exhibited substantially higher average mental health symptom scores than the PFS group, with an adjusted score difference of 12.6 points on the CES-D, 6.5 on the GAD-7, and 9.4 on the PSS. Among those with T2D, PFI and IFI were associated with more depressive symptoms, with a score difference of 7.5 points on the CES-D scale after adjustment for socio-demographic and clinical covariates. In contrast, no association with anxiety or stress symptoms was observed among those with T2D.Table 2Clinical characteristics of YYA with diabetes according to food insecurity pattern over time. The SEARCH Food Security Cohort StudyOutcomePFS, n (%)546 (73.1)IFI, n (%)152 (20.3)PFI, n (%)49 (6.6)Type 1 diabetes  Mental health outcomes  (third time point) mean (SD) Depressive symptoms*12.6 (10.7)18.8 (11.7)25.3 (14.7)Anxiety symptoms*6.1 (5.2)8.6 (5.8)11.7 (6.5)Stress symptoms*22.9 (8.3)26.9 (8.7)32.1 (9.5)  Mental health outcomes  (change between second and third time points) mean (SD)Depressive symptoms0.5 (9.3)0.02 (11.1)0.7 (8.46)Anxiety symptoms0.2 (4.8)−0.01 (4.4)0.3 (4.5)Stress symptoms−0.1 (7.1)−0.6 (7.9)−0.1 (5.9)OutcomePFS40 (41.2)IFI41 (42.3)PFI16 (16.5)Type 2 diabetes  Mental health outcomes  (third time point) mean (SD)Depressive symptoms*12.5 (10.3)18.6 (10.4)22.1 (10.1)Anxiety symptoms*5.9 (6.1)8.5 (6.8)10.6 (5.3)Stress symptoms*22.4 (8.3)27.3 (8.3)28.2 (7.6)  Mental health outcomes  (change between second and third time points) mean (SD)Depressive symptoms−0.4 (9.7)−0.9 (11.7)−1.6 (9.5)Anxiety symptoms−0.4 (5.9)0.3 (5.6)0.3 (6.6)Stress symptoms−0.1 (5.5)−0.6 (7.8)−2.9 (7.4)The p values for mental health outcomes were calculated using analysis of variance.*Represents associations with p<0.05. IFI, intermittent food insecurity; PFI, persistent food insecurity; PFS, persistent food security; YYAs, youth and young adults.Table 3Associations of food insecurity patterns with depressive symptoms, anxiety, and stress symptoms at the third time point* in YYA with diabetes. The SEARCH Food Security Cohort StudyUnadjustedAdjusted†β‡P valueβP valueDepressive symptomsType 1 diabetes  PFI vs PFS12.7<0.000112.6<0.0001  IFI vs PFS6.0<0.00016.1<0.0001Type 2 diabetes  PFI vs PFS9.40.00267.50.0432  IFI vs PFS6.10.00874.70.0724Anxiety symptomsType 1 diabetes  PFI vs PFS6.5<0.00016.5<0.0001  IFI vs PFS2.9<0.00012.9<0.0001Type 2 diabetes  PFI vs PFS3.50.07021.30.5628  IFI vs PFS2.30.09460.70.6377Stress symptomsType 1 diabetes  PFI vs PFS9.9<0.00019.4<0.0001  IFI vs PFS4.3<0.00014.2<0.0001Type 2 diabetes  PFI vs PFS4.00.12574.00.2015  IFI vs PFS4.20.02603.90.0691*Represents the third and final time point of the cohort.†Adjusted: Using age, sex, race, ethnicity, study site, continuous glucose monitor usage, insulin regimen, duration of diabetes, health insurance, annual household income, and parental education level.‡β: Unstandardized regression coefficient, representing the estimated mean difference in mental health scale scores (CES-D, GAD-7, PSS-10) by food insecurity category.CES-D, Center for Epidemiologic Studies Depression Scale; GAD-7, 7-item Generalized Anxiety Disorder; IFI, intermittent food insecurity; PFI, persistent food insecurity; PFS, persistent food security; PSS, Perceived Stress Scale; YYAs, youth and young adults.The average change of depressive, anxiety, and stress scores between the second and third time points were (0.4, 0.2, –0.2 for T1D) and (−0.9, –0.2, –0.9 for T2D). In T1D and T2D, FI status was initially not associated with changes in depressive, anxiety, or stress symptoms after adjusting for socio-demographic and clinical factors (table 4). Including the respective outcome’s value at the second time point as a covariate, however, revealed a positive association between PFI and change in all three mental health outcomes in YYA with T1D, such that those with PFI had greater increases in depressive, anxiety, and stress symptoms than those with PFS. Similar associations were observed for depressive symptoms and anxiety for the comparison between IFI and PFS among those with T1D, although at a lower magnitude. Including the outcome value at the second time point in the models for YYA with T2D did not alter the results.Table 4Associations between food insecurity patterns and changes* in depressive symptoms, anxiety, and stress in YYA with diabetes. The SEARCH Food Security Cohort StudyChange unadjustedChange adjusted†Change adjusted‡β§P valueβP valueβP valueDepressive symptomsType 1 diabetes  PFI vs PFS0.20.9170.70.6545.80.001  IFI vs PFS−0.80.412−0.30.7892.50.008Type 2 diabetes  PFI vs PFS−1.60.6432.10.5694.60.197  IFI vs PFS0.50.8312.60.3063.60.129Anxiety symptomsType 1 diabetes  PFI vs PFS0.10.9230.30.6522.70.001  IFI vs PFS−0.20.578−0.10.7891.00.023Type 2 diabetes  PFI vs PFS−0.40.8180.30.8721.40.482  IFI vs PFS1.10.4210.60.6680.40.789Stress symptomsType 1 diabetes  PFI vs PFS−0.10.9670.10.9363.40.005  IFI vs PFS−0.20.422−0.40.6051.30.066Type 2 diabetes  PFI vs PFS−3.60.099−2.00.4270.80.732  IFI vs PFS−0.10.937−0.20.9140.90.603*Represents the difference between the second and third time points within the cohort.†Adjusted: Using age, sex, race, ethnicity, study site, continuous glucose monitor usage, insulin regimen, duration of diabetes, health insurance, annual household income, and parental education level.‡Adjusted: Using all the mentioned covariates plus the second time point as a covariate.§β: Unstandardized regression coefficient, representing the estimated mean difference in mental health scale scores (CES-D, GAD-7, PSS-10) by food insecurity category.CES-D, Center for Epidemiologic Studies Depression Scale; GAD-7, 7-item Generalized Anxiety Disorder; IFI, intermittent food insecurity; PFI, persistent food insecurity; PFS, persistent food security; PSS, Perceived Stress Scale; YYAs, youth and young adults.Discussion A substantial number of YYA with T1D and T2D in the study sample experienced FI. Specifically, 6.6% of YYA with T1D and 16.5% of those with T2D exhibited persistent FI, and 20.3% of YYAs with T1D and 42.3% of those with T2D experienced IFI over time. These longitudinal results extend findings from earlier cross-sectional work on prevalence of FI and illustrate the intermittent nature of FI that is the dominant experience for those who report any FI. 3 29 We observed more PFI and IFI among YYAs with T2D compared with those with T1D, highlighting a distinct FI pattern in PFI and IFI, which is consistent with prior population-based work showing higher FI prevalence among individuals with T2D.4 30 31 Differences in access to T2D monitoring or lower income may also influence psychosocial outcomes. Although we adjusted for CGM use and insulin regimen, residual confounding by insurance coverage or healthcare access may persist.FI experienced as either PFI or IFI was associated with greater depressive, anxiety, and stress symptoms for YYA with T1D. The magnitude of the associations was correlated with the severity of the FI experience, with larger associations found for the PFI group than for the IFI group. Similarly, PFI and IFI were associated with an increase in these symptoms from the second to the third time point after adjustment for the second time point. Findings from a previous study showed that people with diabetes have increased depressive and anxiety symptoms, and poorer glycemic control, leading to higher rates of complications.26 Additionally, FI has an influence on mental health among various age, gender, and ethnicity groups, where the impact is evident among individuals who self-report lower general health status.32 33 Our work extends previous work by establishing a connection between particular patterns of FI and mental health consequences, thus addressing a gap in existing knowledge.In the general US population, approximately 13–15% of young adults experience FI, compared with substantially higher rates among those with diabetes.4 Roughly 14% of FI youth with either type of diabetes had mild depressive symptoms, while 8.6% had moderate or severe symptoms.34 Among FI young adults, those with T1D had average depression scores that were 7.4 points higher than their food-secure counterparts, whereas those with T2D had scores that were 4.8 points higher.35 Observed mean depressive, anxiety, and stress scores generally reflected mild-to-moderate symptoms, consistent with prior research showing elevated but subclinical psychological distress among youth and young adults with diabetes.36 Consideration of the association between FI and mental health outcomes needs to distinguish between those with and without diabetes.7 37 While one study focused on broad demographic groups, such as adolescents and adults with mental health issues, another specifically examined the links between FI and mental health in people with T2D.7 37 While these studies demonstrated that FI is associated with greater mental health issues in individuals with and without diabetes, other studies in populations without diabetes have been inconclusive.38 39 Together with our own findings supporting an association, the body of evidence to date suggests that FI and mental health relationships may be more pronounced among individuals with diabetes, indicating a potentially greater vulnerability in this population.40 41 Depressive disorder, in turn, affects physical well-being, self-care and functioning, resulting in considerable impairment across multiple domains and economic burdens.40 Individuals with significant depressive symptoms may struggle with employment and financial management, potentially leading to a state of FI or exacerbating a tenuous living situation.6 42 Depressive, anxiety and stress symptoms were worse among people with very low food security compared with those who are food secure in low-income populations.38 This emphasizes that severe FI has a greater association with more detrimental mental health, suggesting that addressing food security is crucial in mitigating these negative outcomes.FI can lead to poor dietary choices, depression, stress, and anxiety, which can in turn worsen diabetes management.43 Our study provides support that addressing FI is essential in individuals living with diabetes, demonstrating that FI is associated with frequency of mental health symptoms over time.33 44 A recent study highlighted the link between FI and mental health issues, specifically diabetes distress, independent of changes in the individuals’ diabetes condition throughout the same time frame.45 The study explored American Indian adults with T2D, while our study represented diverse YYA with T1D and T2D in three locations of the continental USA. In addition, their follow-up period was 6 months, while ours was from 18 to 27 months.While our analyses focused on overall associations between FI and mental health, there is evidence from intervention research that specific improvements in these domains may also confer benefits. A study among adults with T2D has shown that increased access to fresh produce and diabetes self-management support is associated with improved mood and metabolic outcomes.8 38 46 For instance, policies that increase access to nutritious food through food assistance programs and subsidies can mitigate FI.47 To address the link between poor mental health and inadequate diabetes management, it is crucial to expand mental health services, integrate mental health screenings into routine diabetes care, and train providers on this intersection.7 48 These actions hold the potential to improve health outcomes and reduce healthcare costs associated with managing diabetes complications exacerbated by poor mental health and FI.49 The primary limitation of this study is its cohort, which included only YYAs living with T1D and T2D. SFS 1 data collection related to FU2 had been initiated shortly before the onset of the COVID-19 pandemic occurred and may have affected access to resources as well as the stability of reported mental health between SFS 1 FU1 and FU2. Ongoing follow-up of the SFS cohort may allow examination of pandemic-related and post-pandemic trends. Because SFS follow-up occurred at three SEARCH sites, contextual factors (such as local healthy food availability, transportation barriers, and state-level social safety net policies) may shape both the prevalence of FI and its relation to mental health. Almost one-third of participants were lost to follow-up from baseline; they were more likely to be male, from South Carolina, and less likely to be Hispanic (online supplemental table 1). Also, the sample size of T2D individuals was noticeably small, which may limit generalizability. Data collection timing likely had minimal effect due to the predominantly young adult sample.At present, there is no chronic disease or nutrition-sensitive disease-specific validated version of the HFSSM. We have evaluated the HFSSM in our study of people with diabetes and found that there is room for improvement in its metrics for this population.50 The use of the general-population HFSSM may not fully reflect diabetes-specific challenges such as the need for daily access to appropriate foods or non-financial barriers like food environment and storage capacity. The 12-month reference period may also obscure intermittent fluctuations in food access. We are working on this issue through the SFS 2 project, which is using a modified questionnaire within a shorter reference period (1 month) and additional items addressing the daily and dietary stability challenges specific to diabetes.The study design does not allow for full generalization to all individuals with youth-onset diabetes in the USA. Although measures were administered in age-appropriate ways, the wide age range introduces potential heterogeneity in both experience and reporting of FI and mental health. Additionally, caregiver stress or mental health could influence both reporting of FI and diabetes management, potentially biasing observed associations between FI and mental health outcomes. The strengths of this study include a longitudinal design that spans several years (2016–2022), a large and racially and ethnically and geographically diverse group of participants, and the use of reliable and validated measures that helped reduce recall bias to assess both FI and mental health.To disrupt the cycle of FI, mental health issues, and diabetes, integrated strategies, including routine screenings, direct food supports, and behavioral interventions are needed. While many system-level strategies, such as integrated behavioral health and social work linkage, are broadly applicable, tailoring interventions is crucial. For T1D, emphasis should be placed on reliable carbohydrate access, insulin and CGM affordability, and hypoglycemia prevention. For T2D, interventions may be most effective when addressing comorbid conditions, medication access, and structured lifestyle supports in combination with food provision. These integrated models of care, which can also include practical approaches like clinic-based food programs and supplemental nutrition assistance program (SNAP), require managed care providers, including Medicaid plans, to cover and implement them.Conclusion The study identified a previously unrecognized link between prolonged exposure to FI and increased incidence of mental health issues. Subsequent research should prioritize interventions that address FI and evaluate their effectiveness in enhancing both physical and psychological well-being in YYA with diabetes. These interventions should be designed to not only address FI, such as access to healthy and balanced foods that are essential for managing diabetes but also provide comprehensive support that includes mental health services, education, and resources for better diabetes management in YYA with diabetes.",
  "title": "Food insecurity patterns and mental health among youth and young adults with diabetes",
  "uid": "3b04939f-f7aa-5445-bacc-e430dcc44533"
}