Daily and Momentary Associations Between Gender Minority Stress and Resilience With Alcohol Outcomes

Abstract

Background and purpose

Minority stressors have been linked with alcohol use among transgender and gender diverse (TGD); however, no ecological momentary assessment studies have examined daily links between minority stress and alcohol use specifically among TGD. This study examined gender minority stressors and resilience as predictors of same-day or momentary alcohol-related outcomes. Feasibility and acceptability of procedures were evaluated.

Methods

Twenty-five TGD adults (mean age = 32.60, SD = 10.82; 88% White) were recruited Canada-wide and participated remotely. They completed 21 days of ecological momentary assessment with daily morning and random surveys (assessing alcohol outcomes, risk processes, gender minority stressors, resilience), and an exit interview eliciting feedback.

Results

Gender minority stress had significant and positive within-person relationships with same-day alcohol use (incidence risk ratio (IRR) = 1.12, 95% confidence interval [CI] [1.02, 1.23]), alcohol-related harms (IRR = 1.14, 95% CI [1.02, 1.28]), and coping motives (IRR = 1.06, 95% CI [1.03, 1.08]), as well as momentary (past 30-min) alcohol craving (IRR = 1.32, 95% CI [1.18, 1.47]), coping motives (IRR = 1.35, 95% CI [1.21, 1.51]), and negative affect (IRR = 1.28, 95% CI [1.20, 1.36]). Gender minority stress indirectly predicted same-day drinking via coping motives (ab = 0.04, 95% CI [0.02, 0.08]). Resilience was positively associated with same-day alcohol use (IRR = 1.25, 95% CI [1.03, 1.51]) but not harms.

Conclusions

TGD adults may use alcohol to cope with gender minority stress, which can increase the risk for alcohol-related harms. Interventions are needed to eliminate gender minority stressors and support adaptive coping strategies.

Transgender and gender diverse adults may use alcohol to cope with gender minority stress, which can increase their risk of experiencing alcohol-related harms.

The health needs of transgender and gender diverse (TGD) adults have been and remain underserved [1, 2]. TGD is an umbrella term referring to individuals whose gender identity differs from the sex they were assigned at birth, including binary transgender (e.g., women, men) and nonbinary (e.g., agender, genderqueer, genderfluid) identities. At least 0.6% of adults in North America identify as TGD; yet, they continue to be underrepresented and underserved in research and health care systems [1, 3, 4]. These unmet needs are a concern because TGD individuals are at risk for experiencing additional stressors due to their minoritized status in society, which can put them at risk for a number of negative mental health and health outcomes [5, 6], such as alcohol use and harms.

TGD adults report problematic drinking at concerning rates. A 2018 review of alcohol-related studies among TGD adults (N = 44, 80% with U.S. samples) [7] found that TGD adults exhibited high prevalence rates of past-year hazardous drinking (47%–48% prevalence based on Alcohol Use Disorder Identification Test [AUDIT] scores), past-year alcohol use disorder (AUD; 11%–26% prevalence), and alcohol-related harms (e.g., being victimized when intoxicated). Furthermore, TGD individuals may be disproportionately impacted by alcohol-related harms. In a large-scale cross-sectional study of 15,637 transgender individuals and 46,911 cisgender individuals in the USA, transgender adults had elevated rates of AUD (2.6%) compared to cisgender adults (0.9%) [8]. In another cross-sectional U.S. sample of university students (N = 255,788), TGD individuals (compared to cisgender women) reported higher rates of nonconsensual sex and mental health symptoms when drinking [9]. Moreover, compared to cisgender women, TGD individuals experienced more alcohol-related harms at lower levels of drinking.

Gender Minority Stress and Alcohol Outcomes

The Gender Minority Stress and Resilience Model (GMSRM) is an existing framework for understanding the psychopathology inequities experienced by TGD adults [6, 10]. The GMSRM is grounded in the empirically supported Sexual Minority Stress Model, which proposes that inequities in psychopathology for sexual minority individuals are due to the added stressors they experience from societal homophobia [11, 12]. Similarly, the GMSRM proposes that TGD individuals experience inequities due to the excess stress they experience from societal transphobia [6]. TGD people report high levels of distal stressors, such as physical and sexual violence, discrimination, and harassment, due to transphobia and stigma related to their gender identity and/or expression [13, 14]. The model proposes that these distal stress processes increase TGD people’s risk of various mental health outcomes. Cross-sectional research has supported relations between distal minority stressors and higher psychological distress, depression, anxiety, suicidality, and heavy drinking [15–19]. Closely related internal risk processes, called proximal processes (e.g., internalized transphobia, anticipation of stressful events, and associated vigilance), can also adversely impact TGD individuals’ mental health and well-being [6]. These proximal processes are correlated with increased distress, mental health symptoms, and identity concealment [15, 20, 21].

Gender minority stressors could influence drinking via several mechanisms. Minority stress could increase alcohol craving, which increases in response to general stress and is linked to increased alcohol use [22]. Minority stress increases distress, such as anxiety and sadness [23], which in turn could increase alcohol use as a coping strategy [24]. In fact, research with sexual minority individuals has supported distress as a factor linking discrimination to alcohol use [25]. Similarly, minority stress can activate motives to drink to cope with distress. Coping motives are an established risk factor for harmful drinking in adults [26, 27]. Furthermore, research has supported that coping motives link stigma and alcohol use in minoritized and marginalized groups [28]. This may also apply to TGD adults, as many as 72% of whom reported coping motives in prior research [29]. Determining how these drinking-specific risk processes impact alcohol outcomes among TGD adults could help identify therapeutic targets to reduce drinking harms.

Resilience Factors and Alcohol Outcomes

Resilience refers to why some who experience significant adversity survive and “bounce back,” while others do not [15, 30]. Resilience factors in the GMSRM include community connectedness, pride, and social support [10]. Resilience factors, like social or peer support, community connectedness, and stronger transgender identity, have been linked to improved mental health [31] and can be life-saving [32]. Furthermore, transgender adults who reported greater perceived social support from family, friends, or significant others exhibited a lower risk of self-injury and lower depressive symptoms [33, 34]. Ultimately, these resilience factors could weaken the link between minority stress and alcohol harms and, therefore, act as key intervention targets. Resilience factors are understudied for alcohol outcomes, but based on existing evidence, could lessen the harmful impacts of minority stress.

Present Study

Nearly all studies focusing on TGD adults’ alcohol outcomes have been cross-sectional (95% of those in a recent review) and descriptive [7]. As a result, very little is known about the processes that impact drinking in the moments when TGD adults drink that could be targeted in alcohol-related interventions. The overarching goal of this investigation was to test the role of GMSRM processes for alcohol outcomes for the first time with TGD adults using ecological momentary assessment (EMA). EMA is a widely used technique in the alcohol field and is capable of examining alcohol outcomes and risk processes in naturalistic settings while reducing memory recall errors [35]. EMA has been used previously to examine associations between minority stress and alcohol outcomes among sexual and gender minorities, with some studies supporting the anticipated associations [23, 36], although others have not [37]. These studies, however, grouped gender minority participants with cisgender sexual minorities, with TGD participants being underrepresented in the samples. It is important to establish the extent to which gender minority stressors can help explain alcohol outcomes among TGD people specifically. Focusing on TGD experiences is important because TGD have tended to be ignored in research and it is important to specifically focus on the impacts of experiences relevant to TGD people (e.g., transphobia, misgendering).

To accomplish this, we used EMA and multilevel models to test specific hypothesized pathways; chiefly, relations between gender minority stress and resilience with drinking risk processes (e.g., coping motives, negative affect) and alcohol outcomes (e.g., craving, use, harms). It was hypothesized that experiencing gender minority stress would be associated with same-day or momentary increases in alcohol craving, coping motives for drinking, negative affect, and increased alcohol use. Furthermore, it was anticipated that coping motives would mediate the associations between gender minority stress and alcohol outcomes. It was also expected that resilience factors would be associated with momentary decreases in alcohol craving, coping motives, negative affect, and alcohol use. Last, the acceptability and feasibility of the design were explored as this was the first known EMA study focused on TGD adults’ alcohol outcomes.

Methods

Participants

Twenty-five TGD adults, ages 19 years or older who drink alcohol, were recruited. This sample size, chosen to be consistent with the size of prior EMA pilot studies [38, 39], allows for establishing the feasibility and acceptability of the design and adequately powers the study to detect key within-person relationships of interest (described in Analytic Approach). Eligibility criteria were: (i) self-identified under the TGD umbrella; (ii) resided in Canada; (iii) consumed alcohol regularly (defined by consuming one or more standard drinks, at least twice per week, in the past 30 days); (iv) drank heavily recently (defined as consuming four or more standard drinks in a sitting at least once in past 30 days); and (v) had a personal device compatible with the EMA app (i.e., Android or iPhone devices). Exclusion criteria were: (i) trying to reduce alcohol use or receiving related treatment; (ii) self-reported diagnosis of severe mental illness(es) (i.e., bipolar, mania, psychosis); (iii) inability to complete the study in English due to a language barrier; (iv) regularly consuming alcohol prior to noon (as this would interfere with completing the daily morning surveys); and (v) using recreational drugs other than cannabis and nicotine-containing products at least once a month. Participants were recruited through social media, research team networks, community flyers, and Canada-wide community partners.

Procedure

The study procedure, aims, hypotheses, and general analytic approach were preregistered on September 15, 2021 (https://osf.io/7wshk). The study received ethics approval from the Toronto Metropolitan University Ethics Board (ID 2021-347). The study was entirely virtual using video-based online appointments. Trained research assistants conducted 2-hr, virtual, baseline appointments with eligible participants, which consisted of informed consent, interviews, and surveys. Participants were guided to download an EMA app onto their smartphones and then were instructed on the EMA protocol. Afterwards, participants began a 21-day EMA period, which started the day after the baseline appointment. After the EMA period, participants completed a 1-hr virtual exit appointment where they repeated a subset of baseline measures and responded to questions about their experiences in the study. Data collection occurred from January to September of 2022.

EMA protocol

Participants were asked to download an app onto their personal devices to complete the EMA surveys (MetricWire; Waterloo, Ontario). The app registered and saved information regardless of whether data connection was available. The daily morning survey (2–5 min in length) was available each day in the app between 6 am and 1 pm, with reminders provided at 9 am and 12 pm. Surveys not completed by 1 pm were considered missing. The random surveys (2 min) were prompted twice (one time per interval 1–6 pm and 6–11 pm). This study analyzes data from the daily morning and random surveys.

Participant compensation

Participants were compensated up to $170 Canadian in e-gift cards, including $40 for the baseline and $20 for the exit visits. Weekly incentives for the EMA were based on the completion rate of the daily and random surveys: $30/week for 90%–100%, $25 for 80%–89%, $20 for 70%–79%, $15 for 50%–69%, $10 for 20%–49%, and $5 for 1%–19% completion. If 90% or more of all daily and random surveys were completed across the 3 weeks, participants also received a $20 bonus.

Measures

Baseline and exit visits

Applicable measures from the baseline visit included demographics, AUDIT [40], baseline alcohol use from the timeline follow-back (TLFB) [41], and DSM-5 AUD diagnosis based on the Diagnostic Assessment Research Tool (DART) [42]. Exit visit measures examined here included an adaptation of the Acceptability of Intervention Measure (AIM), Intervention Appropriateness Measure (IAM), and Feasibility of Intervention Measure (FIM) [43]. The wording of the questions was changed to ask about the EMA (e.g., “The daily app surveys meet my approval”) and virtual appointments (e.g., “The virtual appointments seem easy to use”). Each statement was rated on a 5-point Likert scale, ranging from “1 = completely disagree” to “5 = completely agree.” Multiple-selection questions asked about preferences for in-person/virtual formats for interviews if they were to participate in a study like this again. Open-ended prompts also allowed participants to provide any additional feedback about their experience in the study.

Daily morning EMA measures

Gender minority stress and resilience experiences from “yesterday” were assessed with eight stress and three resilience items adapted from the GMSRM [10], including: “I experienced difficulties getting services, finding or staying in housing, or with employment” (discrimination), “I was rejected or distanced from others (friends, family, and/or work)” (rejection), “I was verbally, sexually, or physically harassed, teased, threatened, or harmed” (victimization), “I was misgendered or my gender identity was not respected” (non-affirmation), “Because of my gender identity or gender expression, I felt like an outcast” (internalized transphobia), “I thought if I expressed my gender identity and/or history, most people would think less of me” (negative expectations), “Because I did not want others to know my gender identity, I changed my way of speaking, dressing, or behaving” (nondisclosure), “I was proud to be a person whose gender identity is different from my sex assigned at birth” (pride), “I felt part of a community of people who share my gender identity” (community connectedness), and two items asking about challenges or benefits not listed. Response options were “No,” “Yes, once,” and “Yes, multiple times.” The respective items were summed to create composite gender minority stress and resilience scores.

The within- and between-person internal consistency of multi-item EMA measures was assessed using McDonald’s ω [44]. From the daily morning survey, within-person ω = 0.80 for stress and 0.57 for resilience, and between-person ω = 0.88 for stress and 0.94 for resilience. The following coping motives items from the Drinking Motives Questionnaire—Revised Short Form (DMQ-R SF) [45] were included in the daily survey asking about motives the day prior: “To feel less depressed or nervous,” “To cheer up,” and “To forget about [their] problems” (within-person ω = 0.79, between-person ω = 0.96). Response options ranged from 0 (“Not at all”) to 4 (“Extremely”). If participants endorsed prior day alcohol use (N = 147), they were asked in the daily surveys about 20 alcohol-related harms, including all items from the Short Inventory of Problems—Revised (SIP-2R) [46] and two items concerning sexual risk taking from the Young Adult Alcohol Consequences Questionnaire (YAACQ) [47]. Participants indicated whether or not they experienced any of these “between the time [they] first started using alcohol yesterday and now.” Ten alcohol-related harms items were not endorsed at all during the EMA (harm to health, family hurt due to drinking, relationships damaged, social harms, negative work performance, physical fights, accidents, blackouts, unwanted sexual situations, and regret related to sexual situations) and thus they were excluded from the index of alcohol harms. Responses to the other 12 harms were summed for each participant at the day level to create a count of the number of harms experienced on each drinking day, which had good within-person (ω = 0.99) and between-person (ω = 0.83) internal consistency.

Random momentary EMA measures

The same GMSRM items were assessed but with binary response options (yes/no) indicating if the experiences occurred in the past 30 min (within-person ω = 0.61 for stress items, 0.64 for resilience items; between-person ω = 0.71 for stress, 0.87 for resilience). The same coping motives items were in the random surveys but asked about the past 30 min (within-person ω = 0.85, between-person ω = 0.97). A visual analog scale with anchor points of 1 and 10 was used for alcohol craving experienced “right now” [48]. Negative affect in the past 30 min was assessed with five items (upset, hostile, ashamed, nervous, afraid) from the Positive and Negative Affect Scale (PANAS; within-person ω = 0.65 and between-person ω = 0.64) with response options ranging from 0 (“Not at all”) to 4 (“Extremely”) [49].

Analytic Approach

Descriptive analyses were used to characterize the baseline characteristics of the sample. To examine the feasibility and acceptability of the study, point estimates (averages) and corresponding 95% confidence intervals (CIs) were computed for: (a) EMA survey compliance (proportion of complete surveys of all daily and random survey prompts), and (b) EMA question compliance (proportion of daily/random questions answered vs. sent). We examined the number of people who completed both baseline and exit visits. For acceptability, we examined means with 95% CIs and ranges of acceptability, appropriateness, and feasibility ratings. To determine whether EMA compliance was associated with key demographic or alcohol-related variables, we tested associations between EMA survey compliance scores and demographic variables, alcohol-related variables at baseline, and average EMA feasibility and acceptability scores.

Analyses were conducted using the following packages with the statistical software R [50]: lmerTest [51], misty [52], glmmTMB [53], and performance [54]. Using multilevel models, we estimated the reliability of multi-item EMA measures, including gender minority stress, gender minority resilience, coping motives, and alcohol harms. Mean daily and momentary levels of alcohol-related outcomes and their standard errors (SEs) were calculated using unconditional multilevel models. Negative binomial regression models were then used to test direct within-person relationships between gender minority stress and resilience (centered within-person Level 1 predictors) with each of the following outcomes in separate models: momentary alcohol craving, coping motives, and psychological distress, as well as same-day alcohol use, alcohol-related harms, and coping motives. The indirect effect of gender minority stress and gender minority resilience on same-day alcohol use and alcohol-related harms via coping motives was analyzed using a within-person multilevel mediation model (all mediation associations at Level 1 with continuous predictors and mediator centered within-person). CIs around the indirect effects were calculated using Monte Carlo simulation [55]. Negative binomial models, rather than linear, were used given that outcomes were count variables. Decisions to include random slopes were also based on relative model fit and likelihood ratio tests. To account for between-person variance in the predictor variables of interest, the grand-mean centered person-level means for gender minority stress and gender minority resilience (and coping motives, for mediation models) were also included in the models at Level 2 [56]. The following uncentered, categorical covariates were also included in all models: a binary day-of-the-week covariate indicating whether a given observation occurred on a weekend (Friday, Saturday, or Sunday) at Level 1, participants’ age, and dummy-coded gender identity (transgender women, transgender men, and gender diverse or nonbinary) at Level 2.

Our number of participants (Level 2) was largely based on the minimum number of participants needed to establish the feasibility of the study design in this population. Statistical power for the planned within-person analyses was roughly estimated based on Table 5 of Arend and Schäfer [57] using a Level 2 sample size of 30 and Level 1 sample size of 16 (as these most closely approximated our anticipated sample size of 25 and a predicted 80% completion rate of the 21 daily morning surveys), which provides 80% power to detect a small effect size of 0.21 across small to large ICCs. Based on these values, we expected that our study would likely be powered to detect small-to-medium within-person effect sizes. As alcohol harms were only examined on drinking days, there were fewer Level 2 observations for this outcome (N = 147), reducing statistical power.

Results

Descriptive Characteristics

In total, 25 TGD adults were eligible, consented, and completed the baseline appointment. Descriptive characteristics of the sample are summarized in Table 1. The sample had a mean age of 32.60 years (SD = 10.82). The represented gender identities included 7 transgender men, 7 transgender women, and 11 gender diverse or nonbinary individuals (i.e., agender, gender questioning, genderqueer).

Table 1.

Demographic Characteristics of Participants (N = 25)

Characteristic	n	%
Race/ethnicity
East Asian, Southeast Asian, South Asian, or Pacific Islander	1	4
American Indian/Native American	1	4
White	22	88
No response	1	4
Gender identity
Trans woman	7	28
Trans man	7	28
Gender fluid/nonbinary	11	44
Sexual orientationa
Straight/heterosexual	2	8
Gay or lesbian	9	36
Bisexual	8	32
Pansexual	7	28
Asexual	1	4
Bicurious/questioning	1	4
Queer	13	52
Household income (CAD)
$0–$10,000	2	8
$10,000–$14,999	2	8
$15,000–$19,999	4	16
$20,000–$29,999	3	12
$30,000–$39,999	2	8
$40,000–$49,999	4	16
$50,000–$74,999	5	20
$75,000 or more	3	12
Highest level of education
Some high school, no diploma	1	4
High school diploma or equivalent (GED)	1	4
Some college credit, no degree	8	32
Trade/technical vocational training	3	12
Bachelor’s degree	9	36
Master’s degree	3	12
Employment status
Working (paid employee)	16	64
Working (self-employed)	2	8
Not working (laid off, looking for work, or disabled)	3	12
Not working (student)	4	16
Province/Territory of residence
Alberta	2	8
British Columbia	6	24
Manitoba	2	8
New Brunswick	2	8
Nova Scotia	4	16
Ontario	7	28
Quebec	1	4
Saskatchewan	1	4
Met alcohol use disorder criteria (DART)	10	40
	Mean	SD
Age at time of survey (years)	32.60	10.82
AUDIT scoreb	10.71	6.95
Drinks per week (TLFB)	9.51	9.16

AUDIT Alcohol Use Disorder Identification Test; DART Diagnostic Assessment Research Tool; TLFB Timeline Follow-back.

^a n identified by the number of selections and therefore exceeds the number of participants.

^bOne participant did not complete these items and was not included in this computation.

Feasibility and Acceptability Data

The average FIM rating for EMA, 4.15 out of 5 (95% CI [3.98, 4.32], and virtual appointments, 4.42 (95% CI [4.28, 4.56]), suggests that participants found the EMA and virtual appointments feasible. The average AIM score for EMA, 3.70 out of 5 (95% CI [3.52, 3.88]), and average IAM score for EMA, 3.95 (95% CI [3.78, 4.12]), suggests that the participants found the EMA component somewhat acceptable and somewhat appropriate. The average AIM score for virtual appointments, 4.41 (95% CI [4.27, 4.55]), and IAM rating of 4.48 out of 5 (95% CI [4.36, 4.60]) suggest the participants found the virtual appointment format acceptable and appropriate. If the participants were to enroll in a similar study again, 92.00% of the participants indicated that they would prefer for the interviews to be completed with video-conferencing, and 8.00% would prefer for the interviews to be replaced by web-based surveys that they could complete on their own. Furthermore, 92.00% indicated that they would prefer for the interviews to occur virtually because they do not reside close enough to the university to attend in person, and 8.00% indicated that they would prefer for the interviews to occur virtually, even though they reside close enough to the university to attend in-person. Written feedback about the EMA and virtual appointments was largely neutral and constructive.

Patterns of Missing EMA Data

Participants completed an average of 79.62% (95% CI [70.00%, 89.23%]) of the daily morning EMA surveys and 75.22% (95% CI [67.57%, 83.10%] of the random surveys. Spearman’s rank-order correlation coefficients were not significant for the associations between survey compliance and participants’ age (ρ = −0.11, p = .587), income (ρ = 0.001, p = .997), alcohol-related harms (ρ = 0.32, p = .124), weekly alcohol use (ρ = 0.04, p = .831), baseline AUDIT scores (ρ = 0.38, p = .062), and average FIM (ρ = 0.28, p = .170), AIM (ρ = 0.23, p = .264), and IAM (ρ = 0.22, p = .287) scores. One-way analysis of variance F-tests were also not significant for models examining associations between compliance and gender (F = 0.21, p = .810), race/ethnicity (F = 0.37, p = .776), sexual orientation (F = 1.79, p = .170), education (F = 1.24, p = .330), employment (F = 0.79, p = .567), and province of residence (F = 1.19, p = .358).

EMA Patterns of Daily Alcohol Use, Alcohol-Related Harms, and Momentary Craving

Over the course of the EMA period, participants’ average daily alcohol use (including drinking and nondrinking days) was 1.23 drinks per day (SE = 0.20). On drinking days only, the mean drinks consumed were 3.69 (SE = 0.34). Daily alcohol-related harms experienced during the EMA period were only evaluated on days on which participants drank, with participants reporting an average of 0.37 (SE = 0.07) alcohol-related harms during the EMA period. The range of harms experienced over the EMA period was between 0 and 3, with participants reporting experiencing any alcohol-related harm on a total of 35 daily morning surveys. Average levels of momentary alcohol craving were 1.70 (on a 10-point visual analog scale; SE = 0.26) and participants reported average momentary distress values of 3.29 (SE = 0.42). Finally, participants reported an average momentary level of drinking to cope of 2.56 (SE = 0.54).

Same-Day Predictors of Daily Alcohol Use, Alcohol-Related Harms, and Coping Motives

Table 2 displays the findings of the multilevel regression models. The results for analyses of same-day predictors are reported for 24 participants, as one participant only completed one daily morning survey. Random slopes models did not perform better than random intercepts-only models for the direct effect of gender minority stress and gender minority resilience on same-day alcohol use, alcohol-related harms, or coping motives (ΔAIC and ΔBIC = 2–22), so the results from the models with fixed slopes are reported. The negative binomial multilevel model results showed that gender minority stress had a significant and positive within-person relationship with same-day alcohol use (IRR = 1.12, SE = 0.05, 95% CI [1.02, 1.23]). Gender minority resilience also had a significant and positive within-person relationship with same-day alcohol use in this model (IRR = 1.25, SE = 0.10, 95% CI [1.03, 1.51]). For alcohol-related harms, the Poisson multilevel model results showed that gender minority stress had a significant and positive within-person relationship with same-day alcohol-related harms (IRR = 1.14, SE = 0.06, 95% CI [1.02, 1.28]) but there was no significant relationship between resilience and alcohol-related harms (IRR = 0.83, SE = 0.15, 95% CI [0.62, 1.12]). The negative binomial model results for coping motives showed a significant, positive within-person relationship between minority stress and coping motives (IRR = 1.06, SE = 0.01, 95% CI [1.03, 1.08]) but no significant relationship between resilience and coping motives (IRR = 0.96, SE = 0.03, 95% CI [0.91, 1.01]).

Table 2.

Direct Effects of GM Stress and Resilience on Alcohol Use, Alcohol-Related Harms, Alcohol Craving, Coping Motives, and Psychological Distress

	Alcohol use		Alcohol-related harms		Coping motives
	IRR (SE)	95% CI	IRR (SE)	95% CI	IRR (SE)	95% CI
Within person
GM stress	1.12 (0.05)*	[1.02, 1.23]	1.14 (0.06)*	[1.02, 1.28]	1.06 (0.01)***	[1.03, 1.08]
GM resilience	1.25 (0.10)*	[1.03, 1.51]	0.83 (0.15)	[0.62, 1.12]	0.96 (0.03)	[0.91, 1.01]
Weekend	1.94 (0.20)**	[1.32, 2.85]	0.90 (0.32)	[0.48, 1.66]	0.96 (0.05)	[0.87, 1.07]
Between person
GM stress	1.06 (0.08)	[0.91, 1.23]	1.03 (0.12)	[0.81, 1.31]	1.21 (0.10)	[0.99, 1.49]
GM resilience	1.09 (0.14)	[0.83, 1.44]	1.01 (0.22)	[0.66, 1.55]	0.80 (0.19)	[0.55, 1.15]
Age	1.04 (0.02)*	[1.01, 1.08]	0.96 (0.03)	[0.90, 1.02]	1.01 (0.02)	[0.97, 1.06]
Transgender man	0.82 (0.46)	[0.33, 1.99]	0.30 (0.75)	[0.07, 1.28]	1.46 (0.61)	[0.44, 4.83]
Nonbinary	0.80 (0.42)	[0.35, 1.82]	0.76 (0.61)	[0.23, 2.51]	3.08 (0.55)	[1.05, 9.04]
	Alcohol craving		Coping motives		Negative affect
	IRR (SE)	95% CI	IRR (SE)	95% CI	IRR (SE)	95% CI
Within person
GM stress	1.32 (0.06)***	[1.18, 1.47]	1.35 (0.06)***	[1.21, 1.51]	1.28 (0.03)***	[1.20, 1.36]
GM resilience	0.95 (0.09)	[0.80, 1.14]	0.95 (0.09)	[0.79, 1.14]	1.00 (0.05)	[0.92, 1.10]
Weekend	1.02 (0.10)	[0.84, 1.23]	0.97 (0.10)	[0.80, 1.19]	0.95 (0.06)	[0.86, 1.06]
Between person
GM stress	2.13 (0.34)*	[1.09, 4.16]	4.00 (0.46)*	[1.62, 9.91]	1.93 (0.19)**	[1.32, 2.82]
GM resilience	0.48 (0.43)	[0.21, 1.11]	0.41 (0.57)	[0.13, 1.25]	1.01 (0.22)	[1.65, 1.56]
Age	1.01 (0.02)	[0.97, 1.05]	1.01 (0.03)	[0.96, 1.07]	1.00 (0.01)	[0.98, 1.02]
Transgender man	1.96 (0.54)	[0.68, 5.71]	2.62 (0.74)	[0.61, 11.14]	1.32 (0.31)	[0.72, 2.43]
Nonbinary	2.83 (0.51)*	[1.05, 7.63]	5.59 (0.68)*	[1.46, 21.35]	1.60 (0.28)	[0.92, 2.78]

CI confidence interval; GM gender minority; SE standard error.

^*Significance at p < .05.

^**Significance at p < .01.

^***Significance at p < .001.

Same-Day Indirect Pathways for Daily Alcohol Use and Harms

Model fit statistics for the within-person multilevel mediation model examining the indirect effect of gender minority stress and gender minority resilience on same-day alcohol use and via coping motives indicated that a random intercepts-only model provided better fit than a random slopes model (ΔAIC and ΔBIC = 6–42). These results showed that gender minority stress indirectly predicted same-day alcohol use within person through its relationship with coping motives (ab = 0.04, 95% CI [0.02, 0.08]) but gender minority resilience did not have a significant relationship with same-day alcohol use within-person via coping motives (ab = −0.02, 95% CI [−0.06, 0.01]). For the within-person multilevel mediation model examining the indirect effect of gender minority stress and gender minority resilience on same-day alcohol-related harms, a random slopes model provided a better fit than a random intercepts-only model (ΔAIC and ΔBIC = 2–17). Gender minority stress did not predict same-day alcohol-related harms via coping motives (ab = 0.02, 95% CI [−0.01, 0.05]), nor did gender minority resilience (ab = −0.02, 95% CI [−0.11, 0.03]).

Momentary Predictors of Alcohol Craving, Drinking to Cope, and Psychological Distress

Results for the negative binomial multilevel models examining momentary relationships between gender minority stress and resilience experienced in the past 30 min and alcohol craving, coping motives, and psychological distress are reported in Table 2. Model fit statistics showed a consistent preference for random intercepts-only models (ΔAIC and ΔBIC = 1–25), with the exception of mixed findings for the model of the direct effect of momentary gender minority stress and resilience on alcohol craving. To maintain consistency across the models examining momentary relationships, results for models with fixed slopes are reported for all momentary outcomes. A zero-inflated negative binomial model showed slightly better fit than the negative binomial model for the coping motives model, but as these results were consistent across both models, results for negative binomial models are reported for parsimony.

Gender minority stress had a significant, positive within-person relationship with alcohol craving (IRR = 1.32, SE = 0.06, 95% CI [1.18, 1.47]), coping motives (IRR = 1.35, SE = 0.06, 95% CI [1.21, 1.51]), and psychological distress (IRR = 1.28, SE = 0.03, 95% CI [1.20, 1.36]). Gender minority resilience was not significantly associated with momentary within-person alcohol craving (IRR = 0.95, SE = 0.09, 95% CI [0.80, 1.14]), coping motives (IRR = 0.95, SE = 0.09, 95% CI [0.79, 1.14]), or psychological distress (IRR = 1.00, SE = 0.05, 95% CI [0.92, 1.10]). For the momentary relationships, the person-level means of gender minority stress included at Level 2 also had significant, positive associations with alcohol craving (IRR = 2.13, SE = 0.34, 95% CI [1.09, 4.16]), coping motives (IRR = 4.00, SE = 0.46, 95% CI [1.62, 9.91]), and psychological distress (IRR = 1.93, SE = 0.19, 95% CI [1.32, 2.82]).

Discussion

This study sought to fill a gap in the literature concerning the daily and momentary processes that impact alcohol outcomes for TGD adults. Data supported the feasibility and acceptability of the research approach, given the acceptable completion rates of EMA surveys, high retention of participants, and feedback from participants in the form of ratings of acceptability and feasibility that fell in favorable ranges. The study findings provide critical support of key components of the proposed adapted GMSRM for alcohol outcomes for the first time in TGD adults. As hypothesized, in moments when participants reported more gender minority stressors, they also reported increased levels of negative affect, coping motives, and alcohol craving. Furthermore, on days when participants reported more gender minority stressors, they reported increased coping motives, alcohol use, and alcohol harms. Mediation analyses suggest that these stressors increase alcohol use through coping motives. This finding diverges from a prior EMA study combining sexual and gender minority participants, which did not support a mediating role of coping motives in the link between stigma and drinking outcomes [37]. Moreover, individuals who reported more gender minority stress on average in the momentary surveys also reported greater levels of alcohol craving, coping motives, and negative affect. These associations are in line with the GMSRM that suggests that experiencing gender minority stress can increase health and mental health concerns [6, 10]. This is the first EMA study to examine these processes in TGD adults, and converging evidence across the momentary and daily levels supports the role of gender minority stress in alcohol motivation, use, and harms.

Understanding Resilience and Alcohol Outcomes

Although resilience factors were also related to alcohol outcomes, the associations were in the opposite direction than expected. We found that on days when an individual’s resilience was higher (i.e., experiencing pride in one’s identity, being part of a community who shares similar gender identity, other unspecified benefit), their alcohol use was also elevated. Despite the association with increased drinking, resilience was not associated with alcohol harms. There has been relatively little published research on resilience and substance use in TGD individuals [7, 58]. Though a prior study has also identified that social support is associated with increased substance use [59], the evidence base is inconsistent, with many studies supporting no associations between examined resilience factors and substance use.

There are several possible explanations for these unexpected findings, which warrant further investigation. First, it is possible that resilience factors that were not examined here, like familial and social support and receiving gender affirming services, could reduce the risk for alcohol outcomes. For instance, some studies in gender diverse youth have found that familial support and obtaining gender affirming services were correlated with less substance use [58]. Second, it is possible that the resilience factors we chose were confounded by unexamined factors that resulted in increased drinking. For instance, the extent to which pride and community connectedness correspond with spending time in social contexts that encourage or normalize alcohol use could explain the positive associations. Third, the resilience questions asked were broad, and could have encompassed dimensions that increase, decrease, or have no relation to alcohol use. A recent review of the role of community involvement in sexual minority men’s sexual behavior highlights that different aspects of community involvement (e.g., involvement in the community nightlife scene vs. political activism) can have unique associations with health behaviors [60]. Future research would benefit from multidimensional examinations of resilience among TGD, ideally with measures that have been previously validated.

As a whole, the study findings support that the GMSRM can be extended to explain day-level alcohol outcomes in TGD adults. With this in mind, we propose an extension of the GMSRM that incorporates alcohol-related risk processes and outcomes, which is depicted in Fig. 1. Specifically, proximal and distal gender minority stressors may increase alcohol use and harms directly. These effects could also occur indirectly via increased psychological distress, alcohol craving, and coping motives. These pathways may be buffered by specific protective or resilience factors. Future research with larger samples would be better positioned to test these interrelationships. The extent to which these processes hold true could inform future intervention efforts to support TGD adults in preventing alcohol-related harms.

Fig. 1.

Proposed adapted GMSRM for alcohol outcomes. Note: The additions to the original GMSRM [10] are indicated with bold boxes. Resilience factors could also buffer links between distal and proximal stress and between drinking risk processes and alcohol outcomes, which is not depicted. GMSRM Gender Minority Stress and Resilience Model.

Strengths and Limitations

This study had key strengths and limitations. First, a strength was the use of EMA to examine within-person associations between gender minority stress and resilience with alcohol-related outcomes and processes. Our longitudinal design and corresponding multilevel analysis enabled us to consider whether individual-level fluctuations in these experiences in a given day corresponded to changes in drinking to cope, alcohol craving, alcohol use, and harms. In addition, the analyses we conducted were focused on largely contemporaneous (rather than lagged) associations. We could not test meaningful lagged associations because we had only two randomly timed assessments per day. Therefore, the time between assessments was inconsistent and could exceed 12 hr (e.g., gap between evening prompt and next day afternoon prompt). This prevents us from establishing the temporal ordering of many of the associations examined, including the mediation analyses. Having more frequent and more closely spaced surveys would better align with lagged effects testing, and would allow for empirical tests of temporal ordering and time course of these associations of interest. Though the findings are observational and insufficient for causal conclusions, this is the first study to establish these within-person associations, which strengthens the evidence base that experiencing gender minority stress could impact alcohol use and harms directly and through intermediate risk processes, like craving and drinking to cope. Another strength of the study was our specific focus on TGD individuals, which allowed us to hone in on the impacts of gender minority stressors specifically.

We anticipated that the study was adequately powered to detect small-to-medium within-person associations. In fact, significant within-person associations were supported between gender minority stress and each outcome examined. However, we were underpowered for between-person effects, such as gender differences in the outcomes. Given that the design appears feasible and acceptable, it would be important to replicate these findings in a larger and more diverse sample. Similarly, we did not examine the buffering effect of resilience factors with gender minority stress. Furthermore, our exclusion criteria led to a relatively homogenous sample of nontreatment seeking TGD adults who drink heavily, do not use other drugs regularly (besides nicotine and cannabis), and do not self-report certain mental health concerns (i.e., bipolar, mania, psychosis). Though these exclusions helped to bolster the internal validity for this pilot study (increasing confidence that processes were alcohol-related), these exclusion criteria should be relaxed in future research to improve the generalizability of findings. We made the assumption that minority stressors and resilience would have the greatest impact on alcohol outcomes measured in close temporal proximity; however, lagged effects could be explored in future research. Last, our narrow consideration of resilience factors could have impacted our findings.

Implications and Future Directions

Considerable research still needs to be conducted to characterize and begin to address the role of gender minority stress and resilience factors in TGD adults. This study helps to establish the feasibility of using a virtual EMA study to investigate these processes in TGD adults. The virtual design had many advantages, such as facilitating recruitment from a broader pool of potential participants, reducing participant burden, and improving accessibility. The next step is to extend this study with a larger and more racially and ethnically diverse sample. Racial and ethnic diversity is particularly important as these identities can lead to minority stress experiences that have also been linked to alcohol outcomes, including in EMA and daily diary studies [61, 62]. Furthermore, a recent EMA study suggests that there may be a stronger association between stress and alcohol craving for Black compared to White adults [63]. Also, having a more racially and ethnically diverse sample would help improve the generalizability and confidence of the findings. Having a larger sample in general would allow for examining the buffering effects of resilience and subgroup differences (e.g., gender and age differences).

Conclusions

The present study observed that increased alcohol use and harms is one outcome experienced by TGD individuals due to gender minority stress. It is critical to reduce and eliminate gender minority stressors. This can be accomplished through structural interventions that dismantle transphobic and discriminatory policies and practices, which would meaningfully improve the health and well-being of TGD people. Structural interventions have been successful in reducing minority stressors experienced by sexual minority individuals [64]. While these structural changes occur, gender-inclusive interventions could help with developing effective coping strategies for ongoing stressors. These interventions could be modeled after existing interventions, like ESTEEM and EQuIP, which are transdiagnostic interventions designed to support the well-being of sexual minority individuals [65, 66]. These structural changes and individual-level supports are urgently needed to prevent and address the harmful effects of alcohol use among TGD adults.

Open Science Foundation preregistration number: https://osf.io/7wshk.

Funding

This research was supported by start-up funds provided to Sarah S. Dermody from Toronto Metropolitan University.

Compliance with Ethical Standards

Authors’ Statement of Conflict of Interest and Adherence to Ethical Standards Sarah S. Dermody, Alexandra Uhrig, Jeffrey D. Wardell, Carmina Tellez, Tara Raessi, Karla Kovacek, Trevor Hart, Christian Hendershot, and Alex Abramovich declare that they have no conflict of interest.

Authors’ Contributions Sarah S. Dermody (Conceptualization [lead], Data curation [lead], Formal analysis [supporting], Methodology [lead], Project administration [lead], Resources [lead], Supervision [lead], Writing – original draft [lead], Writing – review & editing [lead]), Alexandra Uhrig (Data curation [supporting], Formal analysis [lead], Methodology [supporting], Supervision [supporting], Writing – original draft [equal], Writing – review & editing [equal]), Jeffrey D. Wardell (Conceptualization [supporting], Formal analysis [supporting], Investigation [supporting], Methodology [supporting], Writing – review & editing [supporting]), Carmina Tellez (Data curation [supporting], Methodology [supporting], Writing – original draft [supporting], Writing – review & editing [supporting]), Tara Raessi (Data curation [supporting], Formal analysis [supporting], Visualization [supporting], Writing – original draft [supporting], Writing – review & editing [supporting]), Karla Kovacek (Data curation [supporting], Formal analysis [supporting], Methodology [supporting], Visualization [supporting], Writing – original draft [supporting], Writing – review & editing [supporting]), Trevor Hart (Conceptualization [supporting], Investigation [supporting], Methodology [supporting], Writing – review & editing [supporting]), Christian Hendershot (Conceptualization [supporting], Investigation [supporting], Methodology [supporting], Writing – review & editing [supporting]), and Alex Abramovich (Conceptualization [supporting], Data curation [supporting], Investigation [supporting], Methodology [supporting], Writing – review & editing [supporting])

Transparency Statements The study and analytic plan were preregistered at https://osf.io/7wshk. Deidentified data from this study are not yet available in a public archive. Deidentified data from this study will be made available (as allowable according to institutional ethics board standards) by emailing the corresponding author. Analytic code used to conduct the analyses presented in this study is not available in a public archive. They may be available by emailing the corresponding author. Materials used to conduct the study are not publicly available but will be provided by request by emailing the corresponding author.