Psychosocial correlates of alcohol and substance use in college youth with type 1 diabetes

Abstract

Objective

Adolescents and young adults with chronic diseases face unique challenges during the college years and may consume alcohol and other substances to cope with stressors. This study aimed to assess the patterns of substance use and to determine psychosocial correlates of these behaviors among college youth with type 1 diabetes (T1D).

Methods

College youth with T1D were recruited via social media and direct outreach into a web-based study. Participants answered validated questions about substance use, and they completed validated screeners of depressive and anxiety symptoms (PHQ-2 and GAD-2), illness acceptance (ICQ), interpersonal support (ISEL), and grit (Grit scale). Descriptive statistics, bivariate analyses, and multivariable regression evaluated substance use behaviors as a function of psychosocial factors while adjusting for age and sex.

Results

Alcohol (84.06%) and marijuana (41.30%) were the most common substances reported. In bivariate analyses, depressive symptoms were positively associated (p = .01) and illness acceptance was inversely associated (p = .02) with marijuana use. Higher grit scores were inversely associated with marijuana use (p < .001) and prescription drug misuse (p = .04). The significant associations between marijuana use and depressive symptoms (adjusted odds ratio [AOR] 1.31, 95% confidence interval [CI] 1.04–1.66), illness acceptance (AOR 0.96, 95% CI 0.91–0.99), and grit (AOR 0.32, 95% CI 0.17–0.60) persisted after adjustment for age and sex.

Conclusions

Substance use is prevalent among college youth with T1D. While psychosocial factors such as depressive symptoms may confer an increased risk, illness acceptance and grit may be protective—especially against marijuana use. Providers should address both positive and negative psychosocial factors to mitigate substance use in this population.

Alcohol and other substance use are relatively common among adolescents and young adults (AYA). According to surveys conducted by the Substance Abuse and Mental Health Services Administration, an estimated 34% of young adults between the ages of 18 and 25 endorsed binge drinking, 18% had smoked cigarettes, and 35% had used marijuana in the past month (Key Substance Use and Mental Health Indicators in the United States: Results from the 2019 National Survey on Drug Use and Health, 2020). Youth with chronic conditions engage in alcohol, tobacco, and other drug use at similar rates as youth without chronic conditions (Wisk & Weitzman, 2016). This trend extends to those with type 1 diabetes (T1D) (Sannegowda et al., 2023); in one study, more than half (56%) of AYA with T1D reported ever engaging in substance use, and 41% reported substance use in the past 30 days (Bento et al., 2020). These findings are concerning, not only because adolescence presents a critical period of vulnerability to addiction and substance use disorders later in life (Crews et al., 2007; Nelson et al., 2022) but also because substance use increases the risk of both acute and chronic diabetes-related complications in individuals with T1D (Pastor et al., 2020).

The transition to college is an especially vulnerable time for adolescents with T1D, as risk behaviors may increase and access to support and supervision may decrease during this period (Fedor et al., 2017). Additionally, many college youth reside in settings that are unable to manage their health needs and create social pressures to engage in harmful activities, such as alcohol and substance use (Lemly et al., 2014). It is therefore unsurprising that in one study, nearly half (44%) of adolescents with T1D reported stably high or increased alcohol use in the year after graduation from high school, with those who reported increased alcohol use more likely to live independently from their parents (Hanna et al., 2014).

Like alcohol and other substance use behaviors, anxiety and depression are common among AYA with T1D, with prevalence reaching 20%–30% in some studies (Bernstein et al., 2013; Buchberger et al., 2016); these conditions have also been shown to have negative impacts on diabetes management and glycemic outcomes (Bernstein et al., 2013; Buchberger et al., 2016). Prior studies have detected a positive association between alcohol consumption and anxiety and depressive symptoms in AYA with T1D (Knychala et al., 2015; Snyder et al., 2016). Additionally, among Canadian adults with T1D or T2D, those who reported positive mental health, characterized by happiness, psychological well-being, and social well-being, were less likely to smoke tobacco than their counterparts (Burns & Fardfini, 2021). However, less is known about how the use of other substances may be influenced by anxiety and depression in this population.

Moreover, few studies have investigated how positive psychosocial factors may impact substance use in youth with T1D. Positive youth development (PYD) is a framework for examining thriving in youth that uses a strengths-based approach as a key means of promoting positive outcomes (Catalano et al., 2004). Initiatives that incorporate the PYD framework have been shown to promote healthy outcomes in youth with chronic disease (Maslow & Chung, 2013) and have been applied to substance use prevention efforts in youth more generally (Melendez-Torres et al., 2016). Positive psychosocial factors, such as illness acceptance, interpersonal support, and grit, may underscore a commitment to health-protecting behaviors in youth with and without chronic illness (Guerrero et al., 2016; Weitzman et al., 2019) and may be especially relevant to those with T1D, who are particularly vulnerable to complications as a result of engaging in unhealthy behaviors. Accordingly, a nuanced understanding of risk and protective factors for use of various types of substances in college-attending youth with T1D may allow providers to tailor counseling and preventive efforts more specifically to this population.

In this study, we aimed to assess the proportion of college youth with T1D reporting alcohol and other substance use. We then sought to characterize the relationship between these entities by assessing associations between psychosocial screeners and substance use behaviors in this population. We hypothesized that those who had higher levels of depressive and anxiety symptoms would be more likely to use alcohol and other substances, whereas those who endorsed greater illness acceptance, interpersonal support, and grit would be less likely to engage in these risk behaviors.

Materials and methods

Participants

College youth with T1D were recruited into a pilot study that was designed to evaluate a digital health intervention targeting alcohol use in this population (Wisk et al., 2021). To promote recruitment, partnerships were established with two diabetes advocacy groups with large AYA followings. Recruitment messages were posted on the social media accounts, direct email newsletters, and website banners of these advocacy groups. These messages directed individuals to a Research Electronic Data Capture survey hosted on secure servers; the survey was accessible by phone, tablet, or computer. Respondents then completed a series of questions assessing participation eligibility after providing online informed consent. To meet inclusion criteria, participants had to be 17–25 years old, have a diagnosis of T1D, and currently attend or be enrolled in a college or university. This study was approved by the Institutional Review Board. Additional details of recruitment protocols and study implementation have been previously published (Wisk et al., 2019, 2021). Data are available upon reasonable request from the corresponding author.

Measures

Non-duplicative participants meeting inclusion criteria (N = 138) completed a baseline survey soliciting sociodemographic data (age, sex, race and ethnicity, health insurance source, parental education, year in school, and region of college or university attended). Participants were then asked a series of questions about their alcohol and other substance use patterns. Each participant was asked if they had ever had a drink containing alcohol, as well as if they had engaged in binge drinking over the past 2 weeks (defined as having 3, 4, or 5 drinks on one occasion based on age and sex thresholds) (Alcohol Screening and Brief Intervention for Youth: A Practitioner’s Guide, 2015). Additionally, all participants were asked to select the phrase that described their pattern of use for each of several additional substances: no use/have never tried, no use in the past 12 months but have ever used/tried, and have used in the past 12 months. The same coding scheme was used for each of the following substances: tobacco cigarettes or cigars, e-cigarettes/tobacco vaporizers, marijuana (including smoking, using a vaporizer, edibles, and any other method of consumption), or prescription medication (such as Adderall, Xanax, or oxycodone) that was not prescribed (subsequently referred to as prescription drug misuse). Information about both current and lifetime use was obtained due to the different levels of risk associated with each.

Participants were also asked questions targeting various measures of psychosocial health. Specifically, they were prompted to complete validated screeners of depressive symptoms (Patient Health Questionnaire-2 item [PHQ-2]) (Kroenke et al., 2003) and anxiety symptoms (Generalized Anxiety Disorder-2 item [GAD-2]) (Kroenke et al., 2007), which have been previously studied in this age group (Byrd-Bredbenner et al., 2021; Richardson et al., 2010). For both assessments, higher numbers indicated greater depressive and anxiety symptoms, respectively. In order to reduce the length of the overall survey, which was initially designed to evaluate a substance use intervention, these screening tools were chosen instead of the full PHQ and GAD assessments.

Additionally, participants completed validated questions addressing various aspects of the PYD framework: their acceptance of their T1D diagnosis, drawn from the Illness Cognition Questionnaire (ICQ) (Evers et al., 2001); their level of interpersonal support, drawn from the Interpersonal Support Evaluation List (ISEL) (Cohen & Hoberman, 1983); and their level of grit, drawn from the Grit scale (Duckworth et al., 2007). For all of these assessments, which have been studied in populations with similar ages and/or conditions (Brookings & Bolton, 1988; Rymon Lipińska & Nowicka-Sauer, 2023; Traino et al., 2019), higher numbers indicated greater illness acceptance, interpersonal support, and grit, respectively. To keep the survey length manageable, only some dimensions of the PYD framework were chosen for study, and a subset of questions with the highest factor loading was extracted and included in the questionnaire for participants (Boateng et al., 2018). Cronbach’s alphas in our sample for the 9 ICQ, 6 ISEL, and 4 Grit items were 0.86, 0.71, and 0.71, respectively.

Analysis

Quantitative analyses were performed using SAS 9.4 (Cary, NC), and statistical significance was considered at p < .05; all hypothesis tests were two-sided. All individuals who completed the baseline survey were included in the analyses (N = 138). Descriptive statistics, bivariate analyses (t-tests), and multivariable logistic regression were used to evaluate alcohol and substance use behaviors as a function of psychosocial factors. For the purposes of these analyses, the outcome variables were binary in all cases and reflected lifetime use of each substance (as well as past 2-week binge behavior in the case of alcohol use), while the predictor variables were continuous measures of the psychosocial screeners. Sensitivity analyses evaluated current substance use measures as binary outcome variables. Adjusted models controlled for age and sex, given that alcohol and substance use patterns differ by these two sociodemographic factors (Key Substance Use and Mental Health Indicators in the United States: Results from the 2019 National Survey on Drug Use and Health, 2020; Sex and Gender Differences in Substance Use, 2022). Age and psychosocial factors were treated as continuous variables, while sex and measures of substance use were treated as categorical variables.

Results

Sample characteristics

Participants had an average age of 20.49 years (SD 1.53); 80.43% were female, 82.61% were White non-Hispanic, and 55.80% were in their junior year of college or above. Additionally, 68.84% of participants had a parent with a bachelor’s degree or higher, and 86.96% were covered on a parent’s insurance plan.

The majority of participants (84.06%) endorsed lifetime alcohol use, and approximately one third (31.16%) endorsed binge drinking over the past 2 weeks. Fewer participants reported use of substances aside from alcohol. Marijuana was the second most common substance reported, with 41.30% of participants endorsing lifetime or past year use. Less than one quarter (21.74%) stated that they had ever tried cigarettes or cigars, 15.94% reported ever using e-cigarettes, and 10.87% endorsed prescription drug misuse.

Additionally, participants had a mean PHQ-2 score of 1.36 ± 1.54 (range 0–6) and a mean GAD-2 score of 2.22 ± 1.79 (range 0–6). In this sample, the mean ICQ score was 33.66 ± 8.31 (range 0–45), the mean ISEL score was 13.44 ± 3.26 (range 3–18), and the mean Grit score was 2.94 ± 0.62 (range 1.25–4) (Tables 1 and 2).

Table 1.

Sample characteristics (N = 138).

	M (SD)
Age (years)	20.49 (1.53)
	% (n)
Sex
Female	80.43 (111)
Male	19.57 (27)
Race and ethnicity
White, non-Hispanic	82.61 (114)
Hispanic	7.25 (10)
Asian, non-Hispanic	5.07 (7)
American Indian/Alaska Native, non-Hispanic	2.17 (3)
Black, non-Hispanic	1.45 (2)
Other, non-Hispanic	1.45 (2)
Parental education
Less than bachelor degree	31.16 (43)
Bachelor degree or higher	68.84 (95)
Health insurance
Insurance from parent	86.96 (120)
Other insurance plan or uninsured	13.04 (18)
Year in school
Freshman or sophomore	44.20 (61)
Junior or senior	45.65 (63)
Graduate school or other	10.15 (14)
Region of college/university
Northeast	27.54 (38)
Midwest	25.36 (35)
South	36.96 (51)
West	7.25 (10)
Outside the US	2.89 (4)

Table 2.

Substance use behaviors and psychosocial characteristics (N = 138).

	% (n)
Lifetime alcohol use
Yes	84.06 (116)
No	15.94 (22)
Past 2-week binge behaviora
Yes	31.16 (43)
No	68.84 (95)
Cigarette/cigar useb
Past year	8.70 (12)
Lifetime	13.04 (18)
None	78.26 (108)
E-cigarette use
Past year	2.90 (4)
Lifetime	13.04 (18)
None	84.06 (116)
Marijuana use
Past year	23.91 (33)
Lifetime	17.39 (24)
None	58.70 (81)
Prescription drug misusec
Past Year	5.07 (7)
Lifetime	5.80 (8)
None	89.13 (123)
	M (SD)
Illness Cognition Questionnaire (ICQ)d	33.66 (8.31)
Interpersonal Support Evaluation List (ISEL)e	13.44 (3.26)
Grit Scalef	2.94 (0.62)
Patient Health Questionnaire 2-item (PHQ-2)g	1.36 (1.54)
Generalized Anxiety Disorder 2-item (GAD-2)h	2.22 (1.79)

^aPast 2-week binge behavior is defined as having 3, 4, or 5 drinks on one occasion based on age and sex thresholds.

^bCategories (past year, lifetime, and none) are mutually exclusive.

^cPrescription drug misuse refers to taking prescription medication (such as Adderall, Xanax, or oxycodone) that was not prescribed.

^dThe Illness Cognition Questionnaire is a validated measure of illness acceptance. Higher scores indicate greater acceptance.

^eThe Interpersonal Support Evaluation List is a validated measure of interpersonal support. Higher scores indicate greater support.

^fThe Grit scale is a validated measure of grit. Higher scores indicate greater grit.

^gThe Patient Health Questionnaire-2 item is a validated screener of depression. Higher scores indicate greater depressive symptoms.

^hThe Generalized Anxiety Disorder-2 item is a validated screener of anxiety. Higher scores indicate greater anxiety symptoms.

Preliminary bivariate analyses

Differences emerged with regard to substance use behaviors according to several psychosocial variables. In bivariate analyses, higher depressive symptoms had a significant positive association with lifetime marijuana use (MPHQ-2 = 1.77 in users vs. 1.07 in non-users, p = .01), while higher illness acceptance had a significant inverse association with lifetime marijuana use (MICQ = 31.61 in users vs. 35.10 in non-users, p = .02). Additionally, grit had significant inverse associations with both lifetime marijuana use (MGrit = 2.69 in users vs. 3.11 in non-users, p < .001) and lifetime prescription drug misuse (MGrit = 2.60 in users vs. 2.98 in non-users, p = .04) (Table 3).

Table 3.

Bivariate analyses of substance use behaviors and psychosocial characteristics (N = 138).

	ICQa		ISELb		Gritc		PHQ-2d		GAD-2e
	M (SD)	p-value	M (SD)	p-value	M (SD)	p-value	M (SD)	p-value	M (SD)	p-value
Lifetime alcohol use		0.79		0.13		0.66		0.25		0.08
Yes	33.57 (8.13)		13.66 (3.07)		2.93 (0.60)		1.42 (1.57)		2.33 (1.81)
No	34.14 (9.36)		12.27 (4.00)		3.00 (0.73)		1.05 (1.36)		1.64 (1.59)
Lifetime cigarette/cigar use		0.19		0.34		0.07		0.07		0.21
Yes	31.73 (9.11)		12.90 (3.57)		2.74 (0.68)		1.87 (1.76)		2.60 (1.90)
No	34.19 (8.03)		13.59 (3.17)		2.99 (0.59)		1.22 (1.46)		2.11 (1.75)
Lifetime e-cigarette use		0.22		0.74		0.14		0.11		0.08
Yes	31.27 (9.95)		13.23 (3.22)		2.72 (0.78)		1.91 (1.74)		2.82 (1.68)
No	34.11 (7.93)		13.48 (3.28)		2.98 (0.58)		1.26 (1.49)		2.10 (1.80)
Lifetime marijuana use		0.02		0.33		<0.001		0.01		0.19
Yes	31.61 (8.59)		13.12 (3.17)		2.69 (0.61)		1.77 (1.75)		2.46 (1.76)
No	35.10 (7.84)		13.67 (3.33)		3.11 (0.57)		1.07 (1.31)		2.05 (1.80)
Lifetime prescription drug misusef		0.07		0.74		0.04		0.42		0.45
Yes	30.67 (6.20)		13.20 (2.96)		2.60 (0.62)		1.60 (1.12)		2.53 (1.64)
No	34.02 (8.48)		13.47 (3.31)		2.98 (0.61)		1.33 (1.59)		2.18 (1.81)
Past 2-week binge behaviorg		0.58		0.55		0.06		−0.08		0.95
Yes	33.14 (6.56)		13.67 (2.83)		2.78 (0.64)		1.72 (1.67)		2.23 (1.78)
No	33.89 (9.01)		13.34 (3.45)		3.01 (0.60)		1.20 (1.46)		2.21 (1.80)

Note. Bolded p-values indicate significance at p < .05.

^aICQ refers to the Illness Cognition Questionnaire, a validated measure of illness acceptance. Higher scores indicate greater acceptance.

^bISEL refers to the Interpersonal Support Evaluation List, a validated measure of interpersonal support. Higher scores indicate greater support.

^cGrit refers to the Grit scale, a validated measure of grit. Higher scores indicate greater grit.

^dPHQ-2 refers to the Patient Health Questionnaire-2 item, a validated screener of depression. Higher scores indicate greater depressive symptoms.

^eGAD-2 refers to the Generalized Anxiety Disorder-2 item, a validated screener of anxiety. Higher scores indicate greater anxiety symptoms.

^fPrescription drug misuse refers to taking prescription medication (such as Adderall, Xanax, or oxycodone) that was not prescribed.

^gPast 2-week binge behavior is defined as having 3, 4, or 5 drinks on one occasion based on age and sex thresholds.

Regression analyses

In regression models, each of the psychosocial factors was entered independently with adjustment for age and sex. After adjustment, only lifetime marijuana use had significant associations with psychosocial variables. Those who reported greater depressive symptoms were significantly more likely to endorse marijuana use (adjusted odds ratio [AOR] 1.31, 95% confidence interval [CI] 1.04–1.66), while those who endorsed greater illness acceptance (AOR 0.96, 95% CI 0.91–0.99) and grit (AOR 0.32, 95% CI 0.17–0.60) were significantly less likely to engage in marijuana use (Table 4).

Table 4.

Regression models for substance use behaviors (N = 138).^a,b

Adjusted Modelc	AOR	95% CI	p-value
Lifetime alcohol use
ICQd	1.00	(0.94–1.06)	0.87
ISELe	1.10	(0.96–1.27)	0.17
GRITf	0.78	(0.36–1.66)	0.51
GAD-2g	1.27	(0.93–1.73)	0.13
PHQ-2h	1.66	(1.16–2.37)	0.22
Past 2-week binge behaviori
ICQ	0.99	(0.95–1.04)	0.75
ISEL	1.04	(0.93–1.17)	0.50
GRIT	0.56	(0.31–1.02)	0.06
GAD-2	1.01	(0.82–1.23)	0.95
PHQ-2	1.23	(0.97–1.55)	0.08
Lifetime cigarette/cigar use
ICQ	0.97	(0.93–1.02)	0.29
ISEL	0.95	(0.83–1.08)	0.39
GRIT	0.54	(0.28–1.06)	0.08
GAD-2	1.16	(0.93–1.46)	0.19
PHQ-2	1.26	(0.98–1.61)	0.08
Lifetime e-cigarette use
ICQ	0.96	(0.91–1.02)	0.15
ISEL	1.00	(0.87–1.16)	0.99
GRIT	0.54	(0.26–1.12)	0.10
GAD-2	1.25	(0.98–1.60)	0.08
PHQ-2	1.26	(0.96–1.65)	0.10
Lifetime marijuana use
ICQ	0.96	(0.91–0.99)	0.04
ISEL	0.97	(0.87–1.08)	0.60
GRIT	0.32	(0.17–0.60)	<0.001
GAD-2	1.14	(0.94–1.38)	0.19
PHQ-2	1.31	(1.04–1.66)	0.02
Lifetime prescription drug misusej
ICQ	0.98	(0.92–1.05)	0.53
ISEL	1.02	(0.84–1.24)	0.84
GRIT	0.41	(0.16–1.08)	0.07
GAD-2	1.13	(0.82–1.56)	0.46
PHQ-2	1.01	(0.69–1.47)	0.95

Note. CI = confidence interval. Bolded p-values indicate significance at p < .05.

^aAll measures of lifetime substance use were modeled with logistic regression.

^bPsychosocial factors were treated as continuous variables, while measures of substance use were treated as categorical variables.

^cIn the adjusted model, each of the psychosocial predictors was entered independently with adjustment for age and sex.

^dICQ refers to the Illness Cognition Questionnaire, a validated measure of illness acceptance. Higher scores indicate greater acceptance.

^eISEL refers to the Interpersonal Support Evaluation List, a validated measure of interpersonal support. Higher scores indicate greater support.

^fGrit refers to the Grit scale, a validated measure of grit. Higher scores indicate greater grit.

^gGAD-2 refers to the Generalized Anxiety Disorder-2 item, a validated screener of anxiety. Higher scores indicate greater anxiety symptoms.

^hPHQ-2 refers to the Patient Health Questionnaire-2 item, a validated screener of depression. Higher scores indicate greater depressive symptoms.

ⁱPast 2-week binge behavior is defined as having 3, 4, or 5 drinks on one occasion based on age and sex thresholds.

^jPrescription drug misuse refers to taking prescription medication (such as Adderall, Xanax, or oxycodone) that was not prescribed.

In sensitivity analyses employing current substance use behaviors as outcome measures (Supplementary Table S1), grit emerged as a significant predictor of current marijuana use: lower grit scores were associated with higher odds of current marijuana use (AOR 0.28, 95% CI 0.15–0.52). Moreover, depressive symptoms were a significant predictor of current cigarette/cigar use: higher PHQ-2 scores were associated with higher odds of current cigarette/cigar use (AOR 1.30, 95% CI 1.01–1.67).

Discussion

This study reveals that substance use is prevalent among college youth with T1D. In this study sample, the most common substances reported over the course of the participants’ lifetime included alcohol (84.1%) and marijuana (41.3%), followed by tobacco products including cigarettes and cigars (21.7%). These findings are consistent with prior studies reporting substance use patterns in AYA with T1D (Bento et al., 2020; Sannegowda et al., 2023; Snyder et al., 2016), as well as college-aged youth in nationally representative samples (Key Substance Use and Mental Health Indicators in the United States: Results from the 2019 National Survey on Drug Use and Health, 2020; Wisk & Weitzman, 2016). Additionally, this study suggests that substance use patterns may have important associations with both measures of psychological distress and positive psychosocial factors in this population.

Consistent with our hypothesis, we found that psychosocial factors such as depressive symptoms confer an increased risk of engaging in certain substance use behaviors, although not all associations were equally robust. Though other studies have detected relationships between alcohol consumption and anxiety and depressive symptoms in this population (Knychala et al., 2015; Snyder et al., 2016), our study sample showed fewer significant associations between alcohol use and either depressive or anxiety symptoms. As this study was designed as a pilot trial of an alcohol use intervention and powered for that purpose, we may lack the power required to see consistent associations between each of the psychosocial variables and substance use outcomes. Additionally, since alcohol use is so ubiquitous by college, it may be harder to detect meaningful differences in this parameter; notably, prior work has demonstrated success addressing high-volume or binge alcohol use with a psychoeducational intervention designed to improve understanding of disease-specific risks and address the emotional context of living with a chronic disease (Wisk et al., 2021). The fact there was more variation in behaviors related to other substances suggests that there may be more individual predisposing characteristics that emerge as predictive for those substances.

Accordingly, this study demonstrated a significant association between lifetime marijuana use and depressive symptoms, with those reporting greater depressive symptoms more likely to endorse using marijuana. Additionally, there were associations between depressive symptoms and lifetime cigarette/cigar use and e-cigarette use, as well as between anxiety symptoms and lifetime e-cigarette use; while these were not statistically significant, they may still be clinically meaningful in this population. The fact that depressive symptoms were significantly associated with marijuana use corroborates the well-supported notion that psychological distress is an important risk factor for substance use in youth (Bandiera et al., 2016, 2017; Bottorff et al., 2009; Livingston et al., 2022; Nawi et al., 2021; Snyder et al., 2016; Spangler et al., 2001; Sumbe et al., 2022). It is thus important to consider the potential contributions of mental health conditions while addressing substance use behaviors in this population.

In contrast, the PYD framework emphasizes the importance of taking a strengths-based approach and identifying protective factors that can be harnessed to promote positive health behaviors (Lerner et al., 2011). Importantly, this is one of very few studies to evaluate the relationships between positive psychosocial factors and substance use in AYA with T1D. Prior studies have shown inverse associations between substance use and high emotional regulation, life satisfaction, subjective happiness, optimism, and emotional well-being in other groups of AYA (Brumback et al., 2021; Capaldi et al., 2021; Nawi et al., 2021; Schick et al., 2022). As with indicators of psychological distress, indicators of positive psychosocial functioning were most predictive of marijuana use in this study sample. After adjustment for age and sex, lifetime marijuana use showed significant inverse associations with both illness acceptance and grit. There were also inverse associations between grit and a number of other substance use behaviors, including lifetime cigarette/cigar use, e-cigarette use, prescription drug misuse, and past two-week binge behavior; although these associations did not reach significance, they may also have clinical relevance for college youth with T1D.

It is plausible that illness acceptance, and particularly grit, could have protective effects against substance use. Illness acceptance is likely related to an understanding of the unique risks posed by certain substances in individuals with T1D (Kossowsky et al., 2021; Levy et al., 2022; Wisk et al., 2020); an acceptance of these disease-specific risks may promote decreased engagement in behaviors that could lead to illness complications and flares (Weitzman et al., 2019). Additionally, grit has been shown to be a protective factor against alcohol and marijuana use in one study of Latino adolescents (Guerrero et al., 2016); in that study, grit was associated with higher self-efficacy scores, which may be one mechanism underlying responsible decision-making in this domain. Grit has also been associated with self-control, behavioral inhibition, and improved healthcare management skills (Duckworth & Gross, 2014; Liu et al., 2022; Traino et al., 2019)—all of which may contribute to an avoidance of risk behaviors. An understanding of the positive psychosocial factors that may be protective can thus inform the development of interventions to reduce substance use in this population.

There are several limitations of this study. First, this was a selective sample of young adults with T1D who met the inclusion criteria and consented to participate in an intervention designed to reduce alcohol use; as such, they are likely not representative of all college youth with T1D. Indeed, the rates of alcohol and substance use have been shown to vary in different samples of young adults with T1D (Sannegowda et al., 2023), and as the rates in this sample were at the higher end of the range, it is possible that the study selected for individuals who had a history of alcohol or substance use. Moreover, as this sample was recruited from diabetes advocacy groups, sampling bias may have resulted in a particularly engaged group of individuals with different behaviors and psychosocial characteristics than the general population.

Furthermore, our sample of participants was relatively homogenous in terms of race and ethnicity, as the majority of participants were White non-Hispanic. Thus, the results may not be generalizable to other populations with higher proportions of minoritized individuals, particularly as substance use patterns differ by race and ethnicity in youth (Farokhnia et al., 2024; Racial/Ethnic Differences in Substance Use, Substance Use Disorders, and Substance Use Treatment Utilization among People Aged 12 or Older [2015–2019], 2021). Additionally, the sample was predominantly female with a limited number of male participants; this has important implications related to the generalizability of study findings, as prior studies have found different rates and predictors of alcohol and substance use between male and female adolescents (Farokhnia et al., 2024; Rahal et al., 2024; Sex and Gender Differences in Substance Use, 2022). Accordingly, future studies are needed to evaluate the relationship between psychosocial factors and substance use behaviors in more diverse and heterogenous samples.

Additionally, the relatively small sample size may not have had sufficient power to detect certain associations between substance use behaviors and psychosocial factors, as the original study was powered to assess an alcohol use intervention, rather than these secondary analyses. Alpha inflation is another potential limitation, given the number of dependent variables in relation to independent variables, and additional studies with larger samples that incorporate corrections for multiple comparisons are needed. The degrees of alcohol and substance use were also not examined, and as such, analyses employing binary substance use variables may not sufficiently reflect associations between psychosocial factors and high-volume use, aside from binge drinking. The use of psychosocial screeners of depressive and anxiety symptoms, rather than full measures, may have detracted from the ability to detect associations with these constructs. Likewise, not all aspects of the PYD framework were examined, and the selected items from the ICQ, ISEL, and Grit scale were not validated in this specific population. Furthermore, it was not possible to ascertain directionality in this cross-sectional study, and substance use behaviors and psychosocial factors may indeed have a bidirectional relationship. Future work is needed to elicit any causal relationships between these two constructs and to explore potential interactions between psychosocial variables. Finally, all outcomes related to substance use behaviors and psychosocial variables were self-reported and not verified by other means and thus may have been subject to reporting bias among participants.

In spite of these limitations, there are several strengths to this study. First, this study was focused on a population in whom substance use is relatively common but in whom there have been few prior investigations of contributing psychosocial factors. Additionally, this study offers valuable insights into a variety of different substances not limited to alcohol or tobacco, which is important given the frequency of marijuana use (and e-cigarette vapor product and prescription drug misuse to a lesser extent) in this population. Finally, this study investigated the impacts of negative psychosocial factors in addition to positive psychosocial factors, which are less often studied but nonetheless important to incorporate into guidance around substance use. Future studies may benefit from examining additional aspects of the PYD framework in relation to substance use behaviors in this population.

In conclusion, this study is one of the first to evaluate both positive and negative psychosocial correlates of substance use in college youth with T1D. Alcohol and marijuana emerged as the most common substances reported, and marijuana, in particular, showed significant associations with both positive and negative psychosocial factors. Among the psychosocial factors evaluated, depressive symptoms and grit were most predictive of substance use behaviors: those with more depressive symptoms and lower grit scores were more likely to report engaging in substance use. Given the health risks of any substance use in college youth with T1D, it will be important for providers to comprehensively screen for and address these factors to curb substance use and the specific complications that may ensue. Additionally, future efforts should be directed toward developing relevant psychosocial interventions and evaluating their impact on substance use behaviors and health outcomes in this medically vulnerable population.

Supplementary material

Supplementary material is available online at Journal of Pediatric Psychology (https://academic.oup.com/jpepsy/).

Author contributions

Rebecca K. Tsevat (Conceptualization [equal], Formal analysis [lead], Methodology [lead], Writing—original draft [lead], Writing—review & editing [lead]), Elissa R. Weitzman (Conceptualization [equal], Investigation [equal], Methodology [equal], Project administration [equal], Supervision [equal], Writing—review & editing [equal]), and Lauren E. Wisk (Conceptualization [equal], Data curation [lead], Formal analysis [equal], Funding acquisition [lead], Investigation [lead], Methodology [equal], Project administration [lead], Supervision [lead], Validation [lead], Writing—original draft [supporting], Writing—review & editing [equal])

Funding

This work was supported by the Boston Children’s Hospital Awards Committee Pilot Research Project Funding (FP01017994); the UCLA Children’s Discovery and Innovation Institute (CDI) Research Recognition Award; the National Clinician Scholars Program at UCLA; the UCLA-UCSF ACEs Aware Family Resilience Network (UCAAN) (21-10317); and a career development award from the National Institute of Diabetes and Digestive and Kidney Diseases at the National Institutes of Health (NIH/NIDDK K01DK116932 to L.E.W.). The funding sources had no involvement in the study design; the collection, analysis, and interpretation of data; the writing of the report; or the decision to submit the article for publication.

Conflicts of interest: None declared.