Sexual minority stress and substance use: An investigation of when and under what circumstances minority stress predicts alcohol and cannabis use at the event-level

Abstract

Objective:

Sexual minority women and gender diverse (SMWGD) individuals are at elevated risk for alcohol and cannabis use disorders compared to cisgender, heterosexual women. This has been attributed to the unique stressors that SMWGD experience (i.e., sexual minority stress); however, recent studies have found mixed evidence for a link between sexual minority stress and substance use. The current manuscript introduces and tests a novel theoretical model derived from integrating minority stress theory and the multistage model of drug addiction to explain these mixed findings.

Method:

We used data from a 30-day ecological momentary assessment study of substance use among SMWGD to determine whether event-level associations between enacted stigma (bias from others) and same-/next-day alcohol and cannabis use are dependent on an individuals’ typical pattern of substance use (e.g., frequency, quantity, motives, and substance use disorder symptoms).

Results:

Findings indicate that enacted stigma predicted increased likelihood of alcohol and cannabis use among those who used frequently and those who had a probable alcohol or cannabis use disorder and predicted decreased likelihood of use among those who used less frequently. Enacted stigma also predicted cannabis (but not alcohol) use among those who reported high coping motives for use.

Conclusions:

Findings provide initial evidence in support of an integrated model of minority stress theory and the multistage model of drug addiction. Findings suggest that alcohol and cannabis use disorder interventions for SMWGD would benefit from addressing sexual minority stress and coping skill-building.

General Scientific Summary

The results of this study suggest that discrimination and microaggressions are risk factors for same/next-day alcohol and cannabis use among LGBTQ+ individuals who use alcohol and/or cannabis frequently or have a substance use disorder. However, discrimination and microaggressions do not appear to be risk factors for same/next-day alcohol and cannabis use among LGBTQ+ individuals who use alcohol or cannabis less frequently or do not have a substance use disorder.

Introduction

Sexual minority (i.e., individuals who identify as lesbian, gay, bisexual or with another non-heterosexual identity) and gender diverse individuals (i.e., individuals’ whose gender identity does not match their sex assigned at birth) are at elevated risk for substance use disorders (Kerridge et al., 2017; Krueger et al., 2020), and these disparities are particularly pronounced for those assigned female at birth (Kerridge et al., 2017; Krueger et al., 2020; Watson et al., 2020). These disparities have been attributed to sexual minority stress – the chronic stress that sexual minority individuals experience as a result of the stigmatization of non-heterosexuality (Hatzenbuehler, 2009; Meyer, 2003). However, recent longitudinal studies examining the theorized association between sexual minority stress and substance use have provided mixed evidence, particularly for prospective effects (Dyar, Dworkin, et al., 2021; Dyar et al., 2019; Dyar et al., 2020; Ehlke et al., 2021; Lewis et al., 2021; Wilson et al., 2016). The current manuscript aims to propose and test an integrated theoretical model that may help to explain these mixed findings. This model incorporates sexual minority stress theory and the multistage model of drug addiction, which posits that different risk factors are relevant at different levels of substance use, with stress and reducing negative affect becoming risk factors for substance use only when use is frequent and a substance use disorder is present (Koob, 2013; Koob & Volkow, 2010). Drawing from this integrated theoretical framework, we test whether the event-level association between sexual minority stress and substance use is moderated by characteristics of an individual’s typical pattern of substance use (e.g., average frequency of use, substance use disorder symptoms, motives for use) among a sample of sexual minority women and gender diverse individuals assigned female at birth (SMWGD).

Sexual Minority Stress and Substance Use

Minority stress theory posits that sexual minorities are at elevated risk for substance use and related disorders as a result of sexual minority stress (Hatzenbuehler, 2009; Meyer, 2003). Given that sexual minority stress is experienced in addition to general stressors, it has been theorized that sexual minority stressors deplete coping resources and lead to a reliance on substances to cope with negative affect arising from sexual minority stress (Hatzenbuehler, 2009). Several recent semi-annual longitudinal and ecological momentary assessment (EMA) studies have tested the theorized association between sexual minority stress and substance use, providing mixed findings for a prospective association. For instance, two recent semi-annual longitudinal studies found evidence that experiencing more enacted stigma (i.e., biased treatment by others, such as discrimination, victimization, and microaggressions) than usual predicted higher concurrent alcohol and cannabis use consequences but did not prospectively predict changes in consequences (Dyar et al., 2019; Dyar et al., 2020). Another linked enacted stigma with subsequent changes in alcohol consequences but not in consumption (Wilson et al., 2016). Other semi-annual longitudinal studies have linked enacted stigma with concurrent elevations in alcohol use (Newcomb et al., 2012) and subsequent increases in binge drinking (Dermody et al., 2016).

At the daily level, EMA studies have also provided mixed results. Findings have generally supported at least one concurrent association between enacted stigma and an alcohol outcome (e.g., frequency, likelihood of use) in each study; however, the pattern of findings varies considerably across studies. Ehlke et al. (2021) found concurrent associations between enacted stigma and a higher likelihood of drinking, higher quantity consumed, and more consequences experienced. However, Lewis et al. (2021) found evidence that enacted stigma concurrently predicted an increased likelihood of drinking but not quantity or consequences. Dyar, Dworkin, et al. (2021) found yet another pattern with enacted stigma predicting concurrent consequences but not quantity. Prospective associations were less frequent with Dyar, Dworkin, et al. (2021) finding no evidence of prospective effects of enacted stigma on alcohol use and Livingston et al. (2017) demonstrating associations between enacted stigma and subsequent substance use. Lewis et al. (2021) found that enacted stigma prospectively predicted increases in drinking quantity but not likelihood of drinking or consequences. Given these mixed findings (particularly with regard to prospective associations), further research is needed to understand when and under what circumstances enacted stigma predicts substance use. Additionally, the predominate focus of existing research on alcohol use needs to be addressed, particularly given increases in the legalization of cannabis use in recent years.

An Integrated Theoretical Model

What might explain these mixed findings? The multistage model of drug addiction may provide a useful theoretical framework for understanding these mixed results. This theory posits that risk factors for substance use change as individuals move from infrequent to frequent use and develop substance use disorders (SUDs; Koob, 2013; Koob & Volkow, 2010). When use is infrequent and SUDs are absent, substance use is theorized to be motivated by the desire to increase positive affect and sociability (i.e., positive reinforcement). As a result, social motives (i.e., using substances to improve social gatherings) and enhancement motives (i.e., using substances to enhance positive affect) are theorized to be the most relevant at this stage. However, as use becomes more frequent and SUDs develop, substance use becomes motivated by the desire to reduce negative affect (i.e., negative reinforcement) and coping motives (i.e., use to reduce negative affect) become dominant drivers of substance use.

Numerous neuropsychological studies have supported aspects of the multistage model of drug addiction (Koob, 2013; Koob & Volkow, 2010; Kwako & Koob, 2017). Studies with animal models indicate that increases in substance use lead to dysregulation of neurological reward processes (Koob et al., 2004). Initially, the rewarding effects of substance use (e.g., increased dopamine; decreased stress neurochemicals, like corticotropin-releasing factor [CRF]) are theorized to drive positive reinforcement of substance use (Koob et al., 2004). However, this become attenuated as individuals increase their substance use and the brain attempts to return to homeostasis (Koob et al., 2004). However, in the absence of the substance, individuals who use frequently may experience withdrawal, which is characterized by negative affect and a desire to use the substance to reduce the negative affect (negative reinforcement) (Koob et al., 2004). Some of the same pathways that contribute to negative affective states during withdrawal are also activated in response to psychosocial stressors (Kwako & Koob, 2017; Lijffijt et al., 2014). This results in cross-sensitization of these pathways, with stress increasing the motivational value of substance use and reactivity to stress (Lijffijt et al., 2014). Further, noradrenergic systems are sensitized as substance use becomes more frequent, which has been linked to dysregulation of processes implicated in emotion regulation and problem-solving coping (Lijffijt et al., 2014). Together, the transition to negative reinforcement mechanisms contributing to substance use, the cross-sensitization of stress and substance use, and interference with other coping processes are theorized contribute to an increased likelihood of using substances to cope with experiences of enacted stigma among SGM who use substances frequently or have a substance use disorder (Koob, 2013; Koob et al., 2004; Kwako & Koob, 2017; Lijffijt et al., 2014).

In contrast to the numerous animal models and neuropsychological studies of the multistage model, very few studies have tested this theory outside of the laboratory. Two studies to test this theory in real-world contexts found that risk factors differed for individuals with and without alcohol use disorder (AUD). Specifically, risk factors for those with AUD included negative reinforcement motives for use and coping motives, while risk factors for those without AUD included positive reinforcement motives for use and social drinking contexts (Cho et al., 2019; Corbin et al., 2020). However, we are not aware of any studies to test this theory among sexual and gender minorities, despite its potential to explain mixed findings regarding the link between sexual minority stress and substance use.

Integrating minority stress theory and the multistage model of drug addiction provides the following hypotheses. First, sexual minority stress would only be expected to predict increases in substance use among individuals who use substances frequently or have an SUD. Given that the multistage model theorizes that negative reinforcement and coping motives for use are not relevant risk factors for individuals who use less frequently or do not have an SUD, the association between sexual minority stress and substance use would not be expected to be significant in these individuals. Importantly, this does not mean that sexual minority stress is not harmful to sexual minorities who use substances less frequently. Sexual minority stress has been consistently linked with negative affect and symptoms of anxiety and depression (e.g., Birkett et al., 2015; Dyar et al., 2020; Eldahan et al., 2016; Pachankis et al., 2018; Tucker et al., 2016) and these effects likely apply to all sexual minorities regardless of their level of substance use. Individuals who use substances less frequently may simply not be likely to use substances to cope with the negative affect arising from experiences of sexual minority stress.

Second, given that different motives for use are theorized to be relevant in the presence or absence of SUDs, the link between sexual minority stress and substance use would also be expected to be moderated by an individuals’ typical motives for use. Specifically, individuals who tend to use substances to cope (i.e., negative reinforcement) and not to increase sociability or enjoyment (i.e., social and enhancement motives; positive reinforcement) would experience increases in substance use following experiences of sexual minority stress. Individuals who tend to use substances to increase sociability and enjoyment but not for coping are unlikely to use substances in response to sexual minority stress (a negative reinforcement process), given that positive reinforcement is the major driver of their substance use. Conformity motives are generally conceptualized as operating via negative reinforcement (i.e., using to avoid peer rejection; Cooper et al., 2016), and therefore, the multistage model may suggest that conformity motives would influence substance use among those with a SUD. However, conformity motives are rarely endorsed, particularly among individuals who use substance regularly (Blevins & Stephens, 2016; Bonn-Miller et al., 2007; Lee et al., 2009; Lee et al., 2007), suggesting that conformity motives may not be a risk factor when cannabis use is frequent. As a result, it is unclear how whether and how conformity motives may moderate the association between sexual minority stress and substance use. The current study aims to test these hypotheses in a sample at high risk for problematic use – SMWGD.

Current Study

To fill gaps in existing research and explore a potential explanation for mixed findings, the current study aims to examine moderators of associations between enacted stigma and alcohol/cannabis use at the event level. We will examine an individuals’ average substance use frequency, quantity, motives, and AUD/CUD symptoms as moderators of event-level associations between enacted stigma and substance use. We expect the following patterns to emerge from these moderation analyses. First, experiencing enacted stigma during one assessment will predict a higher likelihood of use during the same-/next-assessment among individuals who report higher frequency and quantity of use as well as those who have a substance use disorder. Second, experiencing enacted stigma during one assessment will predict a higher likelihood of substance use during the same-/next-assessment among those with high coping motives and low social/enhancement motives.

Method

Participants and Procedures

The current analyses used data from a longitudinal study of substance use among SMWGD. Participants were recruited via paid advertisements on social media (e.g., Facebook, Instagram). Advertisements included images of same-gender couples, visibly LGBTQ+ individuals, and individuals with Pride colors and flags and asked members of the LGBTQ+ community to share their experiences by participating in the study. Ads were targeted to a range of interests relevant to the LGBTQ+ community (e.g., LGBTQ+ rights, media with LGBTQ+ characters, prominent LGBTQ+ individuals). Recruitment began in August 2020 and was complete in May 2021. Participants who appeared eligible based on their responses to the eligibility survey were text messaged by study team members to verify their eligibility and their access to a mobile phone with text message capabilities. To verify their eligibility, participants were asked to text demographic information (i.e., age, state of residence, email address), which was cross-checked with their responses in the eligibility survey. Participants who passed the eligibility verification were invited to participate and sent a link to the baseline assessment. Stratified sampling was used to ensure that the sample was diverse in race/ethnicity, gender identity, and sexual identity, by setting recruitment goals for each of these characteristics and capping enrollment of specific groups when the recruitment goal for that group was met.

The study included a baseline assessment (day 0), a 30-day ecological momentary assessment study (days 1–30), and a follow-up assessment (completed within two weeks of day 30). This study uses data from the 30-day ecological momentary assessment study and the baseline assessment. During the ecological momentary assessment period, participants were invited (via text message or email) to complete one survey in the morning (at 8:00am in their time zone) and one in the evening (at 6:00pm in their time zone). Participants had from 8:00am-1:00pm to complete the morning survey and between 6:00pm-12:00am to complete the evening survey. Participants who had not completed the survey by three hours after the survey invitation were sent a reminder. Those who missed more than three surveys in a row were contacted by study staff to check in and re-engage participants. Surveys took approximately 2 minutes to complete. The study received IRB approval at Northwestern University.

Eligible participants were 1) U.S. residents, 2) 18–25 years old, 3) identified as lesbian, bisexual, pansexual, or queer, 4) were assigned female at birth, and 5) met alcohol or cannabis use criteria (i.e., reported having 4 or more drinks at least twice and/or using cannabis on at least three days in the past month). Participants were paid up to $150 based on completion rates: $20 for baseline, $20 for follow-up, $1 for each ecological momentary assessment survey, and $5 bonus for each 6 surveys completed in a row.

There was a total of 429 participants. See Table 1 for demographics. The sample was comprised predominately of people of color, with 33.6% of the analytic sample identifying exclusively as non-Latinx White. There were a sizeable number of gender minority participants (26.8%).

Table 1.

Demographics of Analytic Sample at Baseline (N = 429)

Demographic Variable	n	%
Sexual Identity
Lesbian	112	26.1%
Bisexual	111	25.9%
Pansexual	112	26.1%
Queer	94	21.9%
Race/Ethnicitya
White	235	54.8%
Black	102	23.8%
Latinx	129	30.1%
Asian	53	12.4%
Other Race/Ethnicity	34	7.9%
Gender Identity
Cisgender Women	314	73.2%
Gender Minority	115	26.8%
Substance Use Criteria Met
Alcohol Only	110	25.6%
Cannabis Only	107	24.9%
Alcohol and Cannabis	212	49.4%
Age (M, SD)	22.27 (2.01)

^aPercentages add up to more than 100% because participants could select multiple racial/ethnic identities.

Measures

Daily Diary Measures

Enacted Stigma was assessed during morning and evening surveys by asking participants two questions. First, participants were asked an item adapted from Mohr and Sarno (2016): “Did you experience anything stressful or negative related to your sexual orientation since the last survey? This could be something that was relatively minor (e.g., feeling that your sexual identity was not respected) or major (e.g., being physically attacked because of your sexual orientation).” Participants were asked to indicate yes or no. Regardless of the response to this item, participants were asked to “indicate which of the following events you have experienced since the last survey because of your sexual orientation” and provided with a list of 10 common experiences of enacted stigma that have been utilized in previous daily diary studies (e.g., “someone acted uncomfortable around me”; Dyar & London, 2018; Flanders, 2015). Utilizing both measures allowed us to capture a wider range of potential experiences of enacted stigma than either measure alone. Given that few participants endorsed multiple types of enacted stigma on the same day, we created a binary variable from these measures. Surveys during which a participant endorsed the first binary item and/or any of the 10 experiences on the checklist received a score of 1. Surveys during which a participant indicated “no” in response to the first item and “none of the above” in response to the checklist received a score of 0.

Substance Use was assessed by asking participants, “which of the following have you used since the last survey?” Participants could select alcohol and/or marijuana or “none of the above.” Two binary variables were created indicating no use (0) or use (1) of each substance.

Baseline Measures

Drinking Moderators

Drinking frequency and quantity were assessed using the first two items of the Alcohol Use Disorders Identification Test (Saunders et al., 1993). Frequency was assessed with the item “How often do you have a drink containing alcohol?” on a scale of 0 (never) to 4 (4 or more times a week). Quantity was assessed via the item “How many drinks containing alcohol do you have on a typical day when you are drinking?” on a scale of 0 (1 or 2) to 4 (10 or more).

Drinking motives were assessed using an abbreviated version of the Drinking Motives Measure (Cooper, 1994; Grant et al., 2007). The highest loading items were selected for each subscale and participants were asked to indicate how often they drank for the following reasons: coping (6 items; α = .82; “to forget my worries”), enhancement (2 items; α = .60; “because I like the feeling”), social (2 items; α = .88; “because it improves parties and celebrations”), and conformity (2 items; α = .78; “to fit in with a group I like”). Responses were provided on a scale of 1 (almost never/never) to 5 (almost always/always) and were averaged.

Alcohol use disorder symptoms were assessed using the Alcohol Use Disorder Identification Test (AUDIT; Saunders et al., 1993). The AUDIT includes 10 items rated on different scales. For example, the item “How often do you have a drink containing alcohol?” was rated from 0 (never) to 4 (4 or more times a week). Items also assess problems arising from alcohol use (e.g., “How often during the past 6 months have you found that you failed to do what was normally expected of you because of drinking?”). Responses were summed. Total scores ranged from 0 to 40 (a = .81), with scores of 8–14 indicating hazardous use and 15+ indicating the probable presence of DSM-IV alcohol dependence (Saunders et al., 1993).

Cannabis Use Moderators

Cannabis use frequency was assessed by the item “In the past month, how many days did you use marijuana?” Participants could enter an integer value between 0 and 31.

Cannabis use quantity was assessed using one item from the Cannabis Use Inventory (Cuttler & Spradlin, 2017), “How much marijuana have you typically used per week in the past 30 days?” Response options ranged from 1 (< 1/8 of an ounce) to 7 (more than 1 ounce). This item was only administered to participants who indicated using marijuana bud or flower (n = 285).

Cannabis use motives were assessed using a brief version of the Comprehensive Marijuana Motives Questionnaire (Lee et al., 2009). The two highest loading items were selected for each subscale and participants were asked to indicate how often they used marijuana for the following reasons: coping (α = .84; “to forget your problems”), enhancement (α = .81; “because it is fun”), social/celebration (α = .78; “to celebrate”), and conformity (α = .69; “because you didn’t want to be the only one not doing it”). Responses were provided on a scale of 1 (almost never/never) to 5 (almost always/always) and were averaged.

Cannabis use disorder symptoms were assessed using the Cannabis Use Disorder Identification Test Revised (CUDIT-R; Adamson et al., 2010). The CUDIT-R includes 8 items rated on different scales. For example, the item “How often during the past 6 months did you fail to do what was normally expected from you because of using marijuana?” was rated from 0 (never) to 4 (daily or almost daily). Responses were summed. Total scores ranged from 0 to 32 (a = .83), with scores of greater than 10+ indicating a DSM-5 cannabis use disorder (Bonn-Miller et al., 2016) and 13+ indicating probable DSM-IV cannabis dependence (Adamson et al., 2010).

Transparency and Openness Statement

Given the risk of deductive disclose when data from minoritized populations are made publicly available, deidentified data are only available from the study team after the development of a data sharing agreement. Data analysis code is available from the authors upon request. This study was not preregistered.

Analytic Plan

Analyses were conducted in Mplus version 8.6. There was a total of 19,186 completed ecological momentary assessment surveys from 429 participants. The median completion rate was 88.3% (M = 74%, SD = 28%). Individual completion rates ranged from 2% to 100%. All participants completed the baseline survey. Within completed surveys, less than 1% of data were missing. Missing data were handled using Bayesian methods, which produce results similar to full information maximum likelihood (Asparouhov & Muthén, 2010). Only participants who met inclusion criteria for alcohol were included in analyses of alcohol use (14,579 observations from 322 participants), and only those who met criteria for cannabis use were included in analyses of cannabis use (13,772 observations from 319 participants).

Bayesian multilevel structural equation modeling (MSEM) with diffuse (non-informative) priors was used. MSEM utilizes latent variables, rather than group- and grand-mean centering, to separate within- from between-person variance (Lüdtke et al., 2008). By removing the between-person variance from the within-person variance, the within-person variables indicate the extent to which an individual was experiencing more/less of a construct than usual (above/below their person mean) during a particular assessment (e.g., experiencing more/less anxiety than usual). 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 probit link was used for binary outcomes (e.g., cannabis use). Probit regression coefficients represent the variance shared by the predictor and the latent continuous response variables underlying each binary observed item (Agresti, 2003).

First, we examined the concurrent and prospective unmoderated associations between enacted stigma and the likelihood of alcohol or cannabis use. In these models, within and between-person components of enacted stigma predicted within and between-person components of the likelihood of alcohol or cannabis use. For concurrent models, enacted stigma and likelihood of substance use were assessed at the same timepoint, and in prospective models, enacted stigma was assessed at time t-1, while substance use was assessed at t. In all models, we controlled for day of assessment, whether the assessment was on a weekend or weekday, and whether it was a morning or evening assessment at the within-person level. Further, we included the first-order autocorrelations for the outcome in each model (i.e., the correlation between the variable at day t-1 with the variable at day t), which effectively controls for the prior timepoint of the outcome. Within-person associations among minority stress and substance use variables and autocorrelations were allowed to vary across individuals. Age, sexual identity, gender identity, and race/ethnicity were included as covariates at the between-person level as these demographic variables have been linked to both cannabis use and sexual minority stress (e.g., Dyar, Feinstein, Crosby, et al., 2021; Dyar, Feinstein, Sarno, et al., 2021; Swann et al., 2016).

Next, we examined between-person moderators (baseline quantity and frequency of substance use, motives for use, and AUDIT/CUDIT) of the within-person associations between enacted stigma and substance use. In these models, the between-person moderator predicted the random slope for within-person association (cross-level interaction). For significant interactions, simple slopes were calculated and predicted probabilities of substance use graphed at different levels of enacted stigma and moderators. The Johnson-Neyman technique was also used to determine at which precise values of the moderator the association became significantly positive or negative for significant interactions.

We estimated power to detect effects in the proposed models using Monte Carlo simulation studies. Simulated data matched the proportion of missing data and the prevalence and distributions of enacted stigma, alcohol, and cannabis use present in the current dataset. Results indicated that we were adequately powered (power > .80) to detect within-person effects between enacted stigma and alcohol or cannabis use that were small to moderate in size (probit coefficient = .15) and cross-level moderation effects of r = .20.

Results

Participants reported experiencing enacted stigma on 8% of days, drinking on 20% of days, and using cannabis on 28% of days. Table 2 provides means, standard deviations, and intraclass correlations. At baseline, participants drank between 2–3 times a week and 2–3 times a month, consumed an average of 3 or 4 drinks per occasion, and had an average AUDIT score of 8.42 (above the threshold for hazardous drinking). Participants also used cannabis an average of 12 days in the past month, used between an eighth and a quarter of an ounce of cannabis each week, and had a CUDIT-R score of 10.07 (at the proposed threshold for probable DSM-5 CUD).

Table 2.

Correlations, Means, Variances, and Intraclass Correlations

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
1. Enacted Stigma	-	.03*	.01	-	-	-	-	-	-	-	-	-	-	-	-	-	-
2. Alcohol Use	.01	-	.11*	-	-	-	-	-	-	-	-	-	-	-	-	-	-
3. Cannabis Use	.03	.07	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
4. Baseline Drinking Frequency	−.05	.56*	−.12*	-	-	-	-	-	-	-	-	-	-	-	-	-	-
5. Baseline Drinking Quantity	.02	−.05	.003	−.09	-	-	-	-	-	-	-	-	-	-	-	-	-
6. Alcohol Coping Motives	.12*	.18*	−.11*	.29*	.15*	-	-	-	-	-	-	-	-	-	-	-	-
7. Alcohol Social Motives	−.01	−.02	−.11*	.03	.18*	.16*	-	-	-	-	-	-	-	-	-	-	-
8. Alcohol Enhancement Motives	−.01	.08	−.03	.20*	.13*	.47*	.26*	-	-	-	-	-	-	-	-	-	-
9. Alcohol Conformity Motives	.08	−.11*	−.03	−.05	.09	.23*	.36*	.08	-	-	-	-	-	-	-	-	-
10. Baseline AUDIT	.07	.38*	−.01	.55*	.24*	.20*	.13*	.31*	.09	-	-	-	-	-	-	-	-
11. Baseline Cannabis Use Frequency	.01	.02	.80*	−.06	.01	−.10*	−.04	−.01	−.01	-.03	-	-	-	-	-	-	-
12. Baseline Cannabis Quantity	.10	−.02	.17*	−.02	−.03	.08	−.11	.13*	−.05	.01	.19*	-	-	-	-	-	-
13. Cannabis Coping Motives	.23*	−.06	.26*	−.09	.16*	.48*	.04	.17*	.18*	.14*	.30*	.21*	-	-	-	-	-
14. Cannabis Social Motives	.02	−.02	.12*	−.15*	.03	.04	.18*	.12*	.14*	.02	.15*	.13*	.12*	-	-	-	-
15. Cannabis Enhancement Motives	.01	−.02	.22*	−.01	.04	−.01	.13*	.24*	.02	.03	.28*	.03	.13*	.36*	-	-	-
16. Cannabis Conformity Motives	.03	−.01	−.18*	−.03	.06	.17*	.10	.05	.43*	.10	−.19*	.00	.04	.18*	−.02	-	-
17. Baseline CUDIT	.07	−.07	.65*	−.09	−.10*	−.06	.05	.07	.13*	.07	.71*	.23*	.49*	.22*	.28*	−.03	-
Mean	.08	.20	.28	2.44	.93	2.30	3.24	2.77	1.49	8.42	12.03	2.27	2.34	2.24	3.86	1.23	10.07
Standard Deviation	.27	.40	.45	.92	.84	.93	1.27	1.11	.83	5.33	10.80	1.56	1.24	1.07	1.11	.57	7.18
Range	0–1	0–1	0–1	0–4	0–4	1–5	1–5	1–5	1–5	0–40	0–31	1–7	1–5	1–5	1–5	1–5	0–32
Intraclass Correlation	.14	.13	.38	-	-	-	-	-	-	-	-	-	-	-	-	-	-

Unmoderated Models

At the within-person level, none of the concurrent or prospective associations between experiencing enacted stigma and likelihood of alcohol or cannabis use were significant, contrary to expectations (Table 3). This indicates that, on average, participants were no more or less likely to drink or use cannabis on days when they experienced enacted stigma than on days when they did not.

Table 3.

Concurrent and prospective within-person direct effect estimates and moderation

				Concurrent			Prospective
Model	Moderator	Path	b	95% CI	p	b	95% CI	p
Enacted Stigma → Drinking	Unmoderated	Enacted Stigma → Drinking	.02	−.12, .13	.77	−.05	−.19, .06	.37
	Drinking Frequency	Enacted Stigma → Drinking
		Average Slope	−1.11	−1.69, −.76	< .001	−.83	−1.23, −.44	< .001
		Drinking Frequency → Slope	.44	.30, .62	< .001	.30	.15, .44	< .001
	Drinking Quantity	Enacted Stigma → Drinking
		Average Slope	.10	−.14, .27	.32	.07	−.12, .28	.45
		Drinking Quantity → Slope	−.07	−.24, .08	.42	−.12	−.27, .01	.07
	Drinking Motives	Enacted Stigma → Drinking
		Average Slope	−.13	−.56, .36	.62	−.07	−.42, .33	.73
		Coping Motives → Slope	.06	−.14, .22	.57	.01	−.15, .15	.94
		Social Motives → Slope	.04	−.10, .23	.58	−.03	−.15, .09	.59
		Enhancement Motives → Slope	−.10	−.28, .07	.21	−.04	−.19, .11	.52
		Conformity Motives → Slope	−.14	−.32, .03	.10	−.003	−.13, .11	.96
	AUDIT	Enacted Stigma → Drinking
		Average Slope	−.35	−.68, −.02	.03	−.32	−.59, −.06	.01
		AUDIT → Slope	.03	.01, .06	.02	.03	.01, .05	.004
Enacted Stigma → Cannabis Use	Unmoderated	Enacted Stigma → Cannabis Use	−.13	−.29, .06	.22	−.002	−.16, .15	.99
	Cannabis Use Frequency	Enacted Stigma → Cannabis Use
		Average Slope	−1.41	−1.67, −1.15	< .001	−1.08	−1.44, −.76	< .001
		Cannabis Use Frequency → Slope	.08	.07, .10	< .001	.07	.05, .09	< .001
	Cannabis Quantity	Enacted Stigma → Cannabis Use
		Average Slope	−.54	−.98, −.20	< .001	−.07	−.35, .25	.67
		Cannabis Quantity → Slope	.15	.02, .29	.03	.06	−.05, .15	.31
	Cannabis Use Motives	Enacted Stigma → Cannabis Use
		Average Slope	−.82	−1.37, −.42	< .001	−.25	−.65, .15	.21
		Coping Motives → Slope	.27	.10, .40	< .001	.11	−.02, .24	.10
		Social Motives → Slope	.16	−.04, .37	.13	.07	−.09, .24	.38
		Enhancement Motives → Slope	.003	−.19, .18	.97	.03	−.15, .22	.71
		Conformity Motives → Slope	−.61	−1.19, −.21	< .001	−.33	−.70, .03	.06
	CUDIT	Enacted Stigma → Cannabis Use
		Average Slope	−1.64	−2.14, −1.21	< .001	−.97	−1.41, −.59	< .001
		CUDIT → Slope	.11	.09, .15	< .001	.08	.05, .10	< .001

Moderation Models – Alcohol

Baseline drinking frequency moderated both concurrent and prospective within-person associations between enacted stigma and alcohol use. Simple slopes analyses indicated that when individuals who drank monthly or weekly at baseline experienced enacted stigma, they were significantly less likely to drink during the same period or during the subsequent period (e.g., during the next evening) compared to when they did not experience enacted stigma. However, when individuals who drank more than weekly experienced enacted stigma, they were significantly more likely to drink during the same period. Further, when those who drank four or more times per week experienced enacted stigma, they were also more likely to drink during the subsequent period (e.g., during the next evening). See Table 4 for simple slope estimates and Figure 1 for a depiction of the simple slopes for this and subsequent interactions. Baseline drinking quantity and drinking motives did not significantly moderate the within-person association between enacted stigma and alcohol use.

Table 4.

Within-Person Simple Slopes for Significant Interactions

				Concurrent			Prospective
Association	Moderator	Level of Moderator	b	95% CI	p	b	95% CI	p
Enacted Stigma → Drinking	Drinking Frequency	Monthly	−.68	−1.06, −.43	< .001	−.57	−.88, −.31	< .001
		Weekly	−.24	−.46, −.11	< .001	−.26	−.44, −.11	< .001
		2–3 times per week	.20	.01, .31	.04	.05	−.09, .16	.65
		4+ times per week	.65	.35, .90	< .001	.35	.13, .56	.004
	AUDIT	8 (cut point for hazardous use)	−.07	−.24, .09	.39	−.08	−.22, .06	.26
		15 (cut point for DSM-IV dependence)	.17	−.04, .36	.10	.14	−.03, .28	.19
Enacted Stigma → Cannabis Use	Cannabis Use Frequency	−1 SD (3.00 cannabis use days per month)a	−1.16	−1.38, −.91	< .001	−.86	−1.18, −.59	< .001
		Mean (12.03 cannabis use days per month)	−.40	−.54, −.23	< .001	−.22	−.40, −.05	.01
		+1 SD (22.83 cannabis use days per month)	.51	.34, .67	< .001	.55	.35, .79	< .001
	Cannabis Quantity	< 1/8 oz	−.39	−.76, −.10	.01	-	-	-
		1/4 oz	−.09	−.32, .12	.44	-	-	-
		3/4 oz	.23	−.21, .59	.38	-	-	-
		More than 1 oz	.52	−.15, 1.13	.10	-	-	-
	Coping Motives	Almost Never/Never	−.61	−.89, −.33	< .001	-	-	-
		Half of the time	−.05	−.30, .11	.50	-	-	-
		Almost Always/Always	.48	.06, .88	< .001	-	-	-
	Conformity Motives	Almost Never/Never	−.07	−.33, .13	.52	-	-	-
		Half of the time	−1.32	−2.39, −.62	< .001	-	-	-
		Almost Always/Always	−2.56	−4.76, −1.08	< .001	-	-	-
	CUDIT	10 (cut point for DSM-5 CUD)	−.49	−.71, −.28	< .001	−.21	−.41, −.03	.02
		13 (cut point for DSM-IV cannabis dependence)	−.14	−.33, .04	.12	.02	−.14, .18	.83

^alower limit was adjusted to be within window of observed values. “-” indicates interaction was not significant. Simple slopes were calculated at meaningful levels of the moderator when possible (e.g., monthly drinking, coping motives half the time) and at the mean +/− one standard deviation for measures for which meaningful levels were less clear (e.g., cannabis use frequency).

Figure 1.

Visual representations of simple slopes for significant moderation analyses predicting alcohol use. Only moderated concurrent associations are presented because when concurrent and prospective moderations were significant for the same moderator, they had similar simple slope patterns.

Baseline AUDIT scores also moderated both concurrent and prospective associations between enacted stigma and alcohol use, although the moderation of the concurrent association became non-significant (p = .08) when a Benjamini-Hochberg correction was made to reduce the false discovery rate. Simple slopes and Johnson-Neyman analyses indicated that when individuals with AUDIT scores between 1.0 and 2.2¹ experienced enacted stigma, they were less likely to drink during the same period. Further, individuals with AUDIT scores between 1.0 and 4.3 were less likely to drink during the next period when they experienced enacted stigma. On the other hand, when individuals with AUDIT scores above 17.5 experienced enacted stigma, they were more likely to drink during the same period. Similar thresholds were identified in prospective analyses. When individuals with AUDIT scores above 16.4 experienced enacted stigma, they were more likely to drink during the next period. While evidence for cut-points linked to DSM-5 alcohol use disorder criteria is mixed (Hagman, 2016; Moehring et al., 2019), cutpoints of 8 for hazardous drinking and 15 for DSM-IV dependence are supported (Saunders et al., 1993). This suggests that enacted stigma is a risk factor for alcohol use among individuals who have DSM-IV dependence (and likely severe DSM-5 alcohol use disorder). See Table 5 for information about the proportion of the analytic sample that fell within the values of the moderators identified by the Johnson-Neyman analyses.

Table 5.

Percentage of sample at values of the moderator identified in Johnson-Neyman analyses

Association	Moderator	Level of Moderator	Percentage in Range
Enacted Stigma → Drinking	Drinking Frequency	Monthly to Weekly	42.0%
		More than Weekly	58.0%
	AUDIT	1.0 to 2.2	2.2%
		1.0 to 4.3	12.7%
		> 16.4	10.6%
		> 17.5	9.0%
Enacted Stigma → Cannabis Use	Cannabis Use Frequency	3.00 to 15.02	55.0%
		3.00 to 12.74	44.0%
		18.46 to 31.00	43.4%
		17.33 to 31.00	45.0%
	Cannabis Quantity	1.00 to 2.22 (≤ 1/8 oz)	62.7%
		More than 1 oz	3.6%
	Coping Motives	1.00 to 2.63	66.7%
		4.61 to 5.00	7.9%
	Conformity Motives	1.00 to 1.12	79.6%
		1.13 to 5.00	20.4%
	CUDIT	1.0 to 12.6	51.9%
		1.0 to 10.4	42.5%
		> 15.8	26.4%
		> 15.0	33.0%

Moderation Models – Cannabis

Baseline cannabis use frequency moderated the concurrent and prospective within-person associations between enacted stigma and cannabis use. Simple slopes and Johnson-Neyman analyses indicated that when individuals who used cannabis 3.00 to 15.02 days in the past month experienced enacted stigma, they were significantly less likely to use cannabis during the same period (Figure 2). Further, when those who used cannabis 3.00 to 12.74 days in the past month experienced enacted stigma, they were also significantly less likely to use cannabis during the subsequent period (e.g., the next evening). However, the pattern was reversed for those who use cannabis more frequently. When individuals who used cannabis 18.46 to 31.00 days in the past month experienced enacted stigma, they were significantly more likely to use cannabis during the same period than when they had not experienced enacted stigma. The prospective association between enacted sigma and cannabis use was significant and positive for individuals who used cannabis 17.33 to 31.00 days in the past month.

Figure 2.

Visual representations of simple slopes for significant moderation analyses predicting cannabis use. Only moderated concurrent associations are presented because when concurrent and prospective moderations were significant for the same moderator, they had similar simple slope patterns.

Baseline cannabis use quantity also significantly moderated the concurrent (but not prospective) within-person association between enacted stigma and cannabis use. For individuals who reported using an eighth of an ounce or less (values of 1.00 to 2.22) of cannabis per week, experiencing enacted stigma was associated with a decreased likelihood of using cannabis during the same period. The association between enacted stigma and cannabis use was non-significant for those who use a quarter of an ounce or more of cannabis, although the association became positive and approached significance at very high quantities of use (more than one ounce per week). Only 3.6% of the sample reported consuming this amount of cannabis weekly.

Two baseline cannabis use motives, coping and conformity, moderated the concurrent associations between enacted stigma and cannabis use. When individuals who reported almost never/never using cannabis to cope or using cannabis to cope less than half the time (values between 1.00 and 2.63) experienced enacted stigma, they were less likely to use cannabis than on days when they did not experience enacted stigma. The pattern was reversed for those who almost always/always used cannabis to cope (values of 4.61 to 5.00).² These individuals were significantly more likely to use cannabis on days when they experienced enacted stigma than on days when they did not. Among individuals who reported almost never/never using cannabis to conform (values of 1.00 to 1.12), experiencing enacted stigma was not significantly associated with cannabis use. However, when individuals who used cannabis to conform half the time or more (values of 1.13 to 5.00) experienced enacted stigma, they were significantly less likely to use cannabis compared to when they did not experience enacted stigma. Social and enhancement motives did not moderate within-person associations between enacted stigma and cannabis use.

Baseline CUDIT-R scores also moderated concurrent and prospective associations between enacted stigma and cannabis use. Among those who had CUDIT-R scores between 1.0 and 12.6, experiencing enacted stigma was associated with a lower likelihood of using cannabis during the same period. Among individuals who had scores between 1.0 and 10.4, experiencing enacted stigma was associated with a lower likelihood of using cannabis during the next period. Among those who had CUDIT-R scores greater than 15.8, experiencing enacted stigma was associated with a higher likelihood of using cannabis during the same period. Among individuals who had CUDIT-R scores greater than 15.0, experiencing enacted stigma was associated with a higher likelihood of using cannabis during the next period. Prior research has suggested 10 as a cut point for mild DSM-5 CUD on the CUDIT-R (Bonn-Miller et al., 2016), while 13 was proposed as a cut point for DSM-IV dependence (Adamson et al., 2010). This suggests that enacted stigma is a risk factor for cannabis use at the daily level among individuals with moderate to severe CUD, but not for those without CUD or with mild CUD.

Sensitivity Analyses

In the prospective analyses presented above, we allowed enacted stigma reported in a morning survey to predict alcohol or cannabis use reported in the evening survey as well as enacted stigma reported in the evening to predict substance use reported in the morning survey. To determine whether including both types of lags affected results, we conducted sensitivity analyses in which only enacted stigma reported in the evening survey (covers ~8 am to 6pm) predicted alcohol and cannabis use reported in the next morning survey (covers ~6pm to 8am). Analyses demonstrated the same pattern of findings, suggesting that including both evening to morning and morning to evening lags did not change findings.

Discussion

The current study was the first to examine whether the effects of minority stress on alcohol and cannabis use are dependent on an individual’s pattern of substance use and the presence of SUDs. Results largely indicate that enacted stigma is a risk factors for alcohol and cannabis use at the event-level among SMWGD who have already developed an SUD and/or use frequently. However, evidence for the moderating roles of substance use quantity and motives was more mixed. Findings provide insight into potential moderators that may explain mixed evidence for associations between enacted stigma and substance use. Results also suggest a new direction for research on risk factors for substance use among SMWGD – testing further hypotheses derived from the integration of minority stress theory (Hatzenbuehler, 2009; Meyer, 2003) and the multistage model of drug addiction (Koob, 2013; Koob & Volkow, 2010).

Alcohol and cannabis use frequency and AUD/CUD were consistent moderators of the association between enacted stigma and alcohol/cannabis use in the current study. Among those who used cannabis less than every other day, drank weekly or less, or did not have a probable AUD or CUD, enacted stigma predicted a reduced likelihood of same or next-day alcohol or cannabis use. While this is not consistent with minority stress theory, which would predict a positive association between enacted stigma and substance use (Hatzenbuehler, 2009; Meyer, 2003), it is consistent with the multistage model of drug addiction (Koob, 2013; Koob & Volkow, 2010). This suggests that individuals who use substances less frequently and/or do not have an SUD may be less likely to use substances in the hours following experiences of minority stress. This may be because, as theorized by the multistage model, substance use is driven by positive reinforcement (e.g., increasing sociability and positive affect) rather than negative reinforcement (e.g., coping with minority stress) among individuals who use substances less frequently and/or do not have SUDs. As a result, minority stress (a risk factor related to negative reinforcement) may be less likely to trigger substance use on the same or next day in these individuals. Of note, the finding that enacted stigma predicted a reduced likelihood of substance use at the daily level among individuals who used substances less frequently does not indicate that minority stress is not harmful for individuals who use substances less frequently, rather they may be using other techniques for coping with the negative affect arising from minority stress (e.g., rumination), which may increase their risk for developing mood or anxiety disorder (Dyar, Dworkin, et al., 2021; Sarno et al., 2020).

On the other hand, those who drank more than weekly, used cannabis on more than half of days, or had probable AUD or CUD were more likely to use alcohol or cannabis after experiencing enacted stigma. Again, this may be explained by the theorized driving role of negative reinforcement in substance use among individuals who use frequently and/or have SUDs. This may lead these individuals to engage in the use of substances to cope following experiences of enacted stigma. In fact, an individual’s typical coping motives also moderated the event-level association between enacted stigma and cannabis use, with simple slope patterns mirroring those for frequency of use and AUD/CUD. Given that the association between enacted stigma and same-/next-day cannabis use was only significant among individuals who endorsed high coping motives, coping motives appear to play an important role in differentiating those for whom enacted stigma is an event-level risk factor for cannabis use and those for whom it is not. Future research should examine whether individual-level factors like frequency of use or AUD/CUD diagnoses moderate mechanistic event-level associations between enacted stigma, same/next-day coping motives, and subsequent substance use to provide further insight into this process. Importantly, these findings shed light on the experiences of SMWGD with SUDs by demonstrating that enacted stigma is a risk factor for same-/next-day alcohol and cannabis use among those with an SUD, but not among those without a SUD. This demonstrates the importance of addressing minority stress in interventions for SMWGD with SUDs.

Given that the average frequency of substance use and prevalence of SUDs vary widely from study to study, the moderation effects identified by this study may help to explain why some studies find associations between enacted stigma and substance use (Ehlke et al., 2021; Lewis et al., 2021), while others do not (Dyar, Dworkin, et al., 2021). Other factors likely also contribute to disparate findings across studies. Findings also suggest that researchers should carefully consider inclusion criteria based on substance use frequency for EMA studies as setting higher or lower criteria may substantially affect results.

Coping motives did not moderate the association between enacted stigma and alcohol use, contrary to expectations. It is unclear why coping motives moderated the association between enacted stigma and cannabis use but not alcohol use. The means and distributions of coping motives variables were similar across cannabis and alcohol, indicating that these are unlikely to be the source of these different effects. Future research should attempt to replicate these findings and, if demonstrated in another sample, explore potential factors that may help to explain why individual-level coping motives do not moderate the event-level association between enacted stigma and alcohol use.

An individual’s typical social and enhancement motives did not moderate event-level associations between enacted stigma and alcohol or cannabis use. These findings stand in contrast to hypotheses derived from the multistage model, which posit that enacted stigma would predict substance use among individuals who endorsed low levels of positive reinforcement motives for substance use (i.e., social and enhancement motives). The multistage model posits that positive reinforcement motives give way to negative reinforcement motives as SUDs develop and substance use becomes frequent. However, these findings may suggest that positive reinforcement does not become less important as negative reinforcement begins to affect substance use. Rather, for at least some individuals whose substance use is driven by negative reinforcement, positive reinforcement may also remain relevant. Future research examining positive reinforcement processes at the event-level should test this aspect of the multistage model as it plays out in daily life.

Clinical Implications

Given dramatic disparities in alcohol and cannabis use disorders affecting SMWGD (Krueger et al., 2020; Philbin et al., 2019), there is an urgent need for interventions that reduce SUDs in this population. Results of the current study suggest that minority stress is a temporally proximal risk factor for substance use among those who use frequently or have SUDs. This indicates that interventions aiming to treat AUD or CUD in this population would benefit from addressing minority stress and, likely, incorporating skill building around coping with minority stress. Preventive interventions may also likely benefit from coping skill building and addressing the cycle of negative reinforcement that can arise from using substances to cope. This may help to reduce the likelihood that infrequent use will progress to AUD or CUD, although this should be tested in future research.

Two sets of interventions may be most amenable to addressing these factors. First, two effective interventions have been developed to improve mental health among sexual minority individuals by using a minority stress focused cognitive behavioral approach (Pachankis et al., 2015; Pachankis et al., 2020). These interventions include components that address minority stress, building coping skills, and reducing problematic alcohol use. While developed to reduce anxiety, depression, and alcohol use, adaptation may also make these interventions effective in reducing problematic cannabis use. Second, interventions targeting substance use motives developed with the general population have been effective in reducing substance use. For example, interventions using motivational enhancement and brief interventions targeting drinking motives have successfully reduced alcohol use and consequences (Blevins & Stephens, 2016; LaBrie et al., 2008). While fewer interventions have focused on cannabis use, cognitive behavioral therapy and motivational enhancement therapy have reduced coping motives, cannabis use, and consequences (Banes et al., 2014). While these interventions have not been tested with SMWGD, they could be adapted to address negative reinforcement processes contributing to substance use, including minority stress. Finally, while interventions to reduce the impact of minority stress on SMWGD are necessary, there is a critical need for population and system-level interventions to reduce minority stress and promote equity for individuals of all genders and sexual orientations.

Limitations

Study findings should be considered in light of their limitations. First, only SMWGD who used alcohol and cannabis regularly and lived in the US were included in this study. As a result, it is unclear whether findings will generalize to sexual minority men, those assigned male at birth, sexual minorities who live outside the US, or those who use alcohol or cannabis less frequently. An important direction for future research is to test whether these findings generalize to broader experiences of daily stress not related to a minoritized identity among the general population. Second, due to the study design, we were only able to examine whether individual-level factors moderated event-level associations. While this approach has numerous advantages, including the identification of temporally proximal risk factors, it did not allow for a full exploration of the proposed moderation process. For example, we could not examine whether similar moderation effects were present for longer-term associations between minority stress and substance use or whether changes within individuals over time resulted in changes in the impact of minority stress on substance use. Other study designs may also allow for the inclusion of individuals who use substances less frequently (e.g., less than monthly) as such individuals would not provide adequate data to be included in an EMA study. Future research with diverse study designs should examine this moderation effect in different contexts to advance our understanding of processes affecting the associations between minority stress and substance use. Further, we examined the moderation of associations between minority stress and the likelihood of alcohol of cannabis use at the event-level. We did not examine the moderation of associations between minority stress and other alcohol and cannabis use outcomes (e.g., quantity consumed, heavy episodic drinking, etc.). Future research should examine whether the same or different patterns of moderation are present when other types of substance use outcomes are examined. Finally, some of the values of the moderator at which the associations between minority stress and alcohol or cannabis use were significant were represented by a relatively small proportion of the sample. For example, only 2.2% of participants were within the range of values of the AUDIT at which the concurrent association between enacted stigma and alcohol use was significantly negative. However, this was not the case for the majority of analyses.

Conclusions

The current study was the first to test whether the association between enacted stigma and substance use is dependent on an individuals’ pattern of substance use. Findings provide initial evidence in support of an integrated model of minority stress theory and the multistage model of drug addiction by demonstrating that enacted stigma predicts a higher likelihood of alcohol and cannabis use only among those who use frequently or have an SUD. Nuanced findings emerged regarding other moderators, with enacted stigma predicting cannabis use only among those with high coping motives. However, motives did not moderate the association between enacted stigma and alcohol use. Findings suggest that interventions designed to address AUD and CUD use among SMWGD would benefit from addressing minority stress and coping skill-building.