Sexual Orientation and Gender Identity Disparities in Co-Occurring Depressive Symptoms and Probable Substance Use Disorders in a National Cohort of Young Adults

Abstract

This study examined sexual orientation and gender identity differences in co-occurring depressive symptoms and substance use disorders (SUDs) among young adults in the Growing Up Today Study national cohort (n = 12,347; ages 20–35; 93% non-Hispanic white). Self-administered questionnaires assessed recent co-occurring depressive symptoms and probable nicotine dependence, alcohol use disorder, and drug use disorder. Multinomial logistic regressions with generalized estimating equations quantified differences in prevalences of depressive symptoms only, SUDs only, and co-occurrence, among sexual minorities (mostly heterosexual; lesbian, gay, and bisexual [LGB]) compared to completely heterosexual participants, and gender minorities compared to cisgender participants. Analyses stratified by sex assigned at birth revealed sexual minorities evidenced greater odds of co-occurrence than their completely heterosexual counterparts (assigned female AORs: 3.11–9.80, ps < 0.0001; assigned male AORs: 2.90–4.87, ps < 0.001). Sexual orientation differences in co-occurrence were pronounced among LGB participants assigned female at birth who evidenced nearly 10 times the odds of co-occurring depressive symptoms with nicotine dependence and drug use disorders than did heterosexual participants assigned female at birth. Relationships between gender identity and co-occurrence were generally weaker, possibly due to low power. Gender minorities assigned male at birth, however, evidenced greater odds of co-occurring depressive symptoms and alcohol use disorders (AOR 2.75, p = 0.013) than their cisgender counterparts. This study adds to the limited research quantifying sexual orientation or gender identity differences in recent co-occurring depressive symptoms and SUDs among young adults and suggests sexual and gender minority young adults should be prioritized in prevention and treatment of co-occurring depression and SUDs.

1. Introduction

Depression and substance use disorders (SUDs) are leading contributors to disability.^1–4 The disorders commonly co-occur,^5–7 contributing to higher rates of morbidity and mortality than either disorder alone.^8–11 Compared to other age groups, young adults evidence relatively high prevalences of depression and SUDs,^5,12,13 with studies finding that young adults 18 to 25-years-old have among the highest risks for co-occurrence.^5,10

Among young adults, sexual and gender minorities (SGMs) – i.e., those who identify as lesbian, gay, bisexual, transgender, or other minority sexual orientation or gender identity – may be at particularly elevated risk for co-occurring diagnostic depression and SUDs. These health disparities have been explained by minority stress theory which posits that chronic exposure to multilevel SGM-related stressors and discrimination (e.g., internalized homophobia, anti-SGM structural stigma) shape negative mental health outcomes and associated coping behaviors, such as substance use.^14–16 Indeed, epidemiologic research consistently finds SGM young adults in the United States evidence disproportionately greater prevalences of depressive symptoms and probable SUDs than their completely heterosexual or cisgender peers.^17–23

Despite separate streams of literature examining differential risk in depression and SUDs among SGM young adults, few studies have examined sexual orientation differences in co-occurring depressive symptoms and SUDs. Analyses of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) indicated sexual minority adult women²⁴ and men²⁵ with lifetime alcohol use disorders had approximately 1.5 and 2 times the odds of having co-occurring lifetime mood (including depression) and drug use disorders compared to their heterosexual peers, respectively. Another study found similarly elevated risk of co-occurring recent psychological distress (i.e., within the past few weeks) and substance use among sexual minorities ages 16 and older participating in the Swedish National Public Health Survey.²⁶ Together, these studies suggest elevated rates of co-occurring depression and SUDs among sexual minority adults, but do not offer specific estimates for young adults. Furthermore, they do not examine effect modification by sex assigned at birth in statistical modeling nor do they examine differences by gender identity in co-occurrence.

To build on the sparse literature, we examined sexual orientation and gender identity differences in co-occurring recent depressive symptoms and probable SUDs (i.e., nicotine dependence, and alcohol and drug use disorders) during young adulthood. We analyzed data from two national United States longitudinal cohorts from The Growing Up Today Study (GUTS; GUTS I and II) was collected when participants were between ages 20–35 years. We used the Center for Epidemiologic Studies Depression Scale (CES-D)^27,28 to assess past-week depressive symptoms and Diagnostic and Statistical Manual of Mental Disorders, fourth edition²⁹ (DSM-IV) criteria to assess past 12-month probable nicotine dependence, and alcohol and drug use disorders. We hypothesized that, compared to their completely heterosexual peers, sexual minorities would evidence more depressive symptoms, SUDs, and co-occurring depressive symptoms and SUDs. Similarly, we hypothesized that, compared to their cisgender peers, gender minorities would evidence more depressive symptoms, SUDs, and co-occurring depressive symptoms and SUDs.

Research has found differences in depressive symptoms and SUDs by sex assigned at birth, with larger differences among sexual minority young adults assigned female at birth than sexual minority young adults assigned male at birth^17,19,20 (hereafter “assigned female” and “assigned male”). For example, our prior research with GUTS found that sex assigned at birth significantly modified the relationship between sexual orientation and multiple co-occurring SUDs such that differences in risk between sexual minorities and completely heterosexuals were larger for participants assigned female compared to those assigned male.²⁰ Similarly, an analysis of the National Longitudinal Study of Adolescent to Adult Health (Add Health) found consistent sexual orientation disparities in perceived stress and depressive symptoms among sexual minority young adult women, whereas disparities were less consistent among sexual minority young adult men.¹⁷ The Add Health analysis, however, did not explicitly examine gender/sex at birth as an effect modifier of the relationship between sexual orientation and perceived stress or depressive symptoms. Based on this prior evidence, we hypothesized that sex assigned at birth would modify the relationship between sexual orientation and depressive symptoms, SUDs, and co-occurrence such that sexual orientation disparities would be larger in participants assigned female compared to those assigned male. To test this hypothesis, we estimated sex assigned at birth-by-sexual orientation statistical interactions and present sex assigned at birth-stratified results. Due to small numbers in the sample, were unable to examine sex assigned at birth effect modification among gender minorities.

2. Methods

2.1. Participants and procedures

GUTS I and II participants were recruited in 1996 and 2004, respectively, and are children of the Nurses’ Health Study II participants, whose information was provided to GUTS researchers by their mothers. Specifically, GUTS participants were recruited through letters sent to Nurses’ Health Study II participants who were mothers requesting permission to contact their child/children. Willing mothers then provided information regarding each child’s name, age, sex, and address, and each child was sent a packet inviting them to participate in GUTS and a sex-specific questionnaire.³⁰

At baseline, GUTS I participants were 9–14 years old (N=16,875) and GUTS II participants were 10–17 years old (N=10,918). GUTS questionnaires are administered annually or biennially (via pencil and paper or online) and gather information on health-related behaviors and experiences, including substance use, SUDs, and mental health. More information about GUTS and enrollment procedures is available elsewhere.^31,32 Partners Healthcare provided institutional review board approval for data collection procedures.

This study analyzed data provided by 12,347 participants who contributed 17,364 observations (7,330 participants contributed one observation; 5,017 participants contributed two observations). To be included in the analytic sample, participants must have responded to at least one of the following two data collection scenarios: (1) the 2010 general questionnaire (GUTS I only; response rate=51%, n=8,690) or (2) the 2016 general questionnaire (GUTS I, II; response rate=47%, n=13,120) and the 2015–2017 Substance Substudy questionnaire. The Substance Substudy questionnaire was administered over a two-year period to GUTS recent responders (GUTS I, II; response rate=73% of 13,340 invited to participate). More information about the Substudy is available elsewhere.²⁰

Participants assigned male had greater attrition than those are assigned female (p<.0001). Participants from the Northeastern, Southern, and Midwestern U.S. were less likely to be included in the analysis than those from the West (ps<.0001). There were no differences by race/ethnicity among those included in the analyses compared to those not included (p=.43). Younger participants were more likely to be included in the analyses than those who were older at baseline (p<.0001).

2.2. Measures

2.2.1. Sexual orientation

The GUTS 2010 and 2016 general questionnaire assessed identity and sex(es) of romantic attraction using a single question from the Minnesota Adolescent Health Survey³³ that read: “Which one of the following best describes your feelings?”. Response options were: (1) completely heterosexual (attracted to persons of the opposite sex), (2) mostly heterosexual, (3) bisexual (equally attracted to men and women), (4) mostly homosexual, (5) completely homosexual (gay/lesbian, attracted to persons of the same sex), or (6) not sure. We created the following analytic categories: completely heterosexual, mostly heterosexual, and bisexual/lesbian/gay (LGB; we combined bisexual, mostly homosexual, and completely homosexual to maximize precision of estimates). Because GUTS has regularly assessed sexual orientation since 1999, when sexual orientation was missing on the 2010 or 2016 questionnaire, we substituted information on sexual orientation from participants’ most recent previous response (n=305 observations; 1.8%). We modeled sexual orientation as time varying and corresponding with the timing of participants’ assessment of depressive symptoms and SUDs to account for changes between the two waves.

2.2.2. Gender identity

GUTS assessed participants’ sex assigned at birth at baseline and gender identity on the 2010 and 2016 general questionnaire with the question: “How do you describe yourself?”, response options: (1) Male, (2) Female, (3) Transgender, or (4) “None of the above” (2010) / “Do not identify as male, female or transgender” (2016). We classified participants as gender minorities if they selected (3) or (4), or if their most recent gender identity response was discordant with their baseline sex assigned at birth. We classified participants as cisgender when their gender identity response corresponded with their baseline sex assigned at birth.

2.2.3. Depressive symptoms

The 2010 and 2016 general questionnaire assessed past-week depressive symptoms using the 10-item CES-D,³⁴ which has been validated across age groups, and with community and clinic-based samples.^27,28,34,35 Consistent with past research^18,36 and the original recommendation,³⁴ we created a binary variable for analysis by summing scale response (range 0–30), with scores ≥10 meeting criteria for probable past-week depressive symptoms.

2.2.4. Substance use disorders

The 2010 general questionnaire and the 2015–2017 Substance Substudy questionnaire assessed past 12-month nicotine dependence, alcohol abuse/dependence, and drug abuse/dependence using items adapted from the National Survey on Drug Use and Health corresponding to DSM-IV SUD criteria. We classified participants as meeting criteria for having probable past 12-month substance dependence if they endorsed three or more of seven symptoms of dependence. We classified participants as meeting criteria for probable past 12-month substance abuse (alcohol and drug use only) if they endorsed at least one of four symptoms of abuse.²⁹ We created three probable SUD variables: nicotine dependence (yes/no), alcohol use disorder (combining abuse and dependence; yes/no), and drug use disorder (combining abuse and dependence; yes/no) consistent with DSM-IV criteria.²⁹

2.2.5. Co-occurring depressive symptoms and substance use disorders

We created three new variables for co-occurring depressive symptoms and SUDs with four possible categories for each variable:

1. Depressive symptoms and nicotine dependence: (a) no past-week depressive symptoms, no past 12-month nicotine dependence, (b) past-week depressive symptoms only, (c) past 12-month nicotine dependence only, (d) both past-week depressive symptoms and past 12-month nicotine dependence.

2. Depressive symptoms and alcohol use disorder: (a) no past-week depressive symptoms, no past 12-month alcohol use disorder, (b) past-week depressive symptoms only, (c) past-12-month alcohol use disorder only, (d) past-week depressive symptoms and past 12-month alcohol use disorder.

3. Depressive symptoms and drug use disorder: (a) no past-week depressive symptoms, no past 12-month drug use disorder, (b) past-week depressive symptoms only, (c) past 12-month drug use disorder only, (d) past-week depressive symptoms and past 12-month drug use disorder.

The 2010 general questionnaire assessed both depressive symptoms and SUDs. The 2016 general questionnaire assessed depressive symptoms, but not SUDs. Thus, we linked the 2015–2017 Substudy SUDs assessment with the 2016 depressive symptoms assessment. To maximize use of available data, we substituted depressive symptoms assessed on the 2014 general questionnaire in less than 2% (n=153) of the observations that were missing 2016 depressive symptoms information.

2.2.6. Covariates

We adjusted for race/ethnicity (non-Hispanic white versus other), region of residence (North east, West, South, vs. Midwest), cohort (GUTS I vs. II), and age at the time of SUD assessment in the statistical models.

2.3. Statistical Analysis

We first conducted sex assigned at birth-stratified descriptive analyses to estimate unadjusted prevalences of co-occurring depressive symptoms and SUDs for each sexual orientation and gender identity subgroup. To estimate sex assigned at birth-stratified bivariate statistical associations of sexual orientation and gender identity with co-occurring depressive symptoms and SUDs, we conducted unadjusted multinomial logistic regression with generalized estimating equations (GEE). We used GEE to account for the non-independent repeated measures from the same individual and sibling clusters.^37,38

We then used GEE multinomial logistic regression to test for effect modification of sex assigned at birth in associations between sexual orientation and co-occurring depressive symptoms and SUDs by including sex assigned at birth-by-sexual orientation interaction terms and adjusting for covariates (i.e., age, race/ethnicity, cohort, and region of residence). Because we found significant sex assigned at birth-by-sexual orientation statistical interactions (p-values <.05), we conducted sex assigned at birth-stratified multivariable analyses to estimate associations of sexual orientation and gender identity with co-occurring depressive symptoms and SUDs. We estimated adjusted odds ratios (AOR) and 95% confidence intervals (CI). Models included sexual orientation, gender identity, and covariates; “completely heterosexual” and “cisgender” were the referent groups. All analyses were performed with SAS software, version 9.4.³⁹

We chose not to impute missing data on the outcomes. When data on sexual orientation was missing for an observation and could not be substituted from a previous survey response, we removed the observation from analysis (n=6). Missing data for the other covariates was minimal and coded into the most frequent category to facilitate model convergence (gender identity n=4 observations [coded as cisgender], race/ethnicity n=116 observations [coded as non-Hispanic white], and region of residence n=35 observations [coded as Northeast]).

3. Results

3.1. Participant sociodemographics

In Table 1, we present sex assigned at birth-stratified sociodemographic distributions of participants’ observations. The majority of observations came from participants identifying as completely heterosexual (78.2%), followed by mostly heterosexual (16.4%), and lesbian, gay, bisexual (LGB; 5.5%). Among participants assigned female and assigned male, 0.7% and 1.1% of observations came from gender minorities, respectively. Most observations were from non-Hispanic white participants (~93%) and the average age at response was 27 years old (standard deviation: 3.2 years; range: 20–35).

Table 1.

Characteristics of repeated measures observations of Growing Up Today Study participants aged 20–35 years, by sex assigned at birth (2010, 2016, N=17,364)

	Assigned Female at Birth (n=11,755)		Assigned Male at Birth (n=5,609)
Characteristic	n	%	n	%
Sexual Orientation
Completely heterosexual	8,844	75.2	4,731	84.4
Mostly heterosexual	2,284	19.4	557	9.9
Lesbian/gay/bisexual	627	5.3	321	5.7
Gender Identity
Cisgender/non-gender minority	11,671	99.3	5549	98.9
Gender minority	84	0.7	60	1.1
Race/Ethnicity
Non-Hispanic white	10,914	92.9	5,223	93.1
Other	841	7.2	386	6.9
Cohort
GUTS1	9,224	78.5	4,453	79.4
GUTS2	2,531	21.5	1,156	20.6
Region of Residence
West	2,065	17.6	1,107	19.7
Midwest	3,766	32.0	1,797	32.0
South	2,120	18.0	939	16.7
Northeast	3,804	32.4	1,766	31.5
Age, mean (standard deviation)		27.0 (3.2)		26.9 (3.2)

Note: Percentages within variables sum to 100% except for rounding error.

3.2. Prevalence of outcomes of co-occurring depressive symptoms and substance use disorders and bivariate associations with sexual orientation and gender identity

Table 2 presents sex assigned at birth-stratified prevalences of co-occurring depressive symptoms and SUDs by sexual orientation and gender identity. Mostly heterosexual and LGB participants evidenced higher prevalences of depressive symptoms only, SUDs only, and co-occurring SUDs and depressive symptoms compared to completely heterosexual participants. Gender minority participants also evidenced higher prevalences in most of the main outcomes. Among participants assigned female, all co-occurring depressive symptoms and SUD outcomes were significantly higher among gender minorities, whereas among participants assigned male, only co-occurring depressive symptoms and alcohol use disorders was more prevalent among gender minorities.

Table 2.

Prevalences of outcomes of co-occurring depressive symptoms and substance use disorders and bivariate associations with sexual orientation and gender identity among Growing Up Today Study participants aged 20–35 years, by sex assigned at birth (2010, 2016)

	SEXUAL ORIENTATION								GENDER IDENTITY
	Completely Heterosexual		Mostly Heterosexual			Lesbian/Gay/Bisexual			Cisgender		Gender Minority
	n	%	n	%	p-value	n	%	p-value	n	%	n	%	p-value.
Assigned Female at Birth
Depressive Symptoms (DS) & Nicotine Dependence (ND)
No DS, No ND	6536	74.0	1353	59.2	REF	293	46.7	REF	8141	69.8	41	48.8	REF
DS Only	1774	21.1	621	27.2	<.0001	215	34.3	<.0001	2578	22.1	32	38.1	0.001
ND Only	336	3.8	160	7.0	<.0001	48	7.7	<.0001	540	4.6	4	4.8	0.461
DS & ND	186	2.1	150	6.6	<.0001	71	11.3	<.0001	400	3.4	7	8.3	0.013
Depressive Symptoms (DS) & Alcohol Use Disorder (AUD)
No DS, No AUD	6426	72.8	1305	57.2	REF	291	46.6	REF	7985	68.6	37	44.1	REF
DS Only	1662	18.8	599	26.3	<.0001	235	37.7	<.0001	2464	21.2	32	38.1	0.0002
AUD Only	445	5.0	204	9.0	<.0001	50	8.0	<.0001	691	5.9	8	9.5	0.018
DS & AUD	292	3.3	172	7.5	<.0001	48	7.7	<.0001	505	4.3	7	8.3	0.033
Depressive Symptoms (DS) & Drug Use Disorder (DUD)
No DS, No DUD	6761	76.7	1422	62.5	REF	303	48.6	REF	8446	72.6	40	47.6	REF
DS Only	1853	21.0	675	29.6	<.0001	242	38.8	<.0001	2739	23.6	31	36.9	0.002
DUD Only	104	1.2	86	3.8	<.0001	36	5.8	<.0001	221	1.9	5	6.0	0.006
DS & DUD	98	1.1	94	4.1	<.0001	42	6.7	<.0001	226	1.9	8	9.5	<.0001
Assigned Male at Birth
Depressive Symptoms (DS) & Nicotine Dependence (ND)
No DS, No ND	3478	73.7	278	50.0	REF	177	55.1	REF	3903	70.5	30	50.0	REF
DS Only	794	16.8	199	35.8	<.0001	97	30.2	<.0001	1069	19.3	21	35.0	0.004
ND Only	285	6.0	38	6.8	0.007	22	6.9	0.114	340	6.1	5	8.3	0.186
DS & ND	162	3.4	41	7.4	<.0001	25	7.8	<.0001	224	4.1	4	6.7	0.128
Depressive Symptoms (DS) & Alcohol Use disorder (AUD)
No DS, No AUD	3287	69.7	257	46.5	REF	157	49.2	REF	3673	66.5	28	46.7	REF
DS Only	696	14.8	179	32.4	<.0001	82	25.7	<.0001	942	17.1	15	25.0	0.050
AUD Only	472	10.0	58	10.5	0.004	42	13.2	0.002	565	10.2	7	11.7	0.266
DS & AUD	258	5.5	59	10.7	<.0001	38	11.9	<.0001	345	6.2	10	16.7	0.001
Depressive Symptoms (DS) & Drug Use Disorder (DUD)
No DS, No DUD	3533	75.3	278	50.4	REF	179	56.3	REF	3959	71.9	31	51.7	REF
DS Only	822	17.5	191	34.6	<.0001	101	31.8	<.0001	1093	19.9	21	35.0	0.005
DUD Only	212	4.5	36	6.5	0.0001	20	6.3	0.013	264	4.8	4	6.7	0.230
DS & DUD	126	2.7	47	8.5	<.0001	18	5.7	0.0002	187	3.4	4	6.7	0.070

Note: N equals the number of observations over repeated measures. Prevalence shown are unadjusted. Bivariate p-values estimated by generalized estimating equations. “Completely heterosexual” and “cisgender” participants are the referent groups for the sexual orientation and gender identity independent variables. Associations in bold are p-value < .05.

3.3. Estimates of effect modification of sex assigned at birth on associations of sexual orientation and outcomes of co-occurring depressive symptoms and substance use disorders

Results of multivariable GEE multinomial regression indicated that, in some instances, sex assigned at birth significantly modified relationships between sexual orientation and prevalence of co-occurring depressive symptoms and SUDs (Table 3). Supporting our hypothesis, in several instances, differences between sexual minorities and heterosexuals were larger among participants assigned female compared to male. Specifically, differences between LGB and completely heterosexual participants in prevalence of co-occurring depressive symptoms and nicotine dependence (p-value=0.001) and co-occurring depressive symptoms and drug use disorders (p-value=0.001) were larger among participants assigned female. Contrary to our hypothesis, differences between mostly heterosexual and completely heterosexual participants in prevalence of depressive symptoms without co-occurring SUDs were smaller among participants assigned female compared to male (p-values<.0001).

Table 3.

Multivariable multinomial logistic regression estimates of effect modification of birth sex in associations between sexual orientation and outcomes of co-occurring depressive symptoms and substance use disorders among Growing Up Today Study participants aged 20–35 years (2010, 2016)

	Mostly Heterosexual		Lesbian/Gay/Bisexual
	Estimate	p-value	Estimate	p-value
Outcome 1: Depressive Symptoms (DS) & Nicotine Dependence (ND)
No DS, No ND	REF		REF
DS Only	−0.61	<.0001	0.10	0.580
ND Only	0.31	0.153	0.78	0.014
DS & ND	0.20	0.395	1.06	0.001
Outcome 2: Depressive Symptoms (DS) & Alcohol Use Disorder (AUD)
No DS, No AUD	REF		REF
DS Only	−0.61	<.0001	0.22	0.238
AUD Only	0.37	0.039	0.27	0.298
DS & AUD	0.02	0.932	0.13	0.637
Outcome 3: Depressive Symptoms (DS) & Drug Use Disorder (DUD)
No DS, No DUD	REF		REF
DS Only	−0.53	<.0001	0.17	0.325
DUD Only	0.62	0.013	1.40	<.0001
DS & DUD	−0.1	0.955	1.17	0.001

Note: “Completely heterosexual” and “assigned male at birth” are the referent groups for the sexual orientation and birth sex independent variables in the interaction. Generalized estimating equations models adjusted for age, race/ethnicity, cohort, and region of residence. Associations in bold are p-value <.05.

3.4. Sex assigned at birth-stratified multivariable associations of sexual orientation and gender identity with co-occurring depressive symptoms and substance use disorders

Tables 4 and 5 present sex assigned at birth-stratified multivariable associations of sexual orientation and gender identity with depressive symptoms, SUDs, and co-occurring depressive symptoms and SUDs. Among participants assigned female (Table 4) LGB participants and mostly heterosexual participants evidenced elevated odds of all co-occurrence outcomes. Notably, sexual minority females had between three to more than nine times higher odds of experiencing co-occurring depressive symptoms and SUDs, depending on the sexual orientation group and the type of SUD examined. After adjusting for sexual orientation, gender identity was not associated with co-occurring depressive symptoms and SUDs. It is possible that these findings resulted from low power as the odds of co-occurring depressive symptoms and SUDs were elevated among gender minority participants compared to cisgender participants. This was especially true for co-occurring depressive symptoms and drug use disorder (AOR: 2.38; 95% CI: 0.97–5.88).

Table 4.

Results of multivariable multinomial logistic regression models estimating associations of sexual orientation and gender identity with outcomes of co-occurring depressive symptoms and substance use disorders among Growing Up Today Study participants aged 20–35 years, assigned female at birth (2010, 2016)

	SEXUAL ORIENTATION							GENDER IDENTITY
	Completely Heterosexual	Mostly Heterosexual			Lesbian/Gay/Bisexual			Cisgender	Gender Minority
	AOR	AOR	95% CI	p-value	AOR	95% CI	p-value	AOR	AOR	95% CI	p-value
Outcome 1: Depressive Symptoms (DS) & Nicotine Dependence (ND)
No DS, No ND	1.0	1.00			1.00			1.0	1.00
DS Only	1.0	1.72	1.53–1.92	<.0001	2.64	2.17–3.21	<.0001	1.0	1.46	0.87–2.45	0.155
ND Only	1.0	2.55	2.06–3.16	<.0001	3.80	2.64–5.46	<.0001	1.0	0.76	0.25–2.33	0.638
DS & ND	1.0	4.35	3.42–5.53	<.0001	9.80	6.95–13.8	<.0001	1.0	1.10	0.40–3.01	0.857
Outcome 2: Depressive Symptoms (DS) & Alcohol Use Disorder (AUD)
No DS, No AUD	1.0	1.00			1.00			1.0	1.00
DS Only	1.0	1.80	1.60–2.02	<.0001	3.04	2.51–3.70	<.0001	1.0	1.54	0.90–2.63	0.118
AUD Only	1.0	2.33	1.94–2.80	<.0001	2.41	1.71–3.40	<.0001	1.0	1.52	0.69–3.35	0.303
DS & AUD	1.0	3.11	2.53–3.82	<.0001	3.66	2.60–5.15	<.0001	1.0	1.49	0.54–4.10	0.445
Outcome 3: Depressive Symptoms (DS) & Drug Use Disorder (DUD)
No DS, No DUD	1.0	1.00			1.00			1.0	1.00
DS Only	1.0	1.77	1.58–1.97	<.0001	2.89	2.40–3.49	<.0001	1.0	1.37	0.81–2.32	0.246
DUD Only	1.0	4.20	3.11–5.67	<.0001	7.94	5.13–12.3	<.0001	1.0	1.63	0.49–5.44	0.429
DS & DUD	1.0	4.84	3.59–6.52	<.0001	9.04	5.96–13.7	<.0001	1.0	2.38	0.97–5.88	0.059

Note: “Completely heterosexual” and “cisgender” participants are the referent groups for the sexual orientation and gender identity independent variables. Sexual orientation and gender identity included simultaneously in generalized estimating equations models adjusted for age, race/ethnicity, cohort, and region of residence. Associations in bold are p-value <.05

Table 5.

Results of multivariable multinomial logistic regression models estimating associations of sexual orientation and gender identity with outcomes of co-occurring depressive symptoms and substance use disorders among Growing Up Today Study participants aged 20–35 years, assigned male at birth (2010, 2016)

	SEXUAL ORIENTATION							GENDER IDENTITY
	Completely Heterosexual	Mostly Heterosexual			Lesbian/Gay/Bisexual			Cisgender	Gender Minority
	AOR	AOR	95% CI	p-value	AOR	95% CI	p-value	AOR	AOR	95% CI	p-value
Outcome 1: Depressive Symptoms (DS) & Nicotine Dependence (ND)
No DS, No ND	1.0	1.00			1.00			1.0	1.00
DS Only	1.0	3.11	2.53–3.82	<.0001	2.33	1.76–3.08	<.0001	1.0	1.96	1.02–3.76	0.044
ND Only	1.0	1.84	1.26–2.67	0.001	1.63	0.96–2.76	0.069	1.0	1.67	0.62–4.45	0.309
DS & ND	1.0	3.61	2.43–5.35	<.0001	3.37	2.00–5.68	<.0001	1.0	1.53	0.49–4.75	0.460
Outcome 2: Depressive Symptoms (DS) & Alcohol Use Disorder (AUD)
No DS, No AUD	1.0	1.00			1.00			1.0	1.00
DS Only	1.0	3.26	2.63–4.04	<.0001	2.42	1.79–3.28	<.0001	1.0	1.56	0.75–3.22	0.230
AUD Only	1.0	1.64	1.21–2.22	0.002	1.91	1.29–2.83	0.001	1.0	1.37	0.58–3.25	0.477
DS & AUD	1.0	3.09	2.24–4.27	<.0001	3.17	2.07–4.85	<.0001	1.0	2.75	1.24–6.10	0.013
Outcome 3: Depressive Symptoms (DS) & Drug Use Disorder (DUD)
No DS, No DUD	1.0	1.00			1.00			1.0	1.00
DS Only	1.0	2.95	2.40–3.64	<.0001	2.38	1.81–3.15	<.0001	1.0	1.86	0.99–3.50	0.054
DUD Only	1.0	2.17	1.47–3.22	0.0001	1.86	1.13–3.05	0.014	1.0	1.54	0.50–4.72	0.452
DS & DUD	1.0	4.87	3.35–7.08	<.0001	2.90	1.68–5.00	0.0001	1.0	1.86	0.59–5.88	0.292

Note: “Completely heterosexual” and “cisgender” participants are the referent groups for the sexual orientation and gender identity independent variables. Sexual orientation and gender identity included simultaneously in generalized estimating equations models adjusted for age, race/ethnicity, cohort, and region of residence. Associations in bold are p-value <.05

Among participants assigned male (Table 5), LGB and mostly heterosexual participants evidenced significantly elevated odds of depressive symptoms, SUDs, and co-occurrence in all but one instance (i.e., nicotine dependence among LGB males [AOR: 1.63; 95% CI: 0.96–2.76]). Sexual minority males had between three- and four-times higher odds of experiencing co-occurring depressive symptoms and SUDs. Gender minority participants assigned male evidenced 2.75 greater odds of experiencing co-occurring depressive symptoms and alcohol use disorder (95% CI: 1.24–6.10). Other outcomes of co-occurrence were elevated among gender minorities assigned male, though most differences failed to reach statistical significance, likely due to low power.

4. Discussion

This study quantified sexual orientation and gender identity differences in prevalences of co-occurring depressive symptoms and SUDs among the GUTS national cohort of young adults and tested sex assigned at birth as an effect modifier of the relationship between sexual orientation on co-occurrence. Overall, we found large sexual orientation disparities in co-occurring depressive symptoms across SUDs. Our findings are consistent with NESARC, which found greater lifetime prevalence of co-occurring mood and drug use disorders among adult sexual minorities compared to heterosexuals,^24,25 and extends this work by examining more recent assessments of co-occurrence (co-occurring past-week depressive symptoms and past 12-month SUDs) among young adults.

Our findings align with research finding heightened vulnerability for depression and SUDs among sexual minority young adult women.^17,19,20,40 Indeed, LGB participants assigned female in GUTS evidenced more than nine times greater odds of co-occurring depressive symptoms and SUDs than their completely heterosexual counterparts. This suggests that co-occurring depressive symptoms and SUDs represent significant mental and behavioral health disparities among young adult sexual minorities, with perhaps the greatest needs for prevention and intervention among LGBs assigned female. This may be explained through the lens of intersectionality, which elucidates the relationship between macro-level, interlocking systems of oppression, and individual-level health and social inequity.^41,42 Accordingly, LGBs assigned female may be uniquely exposed to the interlocking systems of homophobia and sexism, and in turn, evidence pronounced sexual orientation disparities in depressive symptoms and co-occurring SUDs.^43,44 Relatedly, this may be explained by the higher rates of depression, anxiety, and co-morbid psychiatric disorders among women in the general population,^45,46 which tend to precede the development of SUDs,⁴⁷ and the added stressor of exposure to sexual minority-related stigma and discrimination.^14–16

With respect to depressive symptoms only, we found that mostly heterosexual participants assigned male evidenced larger sexual orientation sexual disparities than did mostly heterosexual participants assigned female. This may be driven by experiences of minority related-stress and discrimination, and resultant depressive symptoms that are unique to mostly heterosexual men.¹⁷ More research is needed to quantify, explain, and mitigate the additional burden in co-occurring depressive symptoms and SUDs faced by LGB young adults assigned female, and depressive symptoms among mostly heterosexual young adults assigned male. In addition, research is needed to examine the intersectional influences of sexual orientation, sex/gender, and other social positions (e.g., race/ethnicity socioeconomic position) on depressive symptoms, SUDs, and co-occurrence.

Aligned with past research with GUTS^18,20 and in other young adult samples,^22,48 this study found evidence of higher prevalences of depressive symptoms and SUDs among gender minority participants compared to cisgender participants. Adding to the literature, we found evidence of elevated prevalences of co-occurring depressive symptoms and SUDs among gender minority young adults; however, in multivariable models, relationships between sexual orientation and co-occurrence were larger than relationships with gender identity. Nonetheless, odds were only significantly elevated in one co-occurrence outcome – depressive symptoms and alcohol use disorder among gender minorities assigned male. This result is likely driven by small numbers of gender minorities in the sample. Thus, these findings should be viewed as preliminary and interpreted with caution. Studies with larger samples of gender minorities are needed to better understand risk for co-occurrence of depression and SUDs in this population.

Researchers and clinicians recommend integrating treatment of depression and SUDs given evidence that integrated treatment is superior to separate approaches.^49,50 Our findings of heightened odds of co-occurring depressive symptoms and SUDs among SGM young adults support national recommendations for universal adult screenings for harmful substance use and depression in primary care settings.^51,52 Additionally, although tobacco screening and cessation services may be more commonly offered in primary care,⁵³ depression and substance use screenings have historically lagged.⁵⁴ This study suggests that primary care providers may play a critical role in detecting both conditions early in the course given that approximately 70% of young adults identify a regular source of primary care⁵⁵ and the difficulty young adults encounter accessing behavioral health treatment.⁵⁶ In addition, past research has found that SGMs use mental health services more than non-SGMs,⁵⁷ perhaps related to higher rates of depression and SUDs. As such, mental health and addiction specialty care providers should assess for co-occurring conditions among SGM populations.^58,59

While there is evidence for the effectiveness of depression and drug screening, alcohol screening, and brief intervention for adult harmful drinking in primary care, and effective pharmacotherapy and behavioral health treatments for mental health and SUDs,^51,52,60,61 there are few behavioral health interventions specifically targeted for young adults.^62–64 Greater attention to the development and testing of interventions for this age group, including those specifically for young adult SGMs, may help reduce disparities in depression, SUDs, and co-occurrence.⁶⁵ It is also critical that culturally-relevant harm reduction approaches be used to mitigate disparities and improve treatment outcomes for SGM young adults with co-occurring depression and SUDs. This means that treatment and interventions should be tailored to individual needs, personal goals, and sociocultural contexts, while simultaneously decreasing the harms associated with substance use, rather than emphasizing substance abstinence as the end-goal.^66,67 Such approaches may serve as a bridge to long-term recovery efforts for those with co-occurring depression and SUDs. Our own qualitative research with SGM GUTS participants supports the need for harm reduction approaches in clinical treatment among SGM young adults with SUDs.⁶⁸ In addition, interventions at the provider/clinic level that promote culturally-humble harm reduction treatment approaches for SGM young adults with depression, SUDs, or co-occurrence are warranted.^68,69

A study limitation is in the temporal assessment of our main outcomes in which we construct co-occurrence as having past-week depressive symptoms and past 12-month SUDs. Although past-week depressive symptoms could co-occur with a past 12-month probable SUD, we may have missclassifed short-term, situational depressive symptoms as co-occurring with a past 12-month SUD. In addition, we used data from the 2010 general GUTS questionnaire (GUTS I only) and 2015–2017 GUTS Substance Substudy questionnaire (GUTS I, II) to assess SUDs, whereas we drew from the 2010 (GUTS I only) and 2016 (GUTS I, II) general questionnaires to assess depressive symptoms. While there is temporal overlap between the 2015–2017 Substudy and the 2016 general questionaire, these assessments were not simultaneous. Despite these limitations, our study is a step towards precise measurements of co-occurrence. These findings elucidate substantial disparities in co-occurrence among SGM young adults in a large national cohort, and provide information germane to clincal screening and detection, and warranting further epidemiologic investigation.

Our findings should be considered in the context of the predominately non-Hispanic white sample of participants whose mothers are nurses. As such, these findings may not generalize to racial/ethnically and socioeconomically diverse groups of young adults in the U.S.; however health-related phenomena documented among GUTS participants are similar to those observed in more diverse samples.⁷⁰ The lack of racial/ethnic diversity in our sample may have missed potentially larger disparities in depressive symptoms, SUDs, and co-occurrence among SGM young adults of color shaped by interlocking systems of homophobia and/or transphobia, and racism.^41,42 Indeed, research finds a multiplicative relationship between sexual orientation and race/ethnicity such that, in some instances, those with multiply marginalized social identities evidence larger disparities in substance use and disorders, and negative mental health outcomes relative to their peers with one or no minority identities.^71,72 Another consideration is loss to follow up, as bias would be introduced in this study if attrition was differentially related to relationships of sexual orientation or gender identity with depressive symptoms, SUDs, or co-occurrence. While we might expect participants with depressive symptoms and/or SUDs to be more likely to be lost to follow up, it is difficult to determine the potential impact of attrition bias on our findings.

Finally, sexual orientation is fluid among young adults⁷³ and past research has shown that sexual orientation fluidity is positively associated with substance use⁷⁴ and depressive symptoms⁷⁵ in young adulthood; however, our measure of sexual orientation did not measure fluidity over time. In addition, while our measure assessed two of three dimensions of sexual orientation in a single question, i.e., identity and attraction, it did not measure sexual behavior. Measurement of sexual orientation fluidity or sexual behavior may have provided additional information on how sexual orientation is related to co-occurring depressive symptoms and SUDs.⁷⁶ Finally, due to low statistical power, we combined lesbian/gay and bisexual participants into one category, potentially obscuring sub-group differences, as research suggests bisexuals may experience mental health and substance use disparities relative to other sexual minority groups.^21,77,78

Future research should overcome these limitations and attend to gaps in the literature by examining sexual orientation and gender identity differences in co-occurring depressive symptoms and SUDs among nationally representative and racially/ethnically diverse samples of young adults. In addition, qualitative and longitudinal research is needed to examine the multilevel drivers of disparities in depression and co-occurring SUD across diverse samples of SGM young adults. Youth-engaged and/or youth-led research is also needed to inform the development of culturally humble interventions to support mental health equity and healthful coping strategies among SGM adolescents and young adult.

In sum, this study examined patterns of depressive symptoms, SUDs, and co-occurrence by sexual orientation and gender identity among young adults participating in a national U.S. cohort, finding large sexual orientation disparities in co-occurrence, particularly among LGB participants assigned female at birth. Despite political gains for and growing acceptance of SGM people in the last decade,^79–82 co-occurrence remained elevated among SGMs between 2010 and 2016. Coupled with past research, these findings suggest a mental health imperative to identify and address structural antecedents and health services barriers that contribute to SGM young adult disparities in co-occurring depression and SUDs.