Longitudinal Associations between Minority Stress, Internalizing Symptoms, and Substance Use among Sexual and Gender Minority Individuals Assigned Female at Birth

Abstract

Objective:

Sexual and gender minority individuals assigned female at birth (SGM-AFAB) are at increased risk for anxiety, depression, and substance use and problems compared to heterosexual cisgender women. Cross-sectional research has demonstrated that minority stressors are associated with anxiety, depression, and substance use. However, longitudinal research is limited and the examination of prospective associations between minority stressors, mental health, and substance use is even more sparse.

Methods:

We utilized four waves of data (6 months between waves) from a diverse (26.0% non-Latinx White; 26.2% gender minorities) longitudinal cohort of 488 SGM-AFAB (16–32 years old at Wave 1) to examine concurrent and prospective associations between three minority stressors (internalized stigma, microaggressions, victimization) and anxiety, depression, and alcohol and cannabis use and problems.

Results:

At the within-person level, results indicated that when SGM-AFAB experienced more minority stressors than usual, they reported more concurrent and prospective anxiety and depression. Additionally, when SGM-AFAB experienced more microaggressions than usual, they were more likely to use alcohol and cannabis, and when they experienced more victimization than usual, they reported more concurrent alcohol and cannabis use problems. No prospective associations between minority stressors and substance use were significant.

Conclusions:

Findings indicate that minority stressors were consistently associated with internalizing symptoms, both concurrently and prospectively, while evidence for associations between minority stressors and substance use/problems was less consistent. These findings highlight the need for interventions that that teach SGM-AFAB skills for effectively coping with minority stress.

Sexual and gender minority is an umbrella term that is used to refer to individuals who are sexual minorities (i.e., individuals who identify as lesbian, bisexual, pansexual, or with another non-heterosexual identity label and/or report same-sex attractions and/or sexual partners) and/or gender minorities (i.e., individuals who do not identify with the gender that is typically associated with their sex assigned at birth). Sexual and gender minority individuals assigned female at birth (SGM-AFAB)¹ are at increased risk for anxiety, depression (King et al., 2008; Meyer, Brown, Herman, Reisner, & Bockting, 2017), and substance use (SU) and related problems (Day, Fish, Perez-Brumer, Hatzenbuehler, & Russell, 2017; McCabe, Hughes, Bostwick, West, & Boyd, 2009) compared to heterosexual cisgender women. Their heightened risk has been attributed to minority stress – the chronic stress they experience due to the stigmatization of SGM (Meyer, 2003). Minority stressors include experiences of biased treatment by others due to one’s sexual or gender identity (e.g., discrimination) and internalized negative attitudes about one’s own sexual or gender identity (e.g., internalized stigma), referred to as enacted and felt stigma, respectively. While a growing body of research has provided support for minority stress theory, this research has limitations. First, the majority of this research has been cross-sectional, with few longitudinal studies (Kidd, Jackman, Wolff, Veldhuis, & Hughes, 2018; Newcomb & Mustanski, 2010). Second, SGM-AFAB remain under-represented, with most research either focusing on SGM-AMAB (assigned male at birth) or treating SGM as a homogeneous group and failing to present sex or gender specific effects (e.g., Coulter, Kenst, & Bowen, 2014; Kidd et al., 2018). The current study aims to address these limitations by examining longitudinal associations between minority stressors and internalizing symptoms, SU, and related problems among a sample of SGM-AFAB.

Minority Stress and Internalizing Symptoms

Studies examining cross-sectional associations between minority stressors and internalizing symptoms have consistently demonstrated that SGM individuals (both SGM-AFAB and -AMAB) who experience more enacted or felt stigma also experience more anxiety and depression (e.g., Everett, Hatzenbuehler, & Hughes, 2016; Hall, 2018; Kaysen et al., 2014; Lehavot & Simoni, 2011; Logie, Lacombe-Duncan, Poteat, & Wagner, 2017; Ramirez & Paz Galupo, 2019; Szymanski, Dunn, & Ikizler, 2014). However, this cross-sectional research can only test between-persons associations. In other words, it can only identify which individuals are likely to have higher levels of internalizing symptoms than others (e.g., those who experience more minority stress; (Bolger & Laurenceau, 2013). Longitudinal research examining within-person associations between minority stressors and internalizing symptoms across time, in contrast, offers the advantage of using each individual as their own control. Specifically, concurrent within-person associations can test whether changes in the level of minority stressors an individual experiences occur around the same time as changes in internalizing symptoms (Bolger & Laurenceau, 2013), effectively ruling out between-persons causes of the outcome (e.g., differences in participants’ ages at the first wave). Prospective within-person associations, in which an individual’s level of minority stressors at each wave is used to predict changes in their internalizing symptoms from that wave to the next, can also test the directionality of effects (Bolger & Laurenceau, 2013). As such, they can determine whether an increase in minority stressors precedes an increase in internalizing symptoms, providing evidence that is consistent or inconsistent with the central tenet of Meyer’s (2003) theory – that minority stress has a causal effect on health. This represents a significant contribution to the minority stress literature because the majority of studies are cross-sectional and thus cannot test this fundamental tenet.

A handful of studies have found evidence of within-person longitudinal associations between minority stressors and internalizing symptoms among SGM individuals. Findings indicate that during waves when SGM experienced more enacted and felt stigma, they also experienced higher anxiety and depression (Birkett, Newcomb, & Mustanski, 2015; Pachankis, Sullivan, Feinstein, & Newcomb, 2018). However, evidence for time-lagged associations between minority stressors and internalizing symptoms is somewhat mixed. Some studies have indicated that when SGM experience more felt stigma at one wave, they experience more anxiety and depression at the subsequent wave (Eldahan et al., 2016; Rendina et al., 2017). Additionally, experiencing more SGM victimization at one wave has predicted higher anxiety and depression at the next wave (6–12 months later) even controlling for internalizing symptoms at the prior wave (Birkett et al., 2015; Burton, Marshal, Chisolm, Sucato, & Friedman, 2013; Tucker et al., 2016). However, at least one study has not found evidence of time-lagged associations between either enacted or felt sigma and depression or anxiety (Pachankis et al., 2018).

Of note, these longitudinal studies have either examined associations between minority stressors and internalizing symptoms in exclusively SGM-AMAB samples (Eldahan et al., 2016; Pachankis et al., 2018; Rendina et al., 2017) or have included both SGM-AMAB and -AFAB but treated them as a single homogeneous group (Birkett et al., 2015; Burton et al., 2013; Tucker et al., 2016). Little to no research has longitudinally examined these associations separately among SGM-AFAB. Given this under-representation, further longitudinal research on associations between minority stressors and internalizing symptoms among SGM-AFAB samples is needed.

Minority Stress and Substance Use

A number of studies have examined cross-sectional associations between minority stressors and SU and related problems. These studies have consistently linked enacted stigma with a higher likelihood of alcohol and drug use (Lowry, Johns, Robin, & Kann, 2017; Marshal, Burton, Chisolm, Sucato, & Friedman, 2013) and more severe SU problems (McCabe, Hughes, West, Veliz, & Boyd, 2019). Felt stigma has also been associated with drug use in several studies (Goldbach, Schrager, Dunlap, & Holloway, 2015; Kelly, Davis, & Schlesinger, 2015), although its associations with alcohol use problems have been found in some samples (Feinstein & Newcomb, 2016; Kuerbis et al., 2016; Lehavot & Simoni, 2011) but not others (Flood, McLaughlin, & Prentice, 2013; Lea, de Wit, & Reynolds, 2014). Overall, cross-sectional studies indicate that SGM who experience more minority stressors tend to be more likely to use substances and have SU problems. However, SGM-AFAB remain under-represented in this research, with one recent meta-analysis indicating that less than one quarter of studies that examined associations between risk and protective factors and SU did so separately for SGM-AFAB (Kidd et al., 2018).

While relatively few studies have longitudinally examined associations between minority stressors and SU, these studies have more frequently examined effects specifically among SGM-AFAB. Findings indicate that when SGM-AFAB experienced more SGM victimization, they drank more alcohol during that wave (Newcomb, Heinz, & Mustanski, 2012) and experienced increases in binge drinking from that wave to the next (i.e., 12 months later) (Dermody, Marshal, Burton, & Chisolm, 2016). Additionally, when SGM-AFAB experienced more minority stressors during one wave, they experienced subsequent increases in negative consequences of alcohol use but not necessarily greater frequency of alcohol use over the next 12 months (Wilson, Gilmore, Rhew, Hodge, & Kaysen, 2016). Longitudinal links between SGM victimization and alcohol use has been slightly less consistent among SGM-AMAB, with one study indicating that SGM-AMAB had more alcohol use problems during waves when they experienced more SGM victimization (Dyar, Newcomb, & Mustanski, 2019) and another finding no association (Newcomb et al., 2012). While SGM-AFAB specific effects are more common in longitudinal studies of minority stressors and SU, these studies have focused almost exclusively on the association between enacted stigma and alcohol use. Little to no research has longitudinally examined associations between other minority stressors and alcohol use or between minority stressors and cannabis use specifically among SGM-AFAB.

Minority Stress and General Stress

An underlying assumption of Meyer’s (2003) minority stress theory is that minority stressors uniquely contribute to mental health outcomes over and above general stressors experienced by all people. Relatively little research on minority stressors among SGM has differentiated the impact of general stress from that of minority stress on SU or internalizing symptoms. A cross-sectional study of depression among transgender older adults found that internalized stigma no longer predicted depression when general stress was controlled (Hoy-Ellis & Fredriksen-Goldsen, 2017). It is unclear, however, how generalizable the results of this study of transgender older adults are to other subgroups of SGM. In fact, authors of a cross-sectional study conducted with sexual minority women found that, after controlling for general stress, enacted and internalized stigma continued to be significantly associated with psychological distress (Szymanski et al., 2014). We are aware of only one observational longitudinal study that has accounted for general stress in concurrent or prospective associations between minority stressors and SU/problems (Dyar et al., 2019) and none that have done so for associations between minority stressors and internalizing symptoms. In a sample of SGM-AMAB, Dyar and colleagues (2019) found that when general stress was controlled, experiencing more enacted stigma than usual at one wave continued to be significantly associated with more alcohol and cannabis use problems than usual among users during the same wave, while a similar association between internalized stigma and alcohol problems was no longer significant. Of note, a recent experimental study found that exposure to stigma-related stress predicted more negative affect and alcohol cravings than exposure to general stress in a small sample of SGM (Mereish & Miranda Jr, 2019). Overall, research that parses the effects of minority stress and general stress is limited. One of the underlying assumptions of the concept of minority stress is that it is distinct from and additive to general stress. Given the limited number of experimental and longitudinal studies to examine this fundamental assumption of minority stress theory, more longitudinal research is needed that differentiates the unique contributions of minority stressors over and above the effects of general stress.

Current Study

The goal of the current study is to extend the existing literature by examining longitudinal associations between three minority stressors (internalized stigma, microaggressions, victimization) and anxiety, depression, and alcohol and cannabis use and problems among a sample of SGM-AFAB. We aimed to examine within-person associations between minority stressors at one wave and internalizing symptoms and SU/problems during the same wave (concurrent associations) and at the next wave (prospective associations). Concurrent within-person analyses were used to determine whether changes in minority stressors tended to co-occur with changes in internalizing symptoms, SU, and substance problems. Prospective analyses were used to determine whether changes in minority stressors precede changes in internalizing symptoms and SU, consistent with a causal model in which minority stressors lead to changes in these outcomes. Additionally, we examined whether associations between minority stressors and anxiety, depression, and SU/problems persist when general stress is controlled. Controlling for general stress allows us to test the underlying assumption that minority stressors impacts mental health over and above general stressors (Meyer, 2003). We proposed the following hypotheses.

1. Within-persons, concurrent associations: During waves when individuals experienced more minority and/or general stress than usual (i.e., more than the individual experienced on average across observations), they would experience more anxiety and depression and be more likely to use alcohol and cannabis. Among those who used substances, experiencing more stress than usual would be associated with more alcohol and cannabis problems (i.e., physical, psychological, and social consequences of SU).

2. Within-persons, prospective associations: When individuals experienced more minority and/or general stress than usual at one wave, they would experience an increase in anxiety and depression and the likelihood of alcohol or cannabis use from that wave to the next. Also, among those who used substances, experiencing more stress than usual during one wave would be associated with subsequent increases in alcohol and cannabis problems.

3. Between-persons: Individuals who experienced more minority and/or general stress across waves would also have more anxiety and depression, be more likely to use alcohol and cannabis, and report more SU problems on average across waves.

4. We hypothesized that when general stress was controlled for, experiencing more minority stressors would continue to concurrently and prospectively predict internalizing symptoms, SU, and related problems at the within-person level.

Method

Participants and Procedures

Current analyses used data from an ongoing longitudinal cohort study of young sexual and gender minorities assigned female at birth (SGM-AFAB), referred to as FAB400. Data collection began in November 2016 and is ongoing. To achieve a multiple cohort, accelerated longitudinal design, SGM-AFAB from a prior cohort study of SGM (originally recruited in 2007) and a new cohort of SGM-AFAB were both recruited in 2016–17 using venue-based recruitment, social media, and incentivized snowball sampling. At original cohort enrollment, participants were 16–20 years old, assigned female at birth, and either identified with a sexual or gender minority label or reported same-sex attractions or sexual behavior. Cohort members could refer up to five peers and were paid $10 for each peer they recruited. Participants completed study visits at 6-month intervals and were paid $50 for each wave. The study protocol was approved by the Institutional Review Board (IRB) at Northwestern University. See Whitton, Dyar, Mustanski, and Newcomb (2019) for further details about the study design.

The current study uses data from the first assessment (Wave 1) and 6-, 12-, and 18-month follow-up assessments (Waves 2–4). Retention rates for Waves 2, 3, and 4 were 96.7%, 94.9%, and 92.8% respectively. Demographic information about the sample from the Wave 1 assessment is presented in Table 1. The sample was comprised predominately of cisgender women (73.8%), with a sizeable subsample of gender minority individuals (26.2%). Of the gender minority participants, 98.5% were also sexual minorities (based on sexual identity and/or sexual attractions). The two transmen in the sample who identified as straight and reported being only attracted to women did not complete measures of internalized sexual minority stigma or sexual minority microaggressions and thus were excluded from analyses including those variables. The sample is diverse in sexual identity and race/ethnicity (26.0% non-Latinx White) and participants were aged 16–32 at Wave 1 of FAB 400.

Table 1.

Baseline Demographics

Demographics	N	%
Cohort
2016–17 Cohort	400	82.0%
2007 Cohort	88	18.0%
Race/Ethnicity
White	127	26.0%
Black	170	34.8%
Latinx	120	24.6%
Other	71	14.5%
Participant Gender
Cisgender Women	360	73.8%
Transgender or Male	44	9.0%
Genderqueer/Non-Binary	84	17.2%
Sexual Identity
Lesbian	115	23.6%
Bisexual	181	37.1%
Queer	64	13.1%
Pansexual	82	16.8%
Other Sexual Identity	46	9.4%
Household Income
< $20,000	95	19.5%
$20,000-$39,999	110	22.5%
$40,000-$59,999	105	21.5%
$60,000-$79,999	65	13.3%
≥ $80,000	109	22.3%
Missing	4	0.8%
Age (M, SD)	20.06 (3.66)

Measures

General stress.

The 4-item Perceived Stress Scale (Cohen, 1988; Cohen, Kamarck, & Mermelstein, 1983) assessed general stress. Participants were asked, “In the past month, how often have you [item]?” Example: “Felt difficulties were piling up so high that you could not overcome them.” Items were rated from 0 (never) to 4 (very often) and summed. α ranged from .73–.79 across waves.

SGM victimization.

Ten items assessed victimization (Mustanski, Andrews, & Puckett, 2016). Participants were asked, “In the past six months, how many times [item] because you are or were thought to be gay, bisexual, or transgender?” Example: “have you had an object thrown at you.” Items were rated from 0 (never) to 3 (three times or more) and averaged (α=.68–.82).

Sexual orientation based microaggressions.

Nineteen items from the Sexual Orientation Microaggression Inventory (Swann, Minshew, Newcomb, & Mustanski, 2016) plus four additional items that assessed unique microaggressions experienced by SGM-AFAB (e.g., “Someone assured you that if you tried or altered your appearance in some way, a man would want to date you”) were used to assess microaggressions based on one’s sexual orientation. Participants were asked, “In the past 6 months, how often have you had the following experiences?” Example: “You heard someone say ‘that’s so gay’ in a negative way.” Items were rated from 1 (not at all) to 5 (about every day) and averaged (α=.91–.94). A bi-factor analysis of the 23 items in the current sample indicated that 79.0% of the variance in the items was accounted for by the general factor, indicating that the use of a total score is appropriate for this measure. Additionally, a moderate to strong correlation between microaggressions and victimization and small to moderate correlations between microaggressions and internalizing symptoms also provide initial evidence of the validity of the measure.

Sexual orientation based internalized stigma.

The 8-item desire to be heterosexual subscale of Puckett et al.’s (2017) validated measure of internalized stigma, which was recommended as the most internally valid subscale, was used to assess internalized stigma. The measure was developed based on prior work (Nungesser, 1983; Ramirez-Valles, Kuhns, Campbell, & Diaz, 2010). Participants were asked how much they agreed with statements like “Sometimes I think that if I were straight, I would probably be happier” on a scale of 1 (strongly disagree) to 4 (strongly agree). Items were averaged (α=.84–.89). Although this measure was developed and validated in sample of SGM-AMAB, initial psychometric analyses indicate that it performed similarly in the current sample of SGM-AFAB by demonstrating a unidimensional factor structure and small to moderate correlations with victimization, anxiety, and depression.

Alcohol problems.

The AUDIT (Saunders, Aasland, Babor, De la Fuente, & Grant, 1993) assessed alcohol use and problems in the past six months. The AUDIT includes 10 items rated on different scales. “How often do you have a drink containing alcohol?” was rated from 1 (never) to 5 (4 or more times a week). Responses were summed. Scores ranged from 0 to 40 (α=.76–.81), with scores of 8–15 indicating moderate alcohol problems and 16+ indicating severe alcohol problems.

Cannabis problems.

The revised CUDIT (Adamson et al., 2010) assessed cannabis use and problems in the past six months. CUDIT includes eight items rated on different scales. For example, the item “How often do you use marijuana?” was rated from 1 (never) to 5 (4 or more times a week). Responses were summed. Scores ranged from 0 to 32 (α=.75–.78), with scores of 8–11 indicating hazardous use and 12+ indicating a possible cannabis use disorder.

Anxiety and depression.

Symptoms of anxiety and depression were measured by the PROMIS Anxiety and Depression Short Forms 8a (Pilkonis et al., 2011). Each scale includes 8 items (e.g., “I felt fearful” and “I felt worthless,” respectively; α = .94–.95) measured on a scale of 1 (never) to 5 (always). These measures were developed using rigorous psychometric methodology and have demonstrated strong internal consistency (α = .93–.95) and content validity (e.g., high correlations with other established measures of symptoms of anxiety and depression). Items are summed. Scores range from 8 to 40.

Data Analysis

Analyses were conducted in Mplus version 8.1. There were a total of 1879 observations from 4 waves and 488 participants (3.7% of observations missing). Within completed assessments, 0.01% of data were missing and were handled using full information maximum likelihood. In each model, within and between-person components of one stress variable (e.g., internalized stigma) predicted a mental health or SU outcome. Associations between within-person stressors and outcomes were allowed to vary across individuals. In all models, the linear association between within-person age and the outcome was included to control for developmental changes in mental health and SU. This effect was modeled as fixed due to non-convergence when the effect was random. Age at Wave 1, sexual identity, race/ethnicity, and gender were controlled. We examined concurrent within-person associations (stressors at each wave predicting outcomes at the same wave) and prospective within-person associations (stressors predicting outcomes 6 months later, controlling for outcome at the previous time point). We also examined a series of models with a minority stressor and general stress entered simultaneously to assess their unique effects.

For models examining internalizing symptoms, multilevel structural equation modeling (MSEM) with a Bayesian estimator and the default of diffuse (non-informative) priors were used. Bayesian MSEM has several advantages over traditional multilevel modeling using maximum likelihood estimation (Depaoli & Clifton, 2015). MSEM estimates between-person variables with more reliability and less bias than standard multilevel approaches (Preacher, Zyphur, & Zhang, 2010). MSEM treats repeated measures as indicators of latent variables, which estimate the between-person level variable while adjusting for non-independence at the within-person level (Marsh et al., 2009). Using a Bayesian estimator can overcome problems with convergence that MSEM models using a maximum likelihood estimator can have (Depaoli & Clifton, 2015). We used Markov Chain Monte Carlo (MCMC) algorithms to generate a series of 10,000 random draws from the multivariate posterior distribution of our sample for each model. Trace plots and the Gelman-Rubin potential scaling reduction (PSR) were used to determine whether convergence was achieved (Depaoli & Clifton, 2015; Muthén, 2010).

A slightly different analytic approach was required for SU models because AUDIT and CUDIT scores combine two qualitatively different pieces of information. First, individuals who have a score of 0 do not use the substance, while individuals who use the substance have a score of 1 or higher. For individuals who use the substance, the size of the non-zero score indicates the severity of their SU problems (i.e., severity of physical and psychological symptoms of heavy SU or dependence and social consequences of SU). To maximize the utility of this data, we used generalized linear mixed models (GLMMs) with negative binomial hurdle distributions to test associations between minority stressors and SU. In each model, logistic regression estimated the odds ratio (OR) for likelihood of having a zero versus non-zero value on AUDIT or CUDIT (i.e., use vs. no use) and a truncated negative binomial model estimated rate ratios (RRs) for the non-zero count (i.e., severity of SU problems among users). This parses the association between a stressor and SU into four associations (Atkins, Baldwin, Zheng, Gallop, & Neighbors, 2013). At the between-person level, associations between an individuals’ average level of stress (across waves) and: (1) their likelihood of using the substance at least once during the study and (2) their average level of SU problems (non-zero count; aggregated across waves) are modeled. At the within-person level, associations between an individuals’ deviation from their usual level of stress (e.g., experiencing more/less stress than their average) and (3) their likelihood of using the substance during the same/next six-month period and (4) their deviation from their average level of SU problems during the same/next period are modeled. As MSEM and Bayesian estimation are not available for outcomes with hurdle distributions in Mplus, multilevel modeling with a robust maximum likelihood estimator was used and predictors were separated into within- and between-persons components by person-(within-persons) and grand-mean centering (between-persons; Enders & Tofighi, 2007).

Results

See Table 2 for correlations, means, standard deviations, and intraclass correlations (ICCs). No alcohol use was reported at 24.5% of observations and 12.1% of participants reported no alcohol use at any wave. No cannabis use was reported at 37.7% of observations and 22.7% of participants reported no cannabis use at any wave. AUDIT and CUDIT means were 3.79 (SD = 3.43) and 4.86 (SD = 4.66), respectively. Multilevel associations between demographics and mental health and SU/problems (Table 3) were examined to identify demographic covariates. Results indicated that participants experienced a decrease in internalizing symptoms over time and an increase in cannabis use problems. Participants who were older at baseline reported less internalizing symptoms, on average (i.e., across waves), more cannabis use problems, and were more likely to use alcohol. Gender minorities reported more anxiety than cisgender women. Compared to White participants, Black participants reported less anxiety and alcohol use problems, more cannabis use problems, and were less likely to use alcohol during the study. Additionally, Latinx participants reported fewer alcohol use problems than White participants and individuals who identified with other racial/ethnic identities reported less anxiety and depression than White participants. Although no significant differences emerged based on sexual identity, several differences approached significance. Given these associations with the outcomes of interest age, sexual identity, gender identity, and race/ethnicity were included as covariates in subsequent analyses. Next, tests of hypotheses regarding concurrent and prospective associations between stress and internalizing symptoms are discussed.

Table 2.

Correlations and Descriptive Statistics

	Internalized Stigma	Microaggressions	Victimization	General Stress	Anxiety	Depression
Internalized Stigma	-	.12*	.10*	.12*	.11*	.13*
Microaggressions	.17*	-	.44*	.17*	.21*	.23*
Victimization	.12*	.66*	-	.16*	.13*	.16*
General Stress	.15*	.26*	.15*	-	.41*	.51*
Anxiety	.28*	.30*	.24*	.65*	-	.67*
Depression	.32*	.34*	.20*	.75*	.80*	-
Mean	1.65	1.61	.19	7.78	19.10	17.56
Standard Deviation	.48	.42	.28	2.27	6.12	5.68
Range	1–4	1–5	0–3	0–16	8–40	8–40
ICC	.70	.58	.59	.47	.59	.53

Correlations above the diagonal are within-persons correlations, while those below the diagonal are between-person correlations. Scores for internalized stigma, microaggressions, and victimization were calculated by averaging item responses, while scores for all other variables were calculated by summing item responses. For all measures higher scores indicate higher levels of the construct.

^*p < .05.

Table 3.

Multilevel Associations between Demographics and Substance Use

				AUDIT		CUDIT
Predictor	Level	Depression (b)	Anxiety (b)	Any Alcohol Use (OR)	Alcohol Use Problems (RR)	Any Cannabis Use (OR)	Cannabis Use Problems (RR)
Within-Person Associations	Age	−.25 [−.30, −.20]	−.19 [−.24, −.14]	1.08 [.99, 1.17]	1.01 [.98, 1.05]	1.00 [.94, 1.08]	1.06 [1.02, 1.09]
Between-Person	Age at Baseline	−.18 [−.28, −.08]	−.16 [−.25, −.06]	1.29 [1.05, 1.58]	1.05 [.95, 1.15]	.94 [.79, 1.12]	1.10 [1.02, 1.18]
Associations	Sexual Identity: Bisexual	.01 [−.11, .13]	−.002 [−.11, .12]	1.21 [.78, 1.88]	1.17 [.94, 1.45]	1.30 [.86, 1.96]	1.00 [.83, 1.20]
	Sexual Identity: Queer	.06 [−.06, .18]	.06 [−.05, .17]	1.45 [.74, 2.85]	1.07 [.79, 1.43]	1.17 [.65, 2.09]	.97 [.75, 1.26]
	Sexual Identity: Pansexual	.11 [−.01, .23]	.09 [−.02, .20]	1.59 [.90, 2.78]	1.24 [.97, 1.59]	1.63 [.96, 2.77]	1.08 [.85, 1.36]
	Sexual Identity: Other	.04 [−.07, .14]	−.01 [−.12, .09]	.59 [.33, 1.07]	.86 [.60, 1.22]	.67 [.37, 1.20]	1.00 [.75, 1.31]
	Gender Minority	.10 [−.004, .20]	.19 [.08, .28]	.77 [.51, 1.19]	.94 [.78, 1.14]	1.00 [.67, 1.49]	1.00 [.83, 1.19]
	Race: Black	−.09 [−.21, .04]	−.23 [−.35, −.11]	.48 [.30, .78]	.62 [.50, .76]	.74 [.48, 1.15]	1.40 [1.15, 1.69]
	Race: Latinx	−.06 [−.18, .06]	−.08 [−.19, .03]	.90 [.53, 1.51]	.68 [.55, .84]	.95 [.60, 1.50]	1.04 [.85, 1.26]
	Race: Other	−.11 [−.22, −.01]	−.11 [−.22, −.06]	.92 [.50, 1.72]	.84 [.64, 1.10]	.98 [.58, 1.65]	1.13 [.89, 1.44]

Note. Sexual identity is dummy coded with lesbian/gay as the reference group. Race is dummy coded with White as the reference group. 95% confidence intervals are presented. Significant associations are bold.

Stress and Internalizing Symptoms (Table 4)

Table 4.

Concurrent and prospective associations between stress and internalizing symptoms

		Depression		Anxiety
Predictor	Level	Concurrent	Prospective	Concurrent	Prospective
Internalized stigma	Within	.08 [.05, .11]	.16 [.08, .23]	.07 [.04, .09]	.17 [.11, .24]
	Between	.34 [.03, .75]	-	.29 [.01, .56]	-
Microaggressions	Within	.11 [.07, .15]	.13 [.05, .20]	.10 [.06, .13]	.13 [.06, .20]
	Between	.46 [.002, .83]	-	.59 [.11, .95]	-
Victimization	Within	.10 [.06, .13]	.10 [.03, .19]	.16 [.09, .22]	.55 [.21, 1.00]
	Between	.56 [.17, .82]	-	.57 [.14, .88]	-
General Stress	Within	.51 [.46, .57]	−.04 [−.43, .42]	.41 [.36, .47]	.05 [−.38, .61]
	Between	.83 [.77, .88]	-	.72 [.65, .79]	-

All models were estimated controlling for linear change in internalizing symptoms over time, age at baseline, sexual identity, gender identity, race/ethnicity, and (for prospective associations) outcome at the previous wave. 95% confidence intervals are presented. Significant associations are bold.

Within-person concurrent associations.

Internalized stigma, microaggressions, victimization, and general stress were concurrently associated with anxiety and depression, such that at waves when an individual experienced more stress than they usually experienced (i.e., more than the individual experienced on average across observations), they reported more concurrent anxiety and depression than usual.

Within-person prospective associations.

Internalized stigma, microaggressions, and victimization were prospectively associated with anxiety and depression, such that when an individual experienced more stress than usual, they experienced an increase in anxiety and depression 6 months later. General stress did not prospectively predict anxiety and depression.

Between-person associations.

All four stressors were associated with anxiety and depression at the between-person level. This indicates that individuals who experienced more of these stressors on average across time points also tended to experience more internalizing symptoms across time points.

Stress and SU/Problems (Table 5)

Table 5.

Generalized mixed-effects hurdle models: Concurrent and prospective associations between stress and substance use

		AUDIT		CUDIT
Predictor	Level	Any Alcohol Use (OR)a	Alcohol Use Problems (RR)b	Any Marijuana Use (OR)a	Marijuana Use Problems (RR)b
Internalized stigma	Within Concurrent	1.16 [1.03, 1.31]	1.03 [.88, 1.20]	1.31 [.96, 1.79]	.91 [.82, 1.01]
	Within Prospective	1.11 [.97, 1.26]	.98 [.93, 1.04]	.95 [.53, 1.71]	.96 [.82, 1.13]
	Between	1.51 [1.19, 1.91]	1.32 [1.12, 1.56]	1.07 [.96, 1.21]	1.08 [.91, 1.28]
Microaggressions	Within Concurrent	1.58 [1.11, 2.24]	.98 [.85, 1.13]	1.08 [1.02, 1.15]	.96 [.91, 1.01]
	Within Prospective	.90 [.70, 1.17]	.82 [.45, 1.48]	1.67 [.59, 4.71]	.94 [.83, 1.07]
	Between	1.63 [1.24, 2.16]	1.28 [1.07, 1.55]	2.22 [1.73, 2.86]	1.27 [1.09, 1.47]
Victimization	Within Concurrent	.92 [.72, 1.18]	1.31 [1.06, 1.63]	1.11 [.99, 1.24]	1.22 [1.18, 1.26]
	Within Prospective	1.12 [.40, 3.19]	.76 [.54, 1.07]	.76 [.60, 1.04]	1.38 [.54, 3.49]
	Between	1.53 [.94, 2.46]	1.73 [1.29, 2.32]	2.70 [1.67, 4.36]	1.40 [1.13, 1.74]
General Stress	Within Concurrent	1.04 [1.03, 1.05]	1.00 [.98, 1.02]	1.03 [1.00, 1.06]	1.01 [1.00, 1.02]
	Within Prospective	1.00 [.92, 1.08]	1.00 [.97, 1.02]	.98 [.91, 1.04]	.99 [.97, 1.01]
	Between	1.03 [.98, 1.08]	1.06 [1.02, 1.09]	1.09 [1.04, 1.13]	1.04 [1.01, 1.07]

All models were estimated controlling for linear change in likelihood of substance use and substance use problems over time, age at baseline, sexual identity, gender identity, and race/ethnicity.

^acoefficient for hurdle portion of model (likelihood of using substance);

^bcoefficient for count portion of model (count of substance use problems). Significant associations are in bold.

Within-person concurrent associations.

Internalized stigma, microaggressions and general stress were associated with the concurrent likelihood of alcohol use, such that individuals who experienced more internalized stigma, microaggressions, or general stress than usual were more likely to use alcohol during the same wave. However, victimization was not concurrently associated with alcohol use. Only microaggressions were associated with concurrent cannabis use, with individuals who experienced more microaggressions than usual being more likely to use cannabis during the same wave. Victimization was concurrently associated with alcohol and cannabis use problems among users, such that when individuals experienced more victimization than usual, they experienced more concurrent alcohol and cannabis use problems than usual. Internalized stigma, microaggressions, and general stress were not associated with concurrent alcohol or cannabis use problems among users.

Within-person prospective associations.

No within-person prospective associations between stressors and SU or related problems were significant.

Between-person associations.

Individuals who experienced more internalized stigma and microaggressions across study waves were more likely to have used alcohol during the study. Results also demonstrated that drinkers who experienced more internalized stigma, microaggressions, victimization, or general stress also experienced more alcohol use problems on average. Microaggressions, victimization, and general stress were associated with cannabis use and problems. Specifically, individuals who experienced more of these three stressors were more likely to have used cannabis during the study and cannabis users who experienced more of these stressors also experienced more cannabis use problems.

Minority Stress and Outcomes: Controlling for General Stress (Tables 6–7)

Table 6.

Concurrent and prospective within-person associations between minority stress and internalizing symptoms controlling for general stress

		Depression		Anxiety
Model	Predictor	Concurrent	Prospective	Concurrent	Prospective
Model 1	Internalized stigma	.04 [.03, .06]	.15 [.06, .24]	.04 [.02, .05]	.16 [.06, .26]
	General Stress	.28 [.24, .32]	.03 [−.06, .13]	.21 [.18, .26]	.05 [−.03, .13]
Model 2	Microaggressions	.07 [.05, .09]	.14 [.04, .25]	.06 [.04, .08]	.16 [.06, .26]
	General Stress	.29 [.25, .33]	.05 [−.05, .15]	.22 [.19, .27]	.06 [−.03, .14]
Model 3	Victimization	.03 [.02, .05]	.12 [−.04, .28]	.03 [.02, .04]	.19 [.01, .36]
	General Stress	.27 [.24, .32]	.02 [−.07, .11]	.21 [.18, .25]	.04 [−.05, .11]

All models were estimated controlling for linear change in internalizing symptoms over time, age at baseline, sexual identity, gender identity, race/ethnicity, and (for prospective associations) outcome at the previous wave. 95% confidence intervals are presented. Significant associations are bold.

Table 7.

Concurrent and prospective within-person associations between minority stress and substance use/problems controlling for general stress

		AUDIT		CUDIT
Model	Predictor	Any Alcohol Use (OR)a	Alcohol Use Problems (RR)b	Any Marijuana Use (OR)a	Marijuana Use Problems (RR)b
Model 1: Concurrent	Internalized stigma	1.21 [1.11, 1.32]	1.01 [.86, 1.18]	1.37 [.90, 2.10]	.97 [.85, 1.10]
	General Stress	1.01 [.99, 1.02]	.99 [.98, 1.01]	1.02 [.99, 1.03]	1.01 [.99, 1.02]
Model 1: Prospective	Internalized stigma	.78 [.56, 1.08]	1.00 [.84, 1.21]	.95 [.53, 1.70]	.95 [.83, 1.10]
	General Stress	.99 [.90, 1.09]	1.00 [.98, 1.02]	1.01 [.98, 1.04]	.99 [.98, 1.01]
Model 2: Concurrent	Microaggressions	1.91 [1.64, 2.23]	1.02 [.88, 1.18]	1.16 [.93, 1.45]	1.01 [.95, 1.09]
	General Stress	1.01 [.96, 1.07]	1.00 [.98, 1.02]	1.01 [1.00, 1.02]	1.00 [.99, 1.02]
Model 2: Prospective	Microaggressions	1.05 [.48, 2.29]	.81 [.64, 1.04]	1.67 [.97, 2.94]	.93 [.79, 1.11]
	General Stress	1.00 [.96, 1.03]	.99 [.97, 1.02]	1.00 [.93, 1.08]	.99 [.97, 1.01]
Model 3: Concurrent	Victimization	1.32 [1.12, 1.45]	1.27 [1.13, 1.43]	1.03 [.74, 1.42]	1.18 [1.05, 1.33]
	General Stress	1.04 [1.02, 1.07]	1.00 [.98, 1.02]	1.02 [.95, 1.09]	1.01 [.99, 1.03]
Model 3: Prospective	Victimization	1.14 [.43, 3.00]	.77 [.53, 1.14]	1.29 [.51, 3.29]	.83 [.68, 1.01]
	General Stress	.99 [.91, 1.08]	.99 [.97, 1.02]	1.00 [.91, 1.09]	.99 [.97, 1.01]

All models were estimated controlling for linear change in substance use/problems over time, age at baseline, sexual identity, gender identity, race/ethnicity, and (for prospective associations) outcome at the previous wave. 95% confidence intervals are presented. Significant associations are in bold.

Next, we tested whether concurrent and prospective within-person associations between minority stressors and internalizing symptoms and SU/problems remained significant when we controlled for general stress. All concurrent and prospective within-person associations between minority stressors and internalizing symptoms and SU/problems remained significant when general stress was controlled, with two exceptions. Specifically, experiencing more SGM victimization than usual at one wave was no longer significantly prospectively associated with depression when general stress was controlled. Experiencing more microaggressions than usual at one wave was also no longer associated with a higher concurrent risk of cannabis use when general stress was controlled. Additionally, one association that was not significant in previous models became significant when general stress was controlled. Experiencing more SGM victimization than usual was concurrently associated with higher likelihood of alcohol use.

Discussion

The current study adds to a small but growing body of methodologically strong, robust evidence that enacted and internalized stigma may increase anxiety and depression and that certain minority stress experiences may increase some forms of problematic SU among SGM-AFAB. Specifically, results indicated that experiencing more enacted and internalized stigma than usual was concurrently and prospectively associated with more anxiety and depression. Associations between minority stressors and SU variables were less consistent. Concurrently, internalized stigma and microaggressions were associated with higher likelihood of alcohol use; microaggressions were associated with higher likelihood of cannabis use; and victimization was associated with more alcohol and cannabis use problems among users.

Within-Person Associations: Internalizing Symptoms

Our results indicated that, within persons, both enacted and felt stigma were concurrently and prospectively associated with anxiety and depression. This indicates that when SGM-AFAB experienced more minority stressors than usual, they also experienced more internalizing symptoms than usual and an increase in internalizing symptoms 6 months later. These findings add to a small but growing body of evidence from longitudinal studies that minority stressors precede increases in anxiety and depression among SGM and likely play a causal role in the mental health disparities affecting SGM populations.

The concurrent associations between minority stressors and internalizing symptoms are consistent with the results of other longitudinal studies conducted with mixed-sex SGM samples and SGM-AMAB samples (Birkett et al., 2015; Pachankis et al., 2018). Our prospective analyses also replicated previously demonstrated associations between victimization and increases in anxiety and depression in mixed-sex SGM samples (Birkett et al., 2015; Burton et al., 2013; Tucker et al., 2016) but also stands in contrast to the lack of significant prospective associations between enacted or felt stigma and internalizing symptoms among SGM-AMAB found by Pachankis and colleagues (2018). It is possible that the lag used by Pachankis and colleagues (1 year) may have been too long to detect a significant prospective effect compared to the shorter lags used by our study and other studies with mixed-sex SGM and SGM-AMAB samples (e.g., 6 months; Birkett et al., 2015; Burton et al., 2013; Rendina et al., 2017). Our findings also extended this line of research by demonstrating that internalized stigma and microaggressions predict subsequent increases in anxiety and depression. This finding makes an important contribution to the recent debate around the utility and impact of microaggressions by addressing a noted limitation of this literature – the limited number of longitudinal studies (Lilienfeld, 2017). To our knowledge, this is one of the first studies to demonstrate a prospective association between sexual minority based microaggressions and subsequent increases in internalizing symptoms among SGM. These findings add to a growing body of literature documenting prospective associations between experiencing microaggressions and subsequent increases in internalizing symptoms (e.g., Ong, Burrow, Fuller-Rowell, Ja, & Sue, 2013), which provide evidence of the directionality of these associations. This study is also one of the first to demonstrate a prospective association between internalized stigma and internalizing symptoms (alongside Rendina et al., 2017). Additionally, as previous longitudinal studies of these associations have either focused on SGM-AMAB or treated all SGM as a single group, the current study may be one of the first to demonstrate that minority stressors concurrently and prospectively predicts internalizing symptoms among SGM-AFAB.

Within-Person Associations: Substance Use

Overall, we found that certain experiences of felt and enacted stigma were associated with concurrent use of alcohol and enacted stigma was associated with concurrent cannabis use and problems related to cannabis and alcohol use among users. These findings support the theory that elevated rates of SU and problems among SGM individuals may be due to chronic stress experienced as a result of stigma (Meyer, 2003). Specifically, minority stressors may deplete SGM’s psychological resources for coping with stress, increasing emotion dysregulation, which may increase the use of substances to cope (Hatzenbuehler, 2009; Hatzenbuehler, Corbin, & Fromme, 2011). However, only about half of the concurrent associations between minority stressors and SU outcomes were significant and no prospective associations were significant. This pattern of findings suggests that minority stressors may be associated with SU and related problems among SGM-AFAB, but not as reliably as they are with internalizing symptoms.

Interestingly, the within-person associations between stigma and SU were not consistent across types of stigma or across SU outcomes. SGM victimization was associated with SU problems among users, but not likelihood of SU. This parallels previous evidence that experiences of minority stressors were associated with more alcohol use problems but not with more alcohol use among SGM-AFAB (Wilson et al., 2016), and that microaggressions were associated with more alcohol and cannabis use problems among users but not with likelihood of SU among SGM-AMAB (Dyar et al., 2019). In the current study, however, individuals were more likely to use alcohol and cannabis when they experienced more microaggressions than usual and more likely to use alcohol when they experienced more internalized stigma than usual. Given that most studies conflate SU and problems, further research is needed to explore how stigma may have different effects on the likelihood of SU versus problems arising from SU.

This study provided no evidence of prospective associations between minority stressors and SU/problems. Overall, existing literature provides consistent evidence of concurrent associations between minority stressors and SU, but there is limited evidence of long-term effects. For example, Dermody and colleagues (2016) found that experiencing more SGM victimization was associated with a subsequent increase in the likelihood of binge drinking among SGM-AFAB, while Dyar and colleagues (2019) did not find a prospective association between minority stressors and SU or problems 6 months later in a sample of SGM-AMAB. However, at least one experience sampling study found that SGM individuals were more likely to use substances several hours after an experience of discrimination (Livingston, Flentje, Heck, Szalda-Petree, & Cochran, 2017). This suggests that acute experiences of minority stressors may increase SU and related problems in the short term, but that some individuals may return to their baseline levels of SU after the acute stress response ends. For example, an individual who experiences discrimination one day would be expected to increase their SU when stress resulting from that event is high – soon after the event. It is likely that this increase in SU or problems may persist over time for some, while others may return to their usual level of SU after a period of time. This pattern will lead to a weakening of the average within-person effect and larger variation in the effect across individuals as temporal distance from the event increases, consistent with findings from the current study. Therefore, studies with short lags between assessments, like daily diary or experience sampling studies, may be more effective for detecting prospective associations between minority stressors and SU. Another potential reason for the limited number of prospective associations between minority stressors and SU is that there are a wide variety of reasons that people use substances, including for social and enhancement reasons (Grant, Stewart, O’Connor, Blackwell, & Conrod, 2007; C. M. Lee, Neighbors, & Woods, 2007). Although individuals may use substances at times of high stress (e.g., after experiencing minority stress), they also do so at other times (e.g., during celebrations and social gatherings with friends), attenuating associations between minority stressors and SU.

Between-Person Associations

Overall, we found that individuals who tended to experience more enacted or internalized stigma were more likely to use substances, have more SU problems, and have more internalizing symptoms on average (i.e., across study visits). This pattern of findings is consistent with the cross-sectional literature, which has linked enacted stigma with internalizing symptoms (Hall, 2018; Puckett, Levitt, Horne, & Hayes-Skelton, 2015), alcohol use (Lowry et al., 2017; Marshal et al., 2013), cannabis use (Lowry et al., 2017), alcohol use problems (Livingston, Christianson, & Cochran, 2016), and cannabis use problems (Feinstein & Newcomb, 2016; J. H. Lee, Gamarel, Bryant, Zaller, & Operario, 2016). Between-person associations with internalized stigma were less consistent, with internalized stigma being associated with internalizing symptoms, a higher likelihood of alcohol use, and more alcohol use problems but not cannabis use or problems. As findings linking internalized stigma with SU have been mixed (Feinstein & Newcomb, 2016; Goldbach et al., 2015; Lea et al., 2014), the current findings may help to provide added clarity.

General Stress, Internalizing Symptoms, and SU

When individuals experienced more general stress than usual, they experienced more concurrent anxiety and depression and had an increased likelihood of concurrent alcohol use. However, general stress was not concurrently associated with cannabis use or SU problems and was not prospectively associated with any SU outcomes. This somewhat sparse pattern of associations with SU outcomes is consistent with that seen for minority stressors. This suggests that stress as a whole, and not just minority stress, may not be a robust predictor of SU/problems in this sample. These findings differ somewhat from those in Dyar et al. (2019), which found that when SGM-AMAB reported more general stress than usual, they experienced more alcohol and cannabis use problems and were more likely to use cannabis.

General stress did not prospectively predict internalizing symptoms, in contrast to minority stress. The lack of prospective associations between general stress and internalizing symptoms 6 months later may be related to the high variability in general stress, with 56% of its variance occurring within-individuals. This high variability over time suggests that elevations in general stress are short-lived. When combined with consistent evidence of concurrent associations with internalizing symptoms and lack of prospective associations in this study, this high variability suggests that the effects of general stress on internalizing symptoms may be acute and limited to the period when general stress is elevated. Of note, the measure we used to assess general stress assessed perceived stress, or the degree to which situations in one’s life are perceived as stressful, uncontrollable, and unpredictable (Roberti et al., 2006). The measure of enacted stigma assessed the frequency of specific types of experiences, rather than the appraisal or perception of these experiences as stressful. The higher level of variability in general (perceived) stress over time compared to minority stressors may be due to these differences in measurement. Research using more parallel measures of general and minority stress is needed to clarify whether the different pattern of findings is due to differences in the source of stress (general vs. minority) or the type of stress (perception of events vs. frequency of events).

We also found that the associations between minority stressors and internalizing symptoms and SU/problems remained significant even when general stress was controlled. This indicates that minority stressors generally predicted internalizing symptoms and SU/problems above and beyond the effect of perceived stress from any source (including minority stressors). Together with similar findings in other samples (Dyar et al., 2019)(Mereish & Miranda Jr, 2019), these results make an important contribution to the literature on Meyer’s (2003) minority stress framework, by providing support for a central assumption that has rarely been empirically tested: minority stressors have a unique effect on health that is additive to the effect general stressors.

Clinical Implications

These findings have important clinical implications as they provide robust evidence that minority stressors contribute to internalizing symptoms among SGM-AFAB, with effects still detectable up to six months later. Findings also suggest that minority stressors may play a role in SU and related problems but that these effects appear to be more temporally limited – co-occurring with minority stressors but not predicting increased SU or problems six-months later.

Overall, findings highlight the need for interventions that reduce the impact of minority stressors on health for SGM-AFAB. While there is a lack of interventions designed specifically for SGM-AFAB, several strategies have been developed to reduce sexual minority stigma, teach skills to cope with minority stress, and reduce the impact of stigma on health (Chaudoir, Wang, & Pachankis, 2017). As minority stressors predict both mental health and SU problems, SGM-AFAB may benefit from an intervention similar to ESTEEM, a transdiagnostic cognitive-behavioral intervention that aims to reduce mental health and SU problems among sexual minority men. ESTEEM targets minority stressors and universal risk factors for mental health problems and has demonstrated its efficacy in reducing internalizing symptoms and alcohol use problems in a randomized controlled trial (Pachankis, Hatzenbuehler, Rendina, Safren, & Parsons, 2015). Although this intervention was developed for sexual minority men, its transdiagnostic approach and focus on minority stress mean that it is likely to be effective among SGM-AFAB. As some of the results of this study differ from previous research with SGM-AMAB samples, it will be important for future research to test the effectiveness of interventions like ESTEEM with sexual minority cisgender women and gender minorities and make adaptations to existing interventions to meet the unique needs of these populations. Finally, the current results replicate previous longitudinal studies showing that psychological distress is highest for LGBT youth during adolescence and decreases as they age into young adulthood (Birkett et al., 2015). As such, they suggest a critical need for interventions targeting minority stress, internalizing symptoms, and SU in SGM-AFAB adolescents versus adults.

Limitations

The current findings should be considered in light of the study’s limitations. First, the current sample included only SGM-AFAB and therefore did not allow us to examine whether these associations differed significantly by sex assigned at birth. Second, the six-month lag between visits may have been too long to observe some prospective effects of minority stressors on SU. Future research with shorter lags between waves may be able to detect effects of minority stressors that do not persist for such a long period. Third, our non-probability sample limits the generalizability of our findings. Many participants were recruited via SGM community events and social media links to SGM-relevant pages. Therefore, this sample may be more out and more connected to the SGM community than the average SGM-AFAB individual. Additionally, all participants were recruited from Chicago and their experiences may not reflect SGM-AFAB individuals’ experiences in other regions of the US or in other countries. Additionally, cannabis use was decriminalized in Chicago throughout data collection, which may have affected results. Given that two of the three minority stressors variables examined in the current study were focused on sexual minority stressors and did not include gender minority stress, it will be important for future research to replicate these results focusing on the effects of gender minority stress. The current manuscript also did not examine the unique minority stress experiences of subgroups of sexual minorities (e.g., bisexual and pansexual individuals). Future research should continue to examine the effects of shared and unique minority stress experiences on health. We did not examine perceived stress arising from minority stressors and our measure of general stress assessed a slightly different aspect of stress (perceived stress) than did our measures of minority stressors, which assessed the frequency of objectively stressful events. Thus it will be important for future research using more parallel measures of general and minority stress to determine whether perceived distress arising from minority stress predicts health outcomes over and above perceived distress from general stressors. While we conducted some initial examination of the psychometric properties of the measures of microaggressions and internalized stigma used in the current study, they have not been fully validated in an SGM-AFAB sample. Additionally, the measure of perceived stress used in the current study has not, to our knowledge, been formally validated in an SGM sample, although it has been validated in a wide range of other samples. These are important area for future psychometric research.

Conclusion

Despite its limitations, the current study extends the limited longitudinal research on minority stress, internalizing symptoms, and SU. Findings provide robust evidence that minority stressors are consistently associated with elevations in anxiety and depression, with effects remaining detectable a full six months later among a sample of SGM-AFAB. Minority stressors were also associated with SU and SU problems, but there was no evidence of prospective effects on SU or problems. Future research is needed to explore what factors may amplify or buffer the effects of minority stressors on SU/problems and to examine whether the same pattern of findings is present in longitudinal studies with different lags between waves. Overall, our findings indicate that minority stressors contribute to internalizing symptoms and alcohol use among SGM-AFAB, suggesting that interventions which target minority stress are likely to be effective in reducing disparities in internalizing and SU disorders affecting SGM-AFAB. While interventions aiming to reduce SGM health disparities by helping SGM cope with minority stress are a necessary tool for addressing health disparities, anti-discrimination policies and interventions that aim to reduce the stigmatization of SGM at the societal level are also critical as effective population level interventions have the potential to eliminate health disparities affecting SGM by removing the underlying cause.

Public Health Significance Statement.

This study strongly suggests that minority stressors contribute to the high levels of anxiety and depression experienced by sexual and gender minority individuals assigned female at birth. It also provides some evidence that minority stressors may also contribute to high rates of alcohol and cannabis use problems in this population. Study findings highlight the need for individual level clinical interventions that aim to reduce the impact of minority stress on health as well as societal interventions that aim to eliminate the anti-SGM stigma that causes minority stress.