Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 May 17.
Published in final edited form as: Am J Prev Med. 2024 Oct 1;68(2):217–226. doi: 10.1016/j.amepre.2024.09.019

Loneliness, Discrimination, Stress and Type 2 Diabetes Risk in Young Adults

Caleb Harrison 1,2, Maya I Ragavan 3, Margaret F Zupa 4, Xu Qin 5, Vicki Helgeson 6, Mary Ellen Vajravelu 1,2
PMCID: PMC12084895  NIHMSID: NIHMS2080909  PMID: 39362614

Abstract

Introduction:

The aim of this study was to determine the associations between type 2 diabetes (T2D) or prediabetes and loneliness and related social experiences in young adults, a population at increasingly high risk of T2D.

Methods:

This was a cross-sectional analysis using data from adults ages 18–35 enrolled in the All of Us Research Program. Exposures included loneliness, social support, discrimination, neighborhood social cohesion, and stress, measured by standardized surveys. The main outcome was T2D or prediabetes by self-report or linked health record. Logistic regression determined odds of T2D/prediabetes for each survey measure, adjusting for age, sex, race or ethnicity, income, and family history. Latent class analysis (LCA) evaluated clustering of social experiences. Data was collected from 2018–2022 and analyzed May 2023-June 2024.

Results:

The cohort included 14,217 young adults (28.2 +/− 4.4 years; 70.3% (n=9,792) women; 64.1% (n=9,111) White, 10.6% (n=1,506) Hispanic, 5.7% (n=806) Black, 9.1% (n=1,299) multiracial). Overall, 5.5% (n=777) had either prediabetes or T2D. The two highest loneliness quartiles were associated with increased odds of prediabetes/T2D (Q3: OR 1.42 [95% CI 1.15–1.76]; Q4: 1.75 [95% CI 1.43–2.16]). Greater stress and discrimination and lower social support and neighborhood social cohesion were also associated with increased odds of prediabetes/T2D. LCA revealed three distinct phenotypes, with elevated odds of prediabetes/T2D in the two with the most adverse social profiles (OR 2.32 [95% CI 1.89–2.84] and OR 1.28 [95% CI 1.04–1.58]).

Conclusions:

Loneliness and related experiences are strongly associated with T2D and prediabetes in young adults. Whether these factors could be leveraged to reduce T2D risk should be investigated.

Introduction

Type 2 diabetes (T2D) in adolescents and young adults, occurring disproportionately in individuals marginalized by race, ethnicity, and socioeconomic status,1,2 is predicted to increase by nearly 700% by 2060.3 T2D that begins in adolescence or early adulthood is associated with numerous severe outcomes, including an 80% cumulative incidence of microvascular complications within 15 years of diagnosis.4 Macrovascular complications are more common in early-onset T2D,5 contributing to a 15-year reduction in life expectancy.6

Although early glycemic control may improve outcomes,7 for adults younger than 35 years, current guidelines recommend no routine screening8 for T2D, or screening in the setting of risk factors such as overweight or obesity, first-degree relative with diabetes, metabolic complications, and higher-risk race and ethnicity.9 However, additional risk factors may be important to consider, including social determinants of health (SDOH) such as loneliness, social support, and discrimination,10–12 each of which is associated with incidence of T2D in older adults. Whether addressing these factors could also reduce T2D risk in young adults, a vulnerable population characterized by important transitions in social support, is unknown.

Young adults report high levels of loneliness, exceeding those reported by adults older than 65.13 As posited in Iovino et al’s framework, loneliness is preceded by factors such as low social support and discrimination and is a determinant of chronic illnesses such as T2D as well as adverse psychosocial outcomes including perceived stress, anxiety, and depression.14 In this model, greater social support can be protective, as it is associated with healthy self-care,15 adaptive coping,16 and successful diabetes self-management.17 Additional SDOH including income, education, and occupation may also influence the experience of loneliness and social support and are associated with T2D incidence.10

This study evaluated the relationship between T2D risk and loneliness and related psychosocial experiences in young adult participants in the National Institutes of Health-funded All of Us Research Program. Due to the interrelatedness of loneliness, social support, social cohesion, stress, and discrimination, patterns of combined experiences across these measures were examined using latent class analysis. The study hypothesis was that adverse psychosocial experiences, both individually and in combination, would be associated with increased odds of prediabetes and T2D, independent of other recognized risk factors.

Methods

Study Sample

The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. This study was a cross-sectional analysis of survey, anthropometric, and electronic health record (EHR) data collected by the All of Us Research Program. The All of Us Research Program enrolled its first participants in May 2018, and recruitment was ongoing at the time the data for this study was obtained; All of Us seeks to recruit more than a million adults with an emphasis on those historically underrepresented in medical research.18 Individuals anywhere in the U.S. can enroll online or join via a participating health care provider organization, including community health and regional medical centers and those operated by the Department of Veterans Affairs. Participants complete a baseline sociodemographic survey and can optionally complete personal and family health history and overall health, lifestyle, and SDOH surveys. Participants have the option to visit an All of Us health center for physical assessments and specimen collection and to share EHR data. All of Us has a single Institutional Review Board, which oversees the protocol and informed consent. For this study, the University of Pittsburgh Institutional Review Board determined the use of de-identified data to be exempt from institutional oversight.

The cohort was identified via the All of Us Researcher Workbench using data from the Controlled Tier Curated Data Repository (v7). Participants aged 18–34 years (consistent with the U.S. Census Bureau definition of young adulthood19) with a completed SDOH survey were included, excluding those with type 1 diabetes (self-reported and via SNOMED/ICD10CM). From this sample of 16,669 participants, 14,217 had complete SDOH survey data and therefore formed the analytic sample (CONSORT diagram, Appendix Figure 1).

Measures

This analysis used 5 scales20–24 within the All of Us SDOH Survey25 to assess the association between adverse psychosocial experiences and prediabetes or T2D in young adulthood. The included scales were: 1) the UCLA Loneliness Scale (ULS-8), measuring companionship and social isolation; 2) the modified Medical Outcomes Study social support scale (mMOS-SS), measuring perceived social support; 3) the Cohen Perceived Stress Scale (CPSS), which measures everyday thoughts and feelings related to stress; 4) the Everyday Discrimination Scale (EDS) and Discrimination in Health Care Settings Scale (DMS), measuring experiences of discrimination in everyday life and in health care settings, respectively; and 5) the Social Cohesion Scale (SCNS), measuring neighborhood level social organization. Full survey descriptions are in Appendix Table 1. Due to skewed distributions and to facilitate interpretation of findings, responses were categorized into quartiles.

Next, latent class analysis (LCA) was performed to explore item response patterns to the SDOH surveys listed above using the All of Us Researcher Workbench Notebook’s and R package poLCA.26 During model selection, three candidate models were considered (number of classes 2–4). The Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) were used to examine model fit. Although three and four-class models had similar AIC and BIC values, the three-class model was chosen based on scientific and clinical judgement and parsimony. Classes were assigned based on modal posterior probabilities, coded as a categorical variable 1–3.

The main outcome was presence of either prediabetes or T2D in linked medical record data or by self-report via All of Us surveys. The All of Us Researcher Workbench Cohort Builder and Concept Set functionalities were used to select participants with available EHR data and a previous prediabetes/T2D diagnosis, defined via SNOMED and ICD10 codes (Prediabetes: 714628002, R73.03, T2D: 44054006, E11). Concept Sets were created to select participants who responded “self” to the following questions, “Including yourself, who in your family has had prediabetes?” and “Including yourself, who in your family has had type 2 diabetes?” Those who did not respond “self” to either and did not have the corresponding EHR codes were assumed to not have prediabetes or T2D. Due to the anticipated small proportion of this young adult cohort having T2D, the primary analysis used a combined outcome of prediabetes or T2D; these were evaluated as separate outcomes in a sensitivity analysis. Individuals with both diagnoses were classified as having T2D.

Covariates in multivariable models included age, sex assigned at birth (male or female), self-reported racial and ethnic identity (White, Asian, Black, Hispanic, Middle Eastern or North African [MENA], Native Hawaiian or Pacific Islander [NHPI], and Multiracial/Multiethnic), gender identity (man, woman, non-binary, transgender, multiple, other), annual income (<$50,000, $50–100,000, $100–150,000, and >$150,000), family history of prediabetes or T2D, and educational attainment (< high school, high school/GED, college for 1–3 years, and college graduate or advanced degree). BMI was not included as a covariate due to the high amount of missing data (n=4,232, 29.8% of cohort missing).

Statistical Analysis

Data was analyzed in the All of Us Researcher Workbench using both R and Python. Demographic, clinical, and social characteristics were reported using summary statistics. Multivariable logistic regression was used to determine associations between individual SDOH scales or latent classes and prediabetes/T2D, adjusting for covariates. For each model using an individual survey, the reference group was set as the most favorable category (e.g., least lonely quartile, never experiencing discrimination, etc.). For models using latent class membership as the primary exposure, the class with the most favorable combination of responses was used as the reference group (e.g., lowest loneliness, greatest social support, least discrimination, etc). For all analyses, a p-value of < 0.05 was used to define significance.

Results

The sample included n=14,217 young adults 18–34 years old (mean age 28.2, SD 4.4 years). Most were women (70.3%), Non-Hispanic White (64.1%), not currently married (53.8%), and college graduates (62.7%). A total of 777 (5.5%) had either T2D (n=262, 1.8%) or prediabetes (n=515, 3.6%). The cohort with available BMI (n = 9,985) had a mean BMI of 28.0 kg/m2 (SD 7.9). Young adults with, versus without, prediabetes/T2D were an average of 1.5 years older and differed in race and ethnicity, sex and gender, income, educational attainment, and marital status (Table 1).

Table 1.

Cohort Characteristics, overall and by prediabetes/T2D status.

Characteristics Total Sample (n=14,217) Either Prediabetes/T2D (n=777) No Prediabetes/T2D (n=13,440) p-value
Age (mean, SD) (n=14,217) 28.17 (4.43) 29.52 (4.10); n=777 28.09 (4.44); n=13,440 <0.001
BMI, kg/m2 (mean, SD) (n=9,985) 28.01 (7.89); n=581 37.69 (9.97); n=581 27.42 (7.34); n=9,404 <0.001
Sex at Birth (n=14,217) n=14,217 n=777 n=13,175
Male 3538 (24.89%) 137 (17.63%) 3401 (25.31%) <0.001
Female 10391 (73.09%) 617 (79.41%) 9774 (72.72%) <0.001
None, Other, or Skip 288 (2.03%) 23 (2.96%) 265 (1.97%) <0.001
Race and Ethnicity (n=14,217) a n=777 n=13,440
Asian (5) 940 (6.61%) 35 (4.5%) 905 (6.73%) <0.001
Black (4) 806 (5.67%) 90 (11.58%) 716 (5.33%) <0.001
Hispanic (2) 1506 (10.59%) 127 (16.34%) 1379 (10.26%) <0.001
MENA (7) <100 (<0.70%) <20 (<2.57%) <100 (<0.70%) <0.001
Multiple (3) 1299 (9.14%) 82 (10.55%) 1217 (9.06%) <0.001
NHPI (8) <20 (<0.14%) <20 (<2.57%) <20 (<0.15%) <0.001
White (1) 9111 (64.09%) 404 (51.99%) 8707 (64.78%) <0.001
None, Skip, or Prefer Not Answer 452 (3.18%) 37 (4.76%) 415 (3.09%) <0.001
Gender Identity (n=13,932) a n=753 n=13,179
Man 3398 (24.39%) 127 (16.87%) 3271 (24.82%) <0.001
Woman 9792 (70.28%) 560 (74.37%) 9232 (70.05%) <0.001
Non-Binary 278 (2.00%) 30 (3.98%) 248 (1.88%) <0.001
Multiple 333 (2.39%) 23 (3.05%) 310 (2.35%) <0.001
Transgender 61 (0.44%) <20 (<2.66%) <60 (<0.45%) <0.001
Other 70 (0.50%) <20 (<2.66%) <70 (<0.50%) <0.001
Household Income (n=14,217) n=777 n=13,440
<$50,000 5514 (38.78%) 405 (52.12%) 5109 (38.01%) <0.001
$50,000–100,000 3816 (26.84%) 186 (23.94%) 3630 (27.01%) <0.001
$100,000–150,000 1830 (12.87%) 58 (7.46%) 1772 (13.18%) <0.001
>$150,000 1596 (11.23%) 38 (4.89%) 1558 (11.59%) <0.001
Skip or Prefer Not Answer 1461 (10.28%) 90 (11.58%) 1371 (10.2%) <0.001
Education (n=14,217) n=777 n=13,440
Less than high school degree or equivalent 253 (1.78%) 31 (3.99%) 222 (1.65%) <0.001
Twelve or GED 1423 (10.01%) 102 (13.13%) 1321 (9.83%) <0.001
College One to Three 3211 (22.59%) 263 (33.85%) 2948 (21.93%) <0.001
College graduate or advanced degree 8915 (62.71%) 352 (45.3%) 8563 (63.71%) <0.001
Skip or Prefer Not Answer 415 (2.92%) 29 (3.73%) 386 (2.87%) <0.001
Birthplace (n=13,938) n=755 n=13,183
USA 12439 (87.49%) 681 (90.20%) 11758 (89.19%) 0.40
Other 1499 (10.54%) 74 (9.80%) 1425 (10.81%) 0.40
Marital Status (n=13,804) n=747 n=13,057
Married/Lives with partner 6376 (46.19%) 329 (44.04%) 6047 (46.31%) 0.01
Never married 6893 (49.93%) 374 (50.07%) 6519 (49.93%) 0.01
Divorced, separated, or widowed 535 (3.88%) 44 (5.89%) 491 (3.76%) 0.01
Open in a new tab

Note: BMI: body mass index; MENA: Middle Eastern/North African.

a

Categories with fewer than 20 individuals are not reported to minimize risk of deidentification. Boldface indicates statistical significance (p<0.05).

Increased odds of prediabetes/T2D was observed for individuals who reported higher loneliness, stress, discrimination, low social support, and low neighborhood social cohesion, after adjustment for age, sex, race and ethnicity, income, and family history (Figure 1). Compared with the least lonely quartile, participants with the highest two quartiles of loneliness scores had increased odds of prediabetes/T2D, with a gradient of response (Q3 OR 1.42 [95% CI 1.15–1.76], Q4 OR 1.75 [95% CI 1.43–2.16]). The two lowest quartiles of social support were associated with higher odds of prediabetes/T2D (Q1 OR 1.66 [95% CI 1.32–2.09]; Q2 OR 1.42 [95% CI 1.12–1.81]), and the lowest quartile of neighborhood social cohesion was associated with higher odds of prediabetes/T2D (OR 1.47 [95% CI 1.16–1.85). The top 3 quartiles of perceived stress were each associated with higher odds of prediabetes/T2D versus the lowest quartile (Q2 OR 1.34 [95% CI 1.05–1.72], Q3 OR 1.80 [95% CI 1.43–2.28], Q4 OR 2.52 [95% CI 2.01–3.15]). Experiences of less-than-full (versus full) social support were associated with higher odds of prediabetes/T2D (OR 1.41 [95% CI 1.14–1.74]). Participants who reported any (versus no) experiences of discrimination in everyday life (OR 1.33 [95% CI 1.03–1.70]) or in health care settings (OR 1.69 [95% CI 1.37–2.09]) had increased odds of prediabetes/T2D.

Figure 1.

Figure 1.

Open in a new tab

Odds ratios (OR) with 95% confidence intervals (CI) from multivariable logistic regression models, adjusted for age, sex at birth, race and ethnicity, and household income, and family history of T2D/prediabetes, evaluating the independent associations between each social risk factor and odds of prediabetes or T2D in young adult participants of the All of Us Research Program. Odds ratios are depicted by quartile of response for each scale, with the most favorable quartile used as the reference group (e.g., least loneliness).

Results of sensitivity analyses were similar to the primary combined outcome models. Higher odds of each disease state (prediabetes, T2D) were again demonstrated in the setting of greater loneliness, stress, and discrimination and lower social support. Lower neighborhood social cohesion was associated with higher odds of T2D but did not meet statistical significance for prediabetes (Appendix Tables 2 and 3).

Three distinct latent classes emerged, with Class 1 (n=3,898, 27.4% of cohort) having the highest reported loneliness, discrimination, and stress and the lowest social support and neighborhood social cohesion. Class 3 (n=4,439, 31.2% of cohort) had the inverse pattern, with Class 2 (n=5,880, 41.4%) representing intermediate scores for all scales. Notably, Class 1 was characterized by younger age, lower educational attainment, lower household income, and lowest proportion married or living with a partner (each p<0.001). Class 1 had approximately twice the proportion of Non-Hispanic Black participants and significantly more gender minority (non-binary, multiple gender) participants versus Class 3 (Table 2).

Table 2.

Cohort Characteristics, by latent class of social risk.

Characteristics Highest Social Risk, LC1 (n=3,898, 27.4%): Either T2D or prediabetes in 357 (9.2%) Intermediate Social Risk, LC2 (n=5,880, 41.4%): Either T2D or prediabetes in 269 (4.6%) Lowest Social Risk, LC3 (n=4,439, 31.2%): Either T2D or prediabetes in 151 (3.4%) p-value
Loneliness (ULS-8) 58.3 (IQR 50, 70.8) 33.3 (IQR 25, 45.8) 16.7 (IQR 8.3, 29.2) <0.001
Social Support (mMOS-SS) 3.1 (IQR 2.4, 3.8) 4.4 (IQR 3.9, 4.8) 4.9 (IQR 4.3, 5) <0.001
Stress (CPSS) 35 (IQR 31, 39) 28 (IQR 25, 32) 22 (IQR 18, 26) <0.001
Everyday Discrimination (EDS) 2.9 (IQR 2.2, 3.6) 2.1 (IQR 1.8, 2.6) 1.2 (IQR 1, 1.6) <0.001
Discrimination in Medical Settings (DMS) 2.3 (IQR 1.6, 2.9) 1.7 (IQR 1.3, 2) 1 (IQR 1, 1.3) <0.001
Neighborhood Social Cohesion (SCNS) 3.3 (IQR 2.8, 3.8) 3.8 (IQR 3.3, 4) 4 (IQR 3.5, 4.5) <0.001
Age (mean, SD) (n=14,217) n=3,898 n=5,880 n=4,439 <0.001
27.86 (4.61) 28.12 (4.36) 28.52 (4.33)
BMI, kg/m2 (mean, SD) (n=9,985) n=2,582 n=4,184 n=3219 <0.001
29.93 (9.22) 27.52 (7.41) 27.13 (7.03)
Sex at Birth (n=14,217) n=3,818 n=5,761 n=4,350
Male 940 (24.11%) 1207 (20.53%) 1391 (31.34%) <0.001
Female 2878 (73.83%) 4554 (77.45%) 2959 (66.66%) <0.001
None, Other, or Skip 80 (2.05%) 119 (2.02%) 89 (2.0%) <0.001
Race and Ethnicity (n=14,217) a n=3,898 n=5,880 n=4,439
Asian (5) 255 (6.54%) 443 (7.53%) 242 (5.45%) <0.001
Black (4) 366 (9.39%) 227 (3.86%) 213 (4.8%) <0.001
Hispanic (2) 415 (10.65%) 551 (9.37%) 540 (12.16%) <0.001
MENA (7) <40 (<1.03%) <30 (<0.51%) <40 (<0.90%) <0.001
Multiple (3) 421 (10.8%) 533 (9.06%) 345 (7.77%) <0.001
NHPI (8) <20 (<0.51%) <20 (<0.34%) <20 (<0.45%) <0.001
White (1) 2252 (57.77%) 3918 (66.63%) 2941 (66.25%) <0.001
None, Skip, or Prefer Not Answer 155 (3.98%) 174 (2.96%) 123 (2.77%) <0.001
Gender Identity (n=13,932) a n=3,816 n=5,762 n=4,354
Man 857 (22.46%) 1164 (20.20%) 1377 (31.63%) <0.001
Woman 2559 (67.06%) 4317 (74.92%) 2916 (66.97%) <0.001
Non-Binary 152 (3.98%) 99 (1.72%) 27 (0.62%) <0.001
Multiple 170 (4.46%) 140 (2.43%) 23 (0.53%) <0.001
Transgender 41 (1.07%) <20 (<0.35%) <20 (<0.46%) <0.001
Other 37 (0.97%) <30 (<0.52%) <20 (<0.46%) <0.001
Household Income (n=14,217) n=3,898 n=5,880 n=4,439
<$50,000 2155 (55.28%) 2021 (34.37%) 1338 (30.14%) <0.001
$50,000–100,000 826 (21.19%) 1762 (29.97%) 1228 (27.66%) <0.001
$100,000–150,000 260 (6.67%) 855 (14.54%) 715 (16.11%) <0.001
>$150,000 180 (4.62%) 695 (11.82%) 721 (16.24%) <0.001
Skip or Prefer Not Answer 477 (12.24%) 547 (9.3%) 437 (9.84%) <0.001
Education (n=14,217) n=3,898 n=5,880 n=4,439
Less than high school degree or equivalent 144 (3.69%) 47 (0.8%) 62 (1.4%) <0.001
Twelve or GED 617 (15.83%) 355 (6.04%) 451 (10.16%) <0.001
College One to Three 1252 (32.12%) 1252 (21.29%) 707 (15.93%) <0.001
College graduate or advanced degree 1768 (45.36%) 4063 (69.1%) 3084 (69.48%) <0.001
Skip or Prefer Not Answer 117 (3.0%) 163 (2.77%) 135 (3.04%) <0.001
Birthplace (n=13,938) n=3,822 n=5,763 n=4,354
USA 3442 (90.06%) 5194 (90.13%) 3803 (87.35%) <0.001
Other 379 (9.92%) 569 (9.87%) 551 (12.66%) <0.001
Marital Status (n=13,804) n=3,772 n=5,727 n=4,305
Married/Lives with partner 1261 (33.43%) 2792 (48.75%) 2323 (53.96%) <0.001
Never married 2278 (60.39%) 2771 (48.39%) 1844 (42.83%) <0.001
Divorced, separated, or widowed 233 (6.18%) 164 (2.86%) 138 (3.21%) <0.001
Open in a new tab

Note: BMI: body mass index; MENA: Middle Eastern/North African.

a

Categories with fewer than 20 individuals are not reported to minimize risk of deidentification. Additional data was obscured to prevent back-calculation of values. Boldface indicates statistical significance (p<0.05).

Class 1 had the highest unadjusted proportion of individuals with prediabetes (6.0%, n=234) and T2D (3.2%, n=123), compared with Class 2 (prediabetes: 3.3%, n=192; T2D: 1.3%, n=77) and Class 3 (prediabetes: 2.0%, n=89; T2D: 1.4%, n=62) (each p<0.001). In models adjusted for age, sex, race and ethnicity, family history of prediabetes and T2D, and income, young adults in Classes 1 (highest social risk) and 2 (intermediate social risk) had significantly elevated odds of prediabetes/T2D, versus Class 3 (Class 1: OR 2.32 [95% CI 1.89–2.84], Class 2: OR 1.28 [95% CI 1.04–1.58]). In sensitivity analysis, prediabetes remained significantly more likely among Classes 1 and 2 than Class 3 (Class 1: OR 2.60 [95% CI 2.01–3.37]; Class 2: OR 1.52 [95% CI 1.17–1.97]). Though odds of T2D was still higher for Class 1 versus Class 3 (OR 1.68 [95% CI 1.22–2.33]), Class 2 had similar odds as Class 3 (OR 0.91 [95% CI 0.65–1.29].

Discussion

In this cross-sectional study of young adults from the All of Us Research Program, adverse psychosocial experiences, both individually and combined as latent classes of social risk phenotypes, were associated with increased odds of prediabetes and T2D. These findings align with the framework by Iovino proposing that loneliness, low social support, low social cohesion, and experiences of discrimination are key drivers of clinical outcomes.14 As loneliness and T2D are both increasingly common in young adults,27,28 these findings have important implications for efforts to reduce risk of T2D onset and to improve prediction of T2D risk in this unique and vulnerable population.

While previous research has evaluated the relationships between loneliness and social support experiences in T2D, the population of interest has primarily been older adults. A study of participants in the Jackson Heart Study and others have demonstrated associations between increased social cohesion and support with lower incidence of T2D and better glycemic control and diabetes self-management,12,29 while loneliness has been shown to double the odds of developing T2D.30 However, the generalizability of these findings to young adults cannot be assummed due to distinct changes in social support systems and connection to the healthcare system that occur during the transitional period between adolescence and adulthood.31 Emergence into young adulthood is often characterized by loneliness and stress and is a time of reductions in parental support and increases in peer or romantic relationships, which have both been associated with diabetes management.31,32 This study highlights that in a young adult population, adverse psychosocial experiences that commonly arise during this transitional period are associated with prediabetes and T2D.

This study identified distinct social risk groups via latent class analysis, with a gradient of prediabetes/T2D risk. This approach overcomes the limitation of considering each factor independently while offering insights into the potential ability to target clusters of social risk to reduce T2D risk. Social support acts as a buffer against both loneliness33 and discrimination,34 and in this study, young adults who reported greater loneliness and discrimination also reported lower social support and neighborhood social cohesion. Discrimination was also directly related to perceived stress, which can have many negative effects on metabolic and cardiovascular health35,36 and may represent the mechanism behind the observed association with prediabetes/T2D. It is notable that, even in a primarily non-Hispanic White sample, a relationship between discrimination and stress was evident; as the discrimination scales used were broad, it is possible that the experienced discrimination may have been related to socioeconomic status, weight bias, or other unmeasured characteristics. It is notable that the social risk groups differed in demographic characteristics, with the highest-risk group consisting of the youngest individuals with the greatest proportion never married, the lowest incomes, and the highest proportion of Black individuals. The finding that this highest-risk social group was the youngest but had a higher prevalence of prediabetes/T2D is notable, as increasing age is more often associated with prediabetes/T2D risk, suggesting that social risk factors may operate independently of traditional biological risk factors. Overall, this study adds to the growing evidence that increased stress is associated with increased T2D risk37 but is the first to show the associations of stress in combination with other adverse psychosocial experiences among a large, diverse sample of young adults.

These findings have implications for future research and intervention development. For example, qualitative research and social network analysis may inform how social support systems could be leveraged to reduce risk of T2D in young adults. Studies of longitudinal changes in social support and loneliness over time are needed to better understand temporal relationships with T2D onset. In young adults, interventions targeting social support as a primary target for T2D risk reduction would represent a paradigm shift away from a current focus on weight reduction via directed changes in nutrition and physical activity. This may include actively involving caregivers, trusted adults, and peers directly in T2D prevention and treatment programs. Based on the demographic characteristics of the highest-risk latent class group in this analysis, such social support interventions may be particularly valuable for younger, unpartnered individuals. Additional research is needed to understand whether the predictive ability of these adverse psychosocial factors and whether they could be used to guide T2D screening efforts in adolescents and young adults. Pending such research, primary care clinicians may more strongly consider screening for T2D in young adults who report low social support, loneliness, or high stress, particularly in the setting of additional risk factors such as overweight/obesity and family history of T2D.

Limitations

This study has limitations, including its cross-sectional nature, which does not allow for conclusions about causality. It is likely that the relationship between social risk and prediabetes/T2D is bidirectional, as young adults with chronic disease may also experience impaired peer support and social relationships, predisposing to greater loneliness.38 Although it was not possible to assess for relationships between loneliness and undiagnosed prediabetes or T2D, the prevalence of undiagnosed T2D is low in young adult populations.39 All of Us seeks to recruit individuals historically underrepresented in biomedical research, but the majority in the cohort were Non-Hispanic White and college-educated individuals, so this sample may not reflect the populations most impacted by prediabetes and T2D. Notably, the prevalence of T2D (1.8%) and prediabetes (3.6%) in this cohort was lower than that reported by adults ages 18–44 years (T2D: 3.0%; diagnosed prediabetes: 13.8%).40 However, estimates that include adults up to age 44 span the age range when routine screening begins, partially explaining the lower prevalence in the study cohort. Nevertheless, the lower prevalence also likely reflects the generally healthier status of the All of Us volunteers. Notably, the prevalence of prediabetes/T2D among the study cohort, who had completed the SDOH survey, was similar though slightly higher than among all 18–34-year-olds without type 1 diabetes (5.5% versus 3.4%). Data collection spanned periods before, during, and after the COVID-19 pandemic. Although social experiences often worsened during the pandemic, these temporal trends would not be expected to alter the observed relationship between adverse social experiences and prediabetes/T2D. In this study, mental health conditions such as depression or anxiety were not evaluated, which are common among young adults with T2D.41 Finally, this study was unable to evaluate the relative independence from parental support, such as ongoing coverage by parental health insurance, which may influence access to health care and clinical outcomes.

Conclusions

Adverse psychosocial experiences and latent class-derived higher social risk phenotypes were associated with elevated odds of prediabetes and T2D. These findings may inform novel strategies to modify diabetes screening, moving toward more actionable, potentially modifiable characteristics. If successfully implemented, approaches that screen for and address adverse social risk factors such as loneliness, lack of social support, and discrimination may ultimately reduce the risks associated with young-onset T2D via prevention as well as early identification and treatment.

Supplementary Material

Online appendix

Funding:

CH, XQ, and MEV were supported by a Medical Research Grant from The Pittsburgh Foundation. MR was supported by an NIH K23HD104925. MZ was supported by NIDDK K23DK135794. MEV was also supported by NIH K23DK125719.

Footnotes

Declaration of Interests: No financial disclosures have been reported by the authors of this paper.

References

  • 1.Divers J, Mayer-Davis EJ, Lawrence JM, et al. Trends in Incidence of Type 1 and Type 2 Diabetes Among Youths — Selected Counties and Indian Reservations, United States, 2002–2015. MMWR Morb Mortal Wkly Rep 2020. 2020;69:161–165. doi: 10.15585/mmwr.mm6906a3external [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Lawrence JM, Divers J, Isom S, et al. Trends in Prevalence of Type 1 and Type 2 Diabetes in Children and Adolescents in the US, 2001–2017. JAMA. Aug 24 2021;326(8):717–727. doi: 10.1001/jama.2021.11165 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Tonnies T, Brinks R, Isom S, et al. Projections of Type 1 and Type 2 Diabetes Burden in the U.S. Population Aged <20 Years Through 2060: The SEARCH for Diabetes in Youth Study. Diabetes Care. Dec 16 2022;doi: 10.2337/dc22-0945 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Today Study Group, Bjornstad P, Drews KL, et al. Long-Term Complications in Youth-Onset Type 2 Diabetes. N Engl J Med. Jul 29 2021;385(5):416–426. doi: 10.1056/NEJMoa2100165 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Nanayakkara N, Curtis AJ, Heritier S, et al. Impact of age at type 2 diabetes mellitus diagnosis on mortality and vascular complications: systematic review and meta-analyses. Diabetologia. Feb 2021;64(2):275–287. doi: 10.1007/s00125-020-05319-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Rhodes ET, Prosser LA, Hoerger TJ, Lieu T, Ludwig DS, Laffel LM. Estimated morbidity and mortality in adolescents and young adults diagnosed with Type 2 diabetes mellitus. Diabet Med. Apr 2012;29(4):453–63. doi: 10.1111/j.1464-5491.2011.03542.x [DOI] [PubMed] [Google Scholar]
  • 7.Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med. Oct 9 2008;359(15):1577–89. doi: 10.1056/NEJMoa0806470 [DOI] [PubMed] [Google Scholar]
  • 8.U. S. Preventive Services Task Force, Davidson KW, Barry MJ, et al. Screening for Prediabetes and Type 2 Diabetes: US Preventive Services Task Force Recommendation Statement. JAMA. Aug 24 2021;326(8):736–743. doi: 10.1001/jama.2021.12531 [DOI] [PubMed] [Google Scholar]
  • 9.American Diabetes Association Professional Practice C. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes-2024. Diabetes Care. Jan 1 2024;47(Suppl 1):S20–S42. doi: 10.2337/dc24-S002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Hill-Briggs F, Fitzpatrick SL. Overview of Social Determinants of Health in the Development of Diabetes. Diabetes Care. Sep 1 2023;46(9):1590–1598. doi: 10.2337/dci23-0001 [DOI] [PubMed] [Google Scholar]
  • 11.U.S. Public Health Service. Our Epidemic of Loneliness and Isolation: The U.S. Surgeon General’s Advisory on the Healing Effects of Social Connection and Community. 2023:1–81. https://www.hhs.gov/sites/default/files/surgeon-general-social-connection-advisory.pdf [PubMed] [Google Scholar]
  • 12.Gebreab SY, Hickson DA, Sims M, et al. Neighborhood social and physical environments and type 2 diabetes mellitus in African Americans: The Jackson Heart Study. Health Place. Jan 2017;43:128–137. doi: 10.1016/j.healthplace.2016.12.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Gallup and Meta. The Global State of Social Connections. 2023. https://www.gallup.com/analytics/509675/state-of-social-connections.aspx [Google Scholar]
  • 14.Iovino P, Vellone E, Cedrone N, Riegel B. A Middle-Range Theory of Social Isolation in Chronic Illness. Int J Environ Res Public Health. Mar 10 2023;20(6)doi: 10.3390/ijerph20064940 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Helgeson VS, Horner FS, Reis HT, Niezink NMD, Libman I. Peer interactions and health among youth with diabetes: An ecological momentary assessment. Health Psychol. Apr 18 2024;doi: 10.1037/hea0001393 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.O’Neal LJ, Scarton L, Dhar B. Group Social Support Facilitates Adoption of Healthier Behaviors Among Black Women in a Community-Initiated National Diabetes Prevention Program. Health Promot Pract. Nov 2022;23(6):916–919. doi: 10.1177/15248399211045989 [DOI] [PubMed] [Google Scholar]
  • 17.Zupa MF, Lee A, Piette JD, et al. Impact of a Dyadic Intervention on Family Supporter Involvement in Helping Adults Manage Type 2 Diabetes. J Gen Intern Med. Mar 2022;37(4):761–768. doi: 10.1007/s11606-021-06946-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.All of Us Research Program I, Denny JC, Rutter JL, et al. The “All of Us” Research Program. N Engl J Med. Aug 15 2019;381(7):668–676. doi: 10.1056/NEJMsr1809937 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Anderson L, Hayward G, McElrath K, Scherer Z. New Data Visualization Shows More Young Adults Worked From Home and Lived Alone During the COVID-19 Pandemic. 2023. Accessed August 8, 2024. https://www.census.gov/library/stories/2023/09/young-adults-work-lifestyle-changed-during-pandemic.html
  • 20.Mujahid MS, Diez Roux AV, Morenoff JD, Raghunathan T. Assessing the measurement properties of neighborhood scales: from psychometrics to ecometrics. Am J Epidemiol. Apr 15 2007;165(8):858–67. doi: 10.1093/aje/kwm040 [DOI] [PubMed] [Google Scholar]
  • 21.Moser A, Stuck AE, Silliman RA, Ganz PA, Clough-Gorr KM. The eight-item modified Medical Outcomes Study Social Support Survey: psychometric evaluation showed excellent performance. J Clin Epidemiol. Oct 2012;65(10):1107–16. doi: 10.1016/j.jclinepi.2012.04.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Williams DR, Yan Y, Jackson JS, Anderson NB. Racial Differences in Physical and Mental Health: Socio-economic Status, Stress and Discrimination. J Health Psychol. Jul 1997;2(3):335–51. doi: 10.1177/135910539700200305 [DOI] [PubMed] [Google Scholar]
  • 23.Peek ME, Nunez-Smith M, Drum M, Lewis TT. Adapting the everyday discrimination scale to medical settings: reliability and validity testing in a sample of African American patients. Ethn Dis. Autumn 2011;21(4):502–9. [PMC free article] [PubMed] [Google Scholar]
  • 24.Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav. Dec 1983;24(4):385–96. [PubMed] [Google Scholar]
  • 25.Tesfaye S, Cronin RM, Lopez-Class M, et al. Measuring social determinants of health in the All of Us Research Program. Sci Rep. Apr 16 2024;14(1):8815. doi: 10.1038/s41598-024-57410-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.poLCA: Polytomous Variable Latent Class Analysis. Version R package version 1.4. 2013. https://dlinzer.github.com/poLCA
  • 27.Wright SL, Silard AG. Loneliness in Young Adult Workers. Int J Environ Res Public Health. Nov 4 2022;19(21)doi: 10.3390/ijerph192114462 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Xie J, Wang M, Long Z, et al. Global burden of type 2 diabetes in adolescents and young adults, 1990–2019: systematic analysis of the Global Burden of Disease Study 2019. BMJ. Dec 7 2022;379:e072385. doi: 10.1136/bmj-2022-072385 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Tang TS, Brown MB, Funnell MM, Anderson RM. Social support, quality of life, and self-care behaviors amongAfrican Americans with type 2 diabetes. Diabetes Educ. Mar-Apr 2008;34(2):266–76. doi: 10.1177/0145721708315680 [DOI] [PubMed] [Google Scholar]
  • 30.Henriksen RE, Nilsen RM, Strandberg RB. Loneliness increases the risk of type 2 diabetes: a 20 year follow-up - results from the HUNT study. Diabetologia. Jan 2023;66(1):82–92. doi: 10.1007/s00125-022-05791-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Carreon SA, Duran B, Tang TS, et al. Here for You: A Review of Social Support Research in Young Adults With Diabetes. Diabetes Spectr. Nov 2021;34(4):363–370. doi: 10.2337/dsi21-0013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Wiebe DJ, Helgeson V, Berg CA. The social context of managing diabetes across the life span. Am Psychol. Oct 2016;71(7):526–538. doi: 10.1037/a0040355 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Wang J, Mann F, Lloyd-Evans B, Ma R, Johnson S. Associations between loneliness and perceived social support and outcomes of mental health problems: a systematic review. BMC Psychiatry. May 29 2018;18(1):156. doi: 10.1186/s12888-018-1736-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Szaflarski M, Bauldry S. The Effects of Perceived Discrimination on Immigrant and Refugee Physical and Mental Health. Adv Med Sociol. 2019;19:173–204. doi: 10.1108/S1057-629020190000019009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Heard-Garris N, Yu T, Brody G, Chen E, Ehrlich KB, Miller GE. Racial Discrimination and Metabolic Syndrome in Young Black Adults. JAMA Netw Open. Apr 1 2024;7(4):e245288. doi: 10.1001/jamanetworkopen.2024.5288 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Chen E, Yu T, Brody GH, Lam PH, Goosby BJ, Miller GE. Discrimination and Inflammation in Adolescents of Color. Biol Psychiatry Glob Open Sci. Apr 2023;3(2):204–212. doi: 10.1016/j.bpsgos.2022.02.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Williams ED, Magliano DJ, Tapp RJ, Oldenburg BF, Shaw JE. Psychosocial stress predicts abnormal glucose metabolism: the Australian Diabetes, Obesity and Lifestyle (AusDiab) study. Ann Behav Med. Aug 2013;46(1):62–72. doi: 10.1007/s12160-013-9473-y [DOI] [PubMed] [Google Scholar]
  • 38.McGlone M, Long E. Are Young Adults with Long-Standing Illness or Disability at Increased Risk of Loneliness? European Journal of Public Health. 2021;31(Supplement_3)doi: 10.1093/eurpub/ckab165.571 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Fang M, Wang D, Coresh J, Selvin E. Undiagnosed Diabetes in U.S. Adults: Prevalence and Trends. Diabetes Care. Sep 1 2022;45(9):1994–2002. doi: 10.2337/dc22-0242 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Centers for Disease Control and Prevention. National Diabetes Statistics Report: Estimates of Diabetes and Its Burden in the United States. Last reviewed November 29, 2023. Accessed May 1, 2024. https://www.cdc.gov/diabetes/data/statistics-report/index.html
  • 41.Roberts AJ, Bao H, Qu P, et al. Mental health comorbidities in adolescents and young adults with type 2 diabetes. J Pediatr Nurs. Nov-Dec 2021;61:280–283. doi: 10.1016/j.pedn.2021.07.028 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Online appendix
NIHMS2080909-supplement-Online_appendix.pdf (240.3KB, pdf)

RESOURCES