An experimental laboratory-based minority stress paradigm for alcohol research among sexual minority women: A pilot study

Abstract

Background:

Sexual minority women (SMW) use alcohol at higher rates and experience greater alcohol-related harms compared to their heterosexual counterparts. Evidence from observational studies suggests minority stress (i.e., stress experienced due to marginalization in society) is an important risk factor among SMW, yet there is a lack of experimental evidence to establish the direct causal role of minority stress on their alcohol use. The current pilot study adapted the preexisting personalized guided stress induction paradigm to examine how minority stress is related to stress response (assessed via subjective measures and salivary cortisol) and mechanisms of alcohol use (craving, demand, and risky decision-making) in SMW.

Methods:

Using a within-subjects design (N = 8) cisgender SMW who endorsed high-risk drinking (≥1 heavy drinking episode in the past 30 days) completed three study visits: a script development session and two in-person imagery induction appointments (minority stress and neutral).

Results:

The paradigm significantly increased subjective stress response (g = 1.32). Data supported the feasibility, acceptability, and appropriateness of using the paradigm with SMW. While the paradigm did not significantly change scores on minority stress and alcohol outcomes measures, effect sizes for craving and minority stress outcomes were in the small-to-medium range (gs = 0.24 to 0.54).

Conclusions:

The adapted minority stress paradigm appears to be feasible and appropriate for use with SMW to induce stress in laboratory settings. Future research can use this paradigm to understand causal effects of minority stress on alcohol use and related outcomes.

Introduction

Sexual minority women (i.e., who identify as lesbian, bisexual, or another sexual identity other than heterosexual; SMW) experience alcohol use at rates three to seven times higher than heterosexual women and are more likely to report hazardous drinking (i.e., heavy episodic drinking, drinking to intoxication, and drinking-related problems) than heterosexual women (Hughes et al., 2020). Further, SMW are six and a half to seven times more likely to meet criteria for alcohol dependence (under DSM-IV criteria) and eight to eleven times more likely to report two or more negative social consequences when drinking as compared to heterosexual women (Hughes et al., 2016). Additionally, these alcohol-related disparities are often larger for SMW than sexual minority (SM) men when compared to their heterosexual counterparts (Dermody et al., 2014; Fish et al., 2017; Hughes et al., 2016; Watson et al., 2018). It is important to identify underlying mechanisms that cause elevated rates of alcohol-related harms in SMW.

Minority stress, or stressful experiences related to one’s minority status in society (Brooks, 1981; Meyer, 2003), is an underlying mechanism which may increase alcohol use in SMW. However, support for this model has primarily been in observational studies (Dermody et al., 2019; Lehavot & Simoni, 2011), indicating a need for rigorous research that directly tests the causal link between minority stress and alcohol use, such as experimental designs (Hughes et al., 2020). Recently, experiments have manipulated minority stressors with SM participants with some success (Hatzenbuehler & McLaughlin, 2014; Mereish & Miranda Jr., 2019; Seager van Dyk et al., 2023), primarily from vicarious induction methods (e.g., standardized images). These studies support the feasibility of inducing stress based on sexual identity, with Mereish and Miranda (2019) providing additional evidence that such paradigms can be used to examine substance use outcomes. However, some limitations of vicarious methodologies are 1) these paradigms tend to focus on distal stressors (i.e. external stressors such as discrimination and victimization) as opposed to proximal stressors (i.e. internal stressors such as internalized heterosexism and concealment), and 2) researchers cannot always guarantee the minority stress induction is equally salient/relevant to all participants. Thus, it is important to develop a paradigm that directly manipulates personally-experienced stress in order to address this gap in the SMW alcohol use literature.

To our knowledge, researchers have not yet leveraged experiments to test the causal effects of personalized minority stress on alcohol-related outcomes, such as alcohol craving, alcohol demand, or risky decision-making. A limiting factor is an applicable paradigm has yet to be developed or tested. A promising candidate procedure to elicit personalized minority stress is the existing and validated personalized stress imagery induction paradigm, which induces a negative mood state utilizing participants’ own recent, stressful experiences (Sinha, 2009). In participants diagnosed with cocaine, alcohol, or tobacco dependence, exposure to the stress imagery task (versus a neutral condition) increased both physiological and subjective stress response (McKee et al., 2010; Seo et al., 2013; Sinha et al., 2009; Sinha et al., 2003). Conceivably, the stress imagery task could be adapted to elicit SM stress by specifically instructing participants to recount a recent event of heterosexism-related stress, rather than a general life stressor.

An adapted stress imagery task is a promising candidate for eliciting changes in alcohol outcomes due to SM stress because the existing paradigm has already been shown to elicit stress-induced alcohol use and craving. The stress imagery paradigm compared to a neutral condition elicits increased drug (including alcohol) craving in both individuals experiencing substance dependence and healthy control participants (Seo et al., 2013; Sinha et al., 2009; Sinha et al., 2003). For individuals in alcohol treatment, the stress imagery paradigm induced greater levels of alcohol craving, use, and demand as compared to a neutral imagery condition (Owens et al., 2015; Sinha et al., 2011). Thus, it effectively elicits stress and substance use craving across a wide range of individuals. Additionally, the paradigm has already been adapted for drug and trauma cues and has been implemented across multiple contexts (Ralevski et al., 2016; Sinha, 2009); therefore, it is expected that it can be adapted to specifically elicit (and experimentally manipulate) personally-relevant minority stress. The goal of this study was to validate this adaptation and establish the first personalized minority stress induction procedure for use in experimental SMW health research.

When conducting this validation work, it is critical to consider the various ways minority stress can impact alcohol use risk. Consistent with the Research Domain Criteria (RDoC) strategy (Insel et al., 2010), it is informative to take a multidimensional understanding of alcohol risk, recognizing multiple underlying mechanisms work together to influence mental and physical outcomes. To understand the role of minority stress in SMW, it is important to consider its impact on physiological, behavioral, cognitive, and psychological alcohol-related predictors. Therefore, this study was designed to take a multidimensional approach by examining several underlying mechanisms of alcohol use, including: alcohol craving (an arousal and subjective experience-related construct); alcohol demand (a positive valence and behavior-related construct); and risky decision-making (a cognition-related construct). All three of these constructs are related to the use of alcohol (Kovacs et al., 2017; MacKillop et al., 2010a; McHugh et al., 2016). Furthermore, these constructs can be manipulated by experimental paradigms (Owens et al., 2015; Preston et al., 2007; van den Bos et al., 2009).

The Current Study

As assessments of appropriateness, acceptability, and feasibility are needed when adapting interventions and paradigms for use in specific populations, (Sekhon et al., 2017), the first aim of the current study was to assess these constructs in a sample of SMW who report high-risk drinking (e.g., one or more heavy drinking episodes in the past 30 days; Aim 1). Given there is strong face validity for the adapted paradigm, there was evidence the original paradigm is effective, can be successfully adapted, and minimal risk is typically associated with utilization of the paradigm (Curran et al., 2012; Sinha, 2009), it was hypothesized participants would be able to visualize the minority stress induction and would indicate they found the paradigm to be feasible, acceptable, and appropriate. The current study also aimed to validate the adapted personalized stress induction paradigm to specifically elicit minority stress (Aim 2). It was hypothesized exposure to minority stress using adapted guided stress imagery would lead to increased subjective distress, minority stress-related distress, and psychophysiological stress reactivity, as assessed by salivary cortisol, in SMW as compared to a neutral induction task. The study also aimed to assess the effect of minority stress on alcohol-related outcomes in SMW by comparing subjective alcohol craving, alcohol demand (i.e., the reinforcing value of alcohol, as determined by the association between cost and consumption), and risky decision-making between the minority stress and neutral conditions (Aim 3). It was hypothesized exposure to minority stress would increase subjective alcohol craving, alcohol demand, and risky decision-making as compared to the neutral imagery task.

Methods

Participants

Participants were eight SMW who 1) identified as a cisgender woman; 2) reported a SM identity; 3) drank seven or more drinks/week in the last 30 days (note: this criterion was removed in December 2022 to facilitate recruitment and only applied to four of eight eligible participants); 4) had at least one heavy drinking episode (defined as four or more drinks in a single drinking event [i.e., evening or outing]) in the last 30 days; 5) was 21 to 60 years of age; 6) reported English literacy. Data collection began March 2022 and concluded February 2023. Exclusion criteria included: 1) seeking/participating in alcohol use treatment; 2) significant medical or psychiatric conditions: bipolar disorder, suicidal ideation, schizophrenia (as determined by a structured diagnostic interview) 3) taking anticonvulsant, high blood pressure, or psychotropic medications; 3) any history of significant head trauma (loss of consciousness for five minutes or more); 4) current pregnancy. Due to the interest in validating the paradigm for use with sexual minority stress and the evidence that sex differences can be present in stress response (Kudielka & Kirschbaum, 2005), the current study chose to restrict the sample to cisgender sexual minority women.

Setting/Recruitment

Participants were recruited from Benton County, Oregon and the Portland metro area. Participants were given the option of completing their in-person appointments at one of two study sites in a similar manner. Participants who resided more than 25 miles away were provided $5 per visit to help offset transportation costs. Recruitment efforts included social media and volunteer websites (Facebook, Twitter, Craigslist, Reddit, etc.); flyers on college campuses and in the local community, including sexual and gender minority (SGM) community centers and SGM friendly locations.

Procedures

Participants typically completed all study visits in a 3-week time period (M_days = 23.13, SD = 6.38; range = [15, 36]). The study consisted of one online screener administered using Qualtrics. Eligible individuals were invited to schedule the script development appointment, and then completed two imagery induction appointments. Script development appointments were conducted in-person and virtually. Script development and the first imagery induction appointment were spaced approximately one week apart to allow for developing induction scripts. Imagery induction appointments were typically one week apart (aiming for consistent time of day across visits; M_days = 8.25, SD = 3.81; range = [3, 14]).

For the imagery induction appointments, participants were asked to abstain from use of nicotine for two hours and any other substance including alcohol for 12 hours as well as not consume food or exercise within one hour of appointment start time to increase quality of the cortisol samples provided (Adam & Kumari, 2009). If participants provided a breath alcohol reading greater than 0.0 g/210L for any in-person appointment, their visit was rescheduled.

Script Development Appointment

This visit confirmed eligibility and collected baseline demographic information. Written informed consent was obtained at the start of the session. Substance and psychiatric related inclusion/exclusion criteria were confirmed using validated interviews. After the screening process, eligible individuals continued to the script development session. The appointment was approximately two hours for eligible participants and 30 minutes for ineligible participants. Eligible participants were compensated $30; ineligible participants were compensated $10.

Imagery Script Development.

Scripts for the imagery conditions were developed using the personalized stress induction method that has been validated for stress and neutral cues (Sinha, 2009; Sinha & Tuit, 2012). Using the validated procedure, Minority Stress imagery scripts were created by asking participants to recall, in detail, a recent stressful experience that occurred within the past six months, which was “very stressful,” and related to their sexual identity. Participants would begin first by briefly describing their chosen situation. The current study only included situations participants rated seven or higher on a 10-point Likert response format where “1 = not at all stressful” and “10 = the most stressed one felt recently in their life,” consistent with the protocol (Sinha, 2009; Sinha & Tuit, 2012). If participants were unable to recall a recent stressor that was seven or higher, an older stressor was used. This only impacted one participant, who provided a stressor that occurred more than six months ago. Also consistent with the protocol, the Neutral condition script was developed from the participant’s description of a neutral/relaxing situation. Participants were instructed to provide scenarios low in arousal (either positive or negative). Once appropriate neutral and minority stress scenarios were identified, participants were queried regarding detailed physical descriptions of these events, as well as sensory (e.g., sights, sounds, smells) and physiological sensations (e.g., racing heart, clenched fists).

Personalized scripts for each condition were then developed in the second person utilizing the captured sensory details. The personalized scripts were audio recorded by a separate team member using a microphone which ensured high playback quality for presentation during the induction sessions. Each script audio recording was approximately five minutes long.

Imagery Induction Appointments

Each participant completed two one-hour induction appointments on different days: Minority Stress and Neutral in counterbalanced order (to control for carry-over effects). Appointments were conducted by undergraduate research assistants who were blinded to the condition. Each appointment began with a brief baseline assessment including questionnaires concerning mood, stress levels, alcohol craving, and provided baseline cortisol and breath alcohol content (BrAC) measurements.

Then, participants completed imagery training (only at the first appointment) and were trained in progressive muscle relaxation techniques (at both visits) in order to aid in visualization (Sinha, 2009; Sinha & Tuit, 2012). Then, participants underwent the personalized imagery induction (Minority Stress or Neutral). For the induction, participants heard a recording of their personalized induction through noise cancelling headphones (Sinha, 2009; Sinha & Tuit, 2012). Following the induction, participants completed measures including subjective alcohol craving, alcohol demand, risky decision-making, and stress. A second cortisol sample was also provided 10 minutes after the induction. At the end of each appointment, participants could choose to complete a guided relaxation period. Research assistants were also trained to monitor each participant to determine if further intervention was needed to reduce stress response. At the end of the second imagery appointment, participants were debriefed and provided ratings on feasibility, acceptability, and appropriateness measures. Participants were compensated $15 for each imagery induction visit, and a $10 bonus for completing both visits. Participants who resided more than 25 miles away were provided $5 per visit to help offset transportation costs.

Measures

Screening Measures

The Diagnostic Assessment Research Tool (DART), a structured interview, assessed presence of SUDs or severe psychiatric disorders (Schneider et al., 2021). Individuals that met criteria for bipolar disorder, posttraumatic or acute stress disorder, any substance disorder excluding alcohol (current), or psychosis in the past six months were excluded from the study and provided a resource and referral packet for mental health services. The Ask Suicide-Screening Questions (ASQ), a 5-item brief questionnaire, was used to assess the presence of suicidality (Horowitz et al., 2012). Participants that answered yes to any of the first four questions were ineligible to participate and were provided a resource and referral packet.

Descriptive Measures

The Timeline Followback (TLFB) method assessed substance use (alcohol, cigarette, cannabis, and illicit substances) in the past 30 days or since last appointment (Sobell & Sobell, 1993). The self-report version of the Alcohol Use Disorder Identification Test (AUDIT), a 10-item Likert response format survey, was assessed alcohol-related problems in the past year (Saunders et al., 1993).

Demographic questions assessed age, sexual identity, race, and ethnicity. Individuals were allowed to select “all that apply” for the race item. Ethnicity was assessed as Hispanic vs not.

Feasibility and Acceptability Measures

Scripts were provided stress and relaxation ratings by two independent team members (Sinha & Tuit, 2012). Participants used a visual analog scale (VAS; ranging from 1–10) to rate vividness of the imagery as a manipulation check and to control for possible sources of variation across both conditions (Sinha & Tuit, 2012). Three brief 4-item measures, the Acceptability of Intervention Measure (AIM, α = .84), Intervention Appropriateness Measure (IAM, α = .64), and Feasibility of Intervention Measure (FIM, α = .84), were administered (Weiner et al., 2017). Participants were instructed: “…We are interested in knowing your feelings and thoughts related to the stress induction you experienced, and how you feel about it being used in research.” Items were also adapted to say “guided stress induction” in place of the word “intervention”.

Subjective Outcome Measures

The following measures were all used at imagery induction appointments, with some also administered at the script development session. For Aim 2, subjective stress was assessed via a single item VAS (Sinha et al., 2011; Sinha & Tuit, 2012). Minority stress that could potentially be manipulated by a minority stress induction paradigm was also assessed via the IHP-R (internalized heterosexism; script development, α = .76), a 5 item Likert response format scale specifically adapted for use in SMW (Herek et al., 2009) and the 12-item Gay-Related Rejection Sensitivity Scale (GRSS; script development, α = .88) adapted for use in SMW (Feinstein et al., 2012; Pachankis et al., 2008).

For Aim 3, the 8-item Alcohol Urge Questionnaire (AUQ; appointment 1, α = .97; appointment 2, α = .95) and a single-item VAS (following protocol developed by Sinha and Tuit, 2012) were used to assess alcohol craving (Bohn et al., 1995).

Alcohol demand observed outcomes of intensity, O_max, P_max, breakpoint, and elasticity were assessed using the state version of Alcohol Purchase Task (APT; MacKillop et al., 2010b). Participants were asked to report how many drinks they would consume when drinks were free, $0.02, $0.05, $0.13, $0.25, $0.50, and $1–15. Intensity is defined as the number of drinks consumed when the cost is zero; breakpoint is the first increment of cost when no alcohol is consumed; O_max is peak expenditure for alcohol; P_max is the price associated with peak expenditure; elasticity is the rate of change in consumption with changes in price, or the slope of the demand curve. The novel R package beezdemand cleaned APT data using the Stein et al (2015) criteria for identifying nonsystematic responding and recoding outliers, and to obtain observed and derived APT parameters (Kaplan et al., 2019; MacKillop et al., 2010a). No nonsystematic data was identified. One participant who reported they would drink one or more drinks at the highest price increment (i.e., recording no breakpoint) was assigned a breakpoint at the highest price ($15) (Murphy & MacKillop, 2006).

VAS Stress, VAS alcohol, the IHP-R, and GRSS were all administered pre- and post-induction at both imagery appointments.

Cognitive Outcome Measures

The Iowa Gambling Task (IGT) was used to assess risky decision-making (Buelow & Suhr, 2009). The IGT is a cognitive task used to reflect real world decision-making and identify decision-making deficits, as it is sensitive to prefrontal cortex functioning. Participants completed a computerized task in which they were told they start with a $2000 pot of money and need to maximize their profit over 100 trials by selecting cards from one of four decks (Buelow & Suhr, 2009). The following outcomes were examined: total money won and net IGT score (the difference between total advantageous and disadvantageous selections).

Salivary Cortisol

Salivary cortisol was collected as a biological measure of stress response (Buchanan et al., 1999). To account for within-person variability in change in diurnal cortisol levels throughout the day (Edwards et al., 2001), changes in cortisol levels were examined by taking two passive drool salivary cortisol samples, one 10 minutes after the start of each imagery appointment, and one 10 minutes after each imagery induction. Imagery induction appointments were scheduled at approximately the same time of day for each participant. After data collection was completed, cortisol samples were shipped to a biomedical lab for analysis.

Statistical Analyses

Descriptive Analyses

Descriptive information is provided for the following baseline measures and assessments: age, race/ethnicity, sexual identity, AUDIT, TLFB, and minority stress measures.

Aim 1 - Feasibility and Acceptability Analyses

Feasibility was determined in part by examining mean score and corresponding 95% confidence interval (CI) of self-reported imagery vividness (VAS-Imagery item) for each induction. CI including seven or higher (out of 10) indicated individuals successfully engaged with the paradigm. Independent ratings on two items assessing scripts’ level of stress and relaxation were examined. Intra-class correlation coefficients for the items are provided. If 95% CIs are above values of 0.75, it was determined scripts reliably contain the intended emotional connotation (Bobak et al., 2018). For the FIM, AIM, and IAM, mean scores and corresponding 95% CIs for each summed and averaged scale score. If the CIs included four or higher (out of 5; indicating agreement to items), this helped indicate feasibility and acceptability.

Aim 2 – Stress Response

To test if minority stress induction increases stress, levels of stress, minority stress, and salivary cortisol were compared in a series of one-way repeated measures ANOVA (analysis of variance) models using the “MIXED” command and diagonal covariance structure in SPSS 28 (IBM Corp, 2021). Induction method (Minority stress and Neutral) was the repeated measures factor. Analyses examined changes from pre to post imagery induction using difference scores for the following outcomes: VAS-Stress item, IHP-R, GRSS, and salivary cortisol. Significant differences between study conditions on the difference scores of stress outcomes indicate feasibility of the paradigm to induce stress. The following covariates were included in each model: order of condition (to account for expectancy effects; time-invariant covariate), vividness of the imagery (VAS Imagery; time varying), and study appointment (Imagery appointment 1 and 2; time-varying [to account for practice effects]).

Aim 3 – Mechanisms of Alcohol Use

To test if minority stress induction increased alcohol-related outcomes, alcohol craving (VAS-Alcohol and AUQ), alcohol demand (APT; intensity, elasticity, O_max, P_max, and breakpoint), and IGT (total money won and net IGT score) were compared in a series of repeated measures ANOVAs where the repeated factor was induction method (Minority stress and Neutral). Analyses examined how induction method related to changes from pre- to post-imagery induction using difference scores for VAS-Alcohol craving. The other alcohol outcomes were only measured and therefore analyzed post-imagery induction. The following covariates were included in each model: order of condition (time invariant covariate), vividness of the imagery (time-varying), and study appointment (time-varying).

Models’ assumptions, including outliers and outcome distributions, were tested. As there was only one repeated measure with two factors, sphericity could not be tested and is assumed. No outliers were identified (determined as any values more than ±3 standard deviations from the mean). Even so, an outlier was clearly present for elasticity, as one value was >10, compared to all other values in a 0–1 range. Findings did not change when the outlier was removed, and thus the outlier was included in the model. This value was present due to one participant reporting very low alcohol consumption across the task (i.e., one drink when drinks were free and zero drinks the remainder of the task). Even though the beezdemand package calculated a value for this participant, elasticity is intended to be reflective of data with at least two non-zero values. Thus, elasticity findings should be interpreted with caution. The assumption of normality was tested using Shapiro-Wilk’s test (Mishra et al., 2019). Assumptions of normality were violated for changes in IHP-R scores, AUQ scores, and APT elasticity scores for the neutral condition as well as APT Omax scores for the stress condition. An exponential transformation with base e was chosen for changes in IHP-R scores given the negative skew of the variable and negative values, as logarithmic and square root transformations are not possible with negative values, (IBM Corp, 2020). A logarithmic transformation with base 10 was used for AUQ, APT Omax, and APT elasticity scores given variables were positively skewed. As untransformed APT Omax scores contained a zero value, a constant of 1 was used to facilitate logarithmic transformation. Transformations did not address assumptions for IHP-R and APT elasticity scores. Therefore, the untransformed variable results are reported. Repeated measures ANOVAs are relatively robust to normality violations (Blanca et al., 2023); even so, findings related to IHP-R and elasticity scores should be interpreted with caution.

Fixed parameter estimates, covariance estimates, and estimated marginal means across conditions are reported for all models. Given the small sample and pilot nature of the study, many of the analyses are underpowered to detect significant effects using significance testing. Therefore, effect sizes are also used to aid in interpretation of findings. Marginal means and correlation between conditions were also used to calculate Hedge’s g repeated measure effect sizes, and the cutoffs of 0.2, 0.5, and 0.8 were used to determine small, medium, and large effect sizes, respectively (Lakens, 2013).

Results

Of the 295 individuals who completed the online screening survey, 42 individuals were eligible and invited to schedule a script development appointment. Of the eighteen participants who scheduled and attended a script development appointment, 11 were determined to be eligible (see Figure 1 for detailed study flow information). When data collection ended, eight participants had completed all study visits and are included in analyses. The final sample (N = 8) included SMW who identified as pansexual (n = 4), bisexual (n = 3), and lesbian/queer (n =1); was mostly White (n = 6), with two participants identifying as Multiracial; and consisted mostly of young adults (M_age = 25.25, SD = 5.18). The sample reported high levels of alcohol use (e.g., n = 7 [87.5%] met criteria for high-risk drinking, as determined by the AUDIT). On average, participants reported 5.88 (SD = 3.27) heavy drinking episodes in the past 30 days, as determined by the TLFB. Use of cannabis was common, with 62.5% of participants reporting cannabis use in the past month. While measures of minority stress collected at baseline are not norm scored, participants on average scored in the lower range of values for each scale, indicating the sample may have lower levels of minority stress. Table 1 includes descriptive and demographic data for the final sample. No participants reported e-cigarette use nor drug use outside of marijuana use.

Figure 1.

Flowchart of study procedures, participant enrollment, and reasons for exclusion. ^a participants were excluded for multiple reasons, and thus totals from individuals inclusion/exclusion criteria may not equal the number of participants excluded, ^b Standard drinks per week was removed as an inclusion criteria in December 2022, ^c These two participants could not be scheduled within the allotted timeframe for this pilot study due to administrative reasons and do not reflect true drop out.

Table 1.

Demographic information for final study sample.

	N (%)
Total	8 (100%)
Sexual Identity
Lesbian/Queer	1 (12.50%)
Bisexual	3 (37.50%)
Pansexual	4 (50.00%)
Racea
White	8 (100%)
Black	0 (0%)
Asian	1 (12.50%)
American Indian/Alaskan Native	0 (0%)
Hispanica	1 (12.50%)
Cigarette/Tobacco Use	1 (12.50%)
Marijuana Use	5 (62.50%)
	M(SD)	Range
Age (in years)	25.25 (5.18)	[21, 32]
AUDITb	9.13 (2.42)	[4, 12]; out of 40
Ave. past 30 day drinks/day	1.67 (0.86)	[0.40, 2.77]
Ave. past 30 day total number of drinks	50.13 (25.79)	[12, 83]
Ave. number past 30 day heavy drinking episodes	5.88 (3.27)	[1, 11]
IHP-R	1.45 (0.54)	[1, 2.60]; out of 5
GRSS	10.44 (4.69)	[3.33, 17.67], out of 36

Note:

^aParticipants were invited to “select all that apply”, thus % may add up to greater than 100%

^bn= 7 participants received AUDIT scores higher than 8, indicating high-risk drinking; N = number, M = mean, SD = standard deviation, AUDIT = Alcohol Use Disorder Identification Test, Ave. = average IHP-R = Internalized Homophobia scale, GRSS = Gay Related Rejection Sensitivity Scale

Aim 1 - Feasibility and Acceptability of the Personalized Minority Stress Induction

Intra-class correlations for stress and relaxation ratings of all scripts were examined. High levels of reliability were observed for stress level (r = 0.95, 95% CI: [0.85, 0.98]) and relaxation level (r = 0.95, 95% CI: [0.86, 0.98]). Average scores on the vividness VAS item were M = 8.75, SD = 1.04 (95% CI: [8.033, 9.467]), suggesting participants were able to successfully visualize the stress scenario.

For participants’ attitudes of acceptability, appropriateness, and feasibility toward the minority stress induction’s use, average scores were: AIM (M = 4.12, SD = 0.75, 95% CI: [3.67, 4.71]); IAM (M = 4.38, SD = 0.53, 95% CI: [4.01, 4.75]); FIM (M = 4.66, SD = 0.44, 95% CI: [4.35, 4.96]). Therefore, participants generally found the paradigm to be acceptable, appropriate, and feasible for use. None of the participants reported lingering feelings of stress when asked, nor did they utilize the stress reduction exercise, also indicating the paradigm was well-tolerated.

Aim 2 – Minority Stress and Stress Response

While 50% of scripts consisted of discrimination events (e.g., discrimination), 50% consisted of proximal stressors (e.g., expectations of rejection). The most common proximal stressor was expectations of rejection (n = 3). Repeated measures ANOVAs were conducted to examine stress response (change in VAS Stress, IHP-R, GRSS, and salivary cortisol) across the Minority Stress and Neutral conditions. While VAS Imagery was proposed as a covariate, results indicated it did not significantly relate to outcomes of interest and was subsequently removed from all models for the sake of parsimony. Order and appointment were kept covariates to address feasibility of the within-person design.

Results indicated the Minority Stress induction was significantly associated with greater increases in VAS Stress scores from pre- to post-induction compared to the Neutral induction (b = 5.38, SE = 1.02, p < .001). This was a large effect (g = 1.32, 95% CI: [−0.24, 2.88]), and 100% of participants displayed higher difference scores in the minority stress condition as compared to the neutral condition (albeit, to varying degrees; Figure 2A). The Minority Stress induction was not significantly associated with changes in IHP-R scores, GRSS scores, or salivary cortisol. Using parallel plots, it was observed that 62.5% of IHP-R, 50% of GRSS, and 12.5% of salivary cortisol difference scores were higher in the minority stress condition as compared to the neutral condition (Figures 2B, C, and D). Additionally, examination of effect sizes indicated small effects for changes in GRSS scores (g = 0.45, 95% CI: [−0.55, 1.45]), changes in cortisol (but in reverse of the expected direction; g = −0.43, 95% CI; [−1.44, 0.58]), and changes in IHP-R scores (g = 0.24, 95% CI: [−0.75, 1.22]). Order and appointment were not significant covariates. Full parameter estimates for models of stress outcomes are reported in Table 2.

Figure 2.

Parallel plots of stress outcomes change in VAS Stress (A), IHP-R (B), GRSS (C), and salivary cortisol (D) across Minority Stress and Neutral conditions (without controlling for covariates). All stress outcomes were collected both pre and post induction. Difference scores were calculated by subtracting pre-induction scores from post-induction scores (i.e., post – pre). Note: Changes in VAS Stress scores significantly differed across conditions.

Table 2.

Repeated measures ANOVA model parameters for stress outcomes across conditions.

	Change in VAS Stress			Change in IHP-R Scores
	Estimate	SE	95% CI	Estimate	SE	95% CI
Fixed Effects
Minority Stress Inductiona	5.38 ***	1.02	[3.02, 7.73]	0.15	0.16	[−0.20, 0.50]
Order of Conditionb	−2.13~	1.02	[−4.48, 0.23]	0.20	0.16	[−0.15, 0.55]
Imagery Appointmentc	−0.38	1.02	[−2.73, 1.98]	0.10	0.16	[−0.25, 0.45]
Covariance
Minority Stress Induction	7.13	4.11	--	0.11	0.07	--
Neutral Induction	1.17	0.67	--	0.09	0.05	--
Marginal Means
Minority Stress Induction	3.13	0.94	[0.82, 5.43]	−0.05	0.12	[−0.34, 0.24]
Neutral Induction	−3.18	0.38	[−3.18, −1.31]	−0.20	0.11	[−0.46, 0.06]
	Changes in GRSS Scores			Changes in Salivary Cortisol
	Estimate	SE	95% CI	Estimate	SE	95% CI
Fixed Effects
Minority Stress Inductiona	1.15	0.64	[−0.26, 2.55]	−0.06	0.03	[−0.13, 0.2]
Order of Conditionb	−0.79	0.64	[−2.20, 0.61]	0.01	0.03	[−0.07, 0.08]
Imagery Appointmentc	1.00	0.64	[−0.40, 2.40]	−0.03	0.03	[−0.10, 0.04]
Covariance
Minority Stress Induction	1.22	0.71	--	0.01	0.003	--
Neutral Induction	2.05	1.19	--	0.003	0.002	--
Marginal Means
Minority Stress Induction	0.54	0.39	[−0.42, 1.50]	−0.06	0.03	[−0.12, 0.01]
Neutral Induction	−0.60	0.51	[−1.84, 0.64]	0.003	0.02	[−0.05, 0.05]

Note:

^aref. = Neutral induction

^bref. = stress condition first

^cref. = imagery appointment 2; Findings significant at the p < .05 level are bolded; VAS = Visual Analog Scale, IHP-R = Internalized Homophobia Scale, GRSS = Gay Related Rejection Sensitivity Scale, SE = Standard error, CI = confidence Interval

^~p < .10

^*p< .05

^**p < .01

^***p < .001

Aim 3 – Minority Stress and Mechanisms of Alcohol Use

As in the stress models, VAS Imagery scores across conditions did not significantly predict outcomes and were thus removed from all models as a covariate. Minority Stress induction was not significantly associated with any examined mechanisms of alcohol use. However, a medium effect was observed for changes in VAS Alcohol craving (g = 0.54, 95% CI: [−0.53, 1.61]), indicating the Minority Stress induction was related to greater increases in craving compared to the Neutral induction. Small effects were observed for AUQ scores (g = 0.36, 95% CI: [−0.64, 1.35]), APT intensity (g = 0.28, 95% CI: [−0.72, 1.28]), and APT P_max (g = 0.46, 95% CI: [−0.55, 1.46]). Using parallel plots, it was observed that 62.5% of participants displayed higher VAS craving difference scores in the minority stress condition as compared to the neutral condition and 87.5% displayed higher AUQ scores (Figures 3A and B). For demand outcomes, it was observed using parallel plots that 50% of intensity, 37.5% of breakpoint, 50% of Omax, and 62.5% of Pmax scores were higher in the minority stress condition as compared to the neutral condition (Figures 4A, B, C, and D). For IGT outcomes, parallel plots indicated that 50% of participants displayed higher IGT net scores in the minority stress condition as compared to the neutral condition and 50% displayed higher IGT total money won scores (Figures 5A and B). Experiencing the Neutral induction at the first imagery appointment was significantly associated with lower APT P_max (b = −4.38, SE = 1.30, p = .006), indicating these participants were spending more on alcohol at lower price points. The first imagery appointment was significantly associated with lower Net IGT scores (b = −21.50, SE = 9.58, p = .045), indicating practice effects were present. Full parameter estimates for models of alcohol outcomes are reported in Table 3 and effect sizes for stress and alcohol outcomes are reported in Table 4.

Figure 3.

Parallel plots of craving outcomes change in VAS Alcohol Craving (A) and log-transformed AUQ scores (B), across Minority Stress and Neutral conditions (without controlling for covariates). VAS Alcohol Craving ratings were collected both pre and post induction. A difference score was calculated by subtracting pre-induction values from post-induction values (i.e., post – pre).

Figure 4.

Parallel plots of alcohol demand outcomes Intensity (A), Breakpoint (B), log-transformed O_max (C), P_max (D), and elasticity (E), across Minority Stress and Neutral conditions (without controlling for covariates). One participant reporting an extreme value was excluded from the elasticity figure to allow for a more accurate visualization of the potential changes observed across conditions.

Figure 5.

Parallel plots of risky decision-making outcomes IGT Net Score (A) and IGT Total Money Won (B) across Minority Stress and Neutral conditions (without controlling for covariates).

Table 3.

Repeated measures ANOVA model parameters for alcohol craving, alcohol demand, and risky decision making outcomes across conditions.

	Change in VAS Alcohol Craving			AUQ Scoresd
	Estimate	SE	95% CI	Estimate	SE	95% CI
Fixed Effects
Minority Stress Inductiona	2.50~	1.16	[−1.44, 5.14]	0.18	0.13	[−0.10, 0.47]
Order of Conditionb	<.001	1.16	[−2.64, 2.64]	−0.08	0.13	[−0.36, 0.20]
Imagery Appointmentc	−0.25	1.16	[−2.89, 2.39]	0.04	0.13	[−0.24, 0.32]
Covariance
Minority Stress Induction	8.79	5.08	--	0.07	0.04	--
Neutral Induction	1.96	1.13	--	0.06	0.04	--
Marginal Means
Minority Stress Induction	1.88	1.05	[−0.69, 4.44]	0.44	0.10	[0.20, 0.67]
Neutral Induction	−0.63	0.50	[−1.84, 0.59]	0.25	0.09	[0.04, 0.47]
	APT Intensity			APT Breakpoint
	Estimate	SE	95% CI	Estimate	SE	95% CI
Fixed Effects
Minority Stress Inductiona	1.63	1.44	[−1.58, 4.83]	1.37	1.97	[−3.04, 5.78]
Order of Conditionb	1.88	1.44	[−1.33, 5.08]	−2.62	1.97	[−7.03, 1.79]
Imagery Appointmentc	0.13	1.44	[−3.08, 3.33]	0.88	1.97	[−3.53, 5.29]
Covariance
Minority Stress Induction	11.96	6.90	--	8.25	4.76	--
Neutral Induction	4.58	2.65	--	22.91	13.23	--
Marginal Means
Minority Stress Induction	5.38	1.22	[2.38, 8.37]	10.00	1.02	[7.52, 12.49]
Neutral Induction	3.75	0.76	[1.90, 5.60]	8.63	1.69	[4.49, 12.77]
	APT Omaxe			APT Pmax
	Estimate	SE	95% CI	Estimate	SE	95% CI
Fixed Effects
Minority Stress Inductiona	0.07	0.21	[−0.39, 0.54]	2.38~	1.30	[−0.45, 5.20]
Order of Conditionb	0.02	0.21	[−0.45, 0.49]	−4.38 **	1.30	[−7.20, −1.55]
Imagery Appointmentc	−0.4	0.21	[−0.51, 0.43]	1.63	1.30	[−1.20, 4.45]
Covariance
Minority Stress Induction	0.07	0.04	--	6.33	3.66	--
Neutral Induction	0.27	0.16	--	7.13	4.11	--
Marginal Means
Minority Stress Induction	1.10	0.09	[0.86, 1.33]	8.00	0.89	[5.82, 10.18]
Neutral Induction	1.02	0.18	[0.57, 1.47]	5.63	0.94	[3.31, 7.93]
	APT Elasticity			IGT Net Total
	Estimate	SE	95% CI	Estimate	SE	95% CI
Fixed Effects
Minority Stress Inductiona	−1.32	1.32	[−4.55, 1.92]	6.00	9.58	[−14.89, 26.89]
Order of Conditionb	1.32	1.32	[−1.92, 4.55]	−16.50	9.58	[−37.39, 4.39]
Imagery Appointmentc	1.31	1.32	[−1.92, 455]	−21.50 *	9.58	[−42.39, −0.61]
Covariance
Minority Stress Induction	<0.001	<0.001	--	394.33	227.67	--
Neutral Induction	14.00	8.06	--	340.00	196.30	--
Marginal Means
Minority Stress Induction	0.02	0.01	[0.003, 0.03]	29.00	7.02	[11.82, 46.18]
Neutral Induction	1.34	1.32	[−1.90, 4.57]	23.00	6.52	[7.05, 38.95]
	IGT Total Money Won
	Estimate	SE	95% CI
Fixed Effects
Minority Stress Inductiona	50.63	486.82	[−1,012.02, 1,113.27]
Order of Conditionb	−435.63	486.82	[−1,498.27, 627.02]
Imagery Appointmentc	−965.63~	486.82	[−2,028.27, 97.02]
Covariance
Minority Stress Induction	69,983.33	617,755.17	--
Neutral Induction	825,986.46	476,883.50	--
Marginal Means
Minority Stress Induction	130.00	365.72	[−764.87, 1,024.87]
Neutral Induction	79.38	321.32	[−706.87, 865.62]

Note:

^aref. = Neutral induction

^bref. = stress condition first

^cref. = imagery appointment 2

^dAUQ scores log-transformed to account for non-normality

^eO_max scores log-transformed to account for non-normality; Findings significant at the p < .05 level are bolded; VAS = Visual Analog Scale, AUQ = Alcohol Urge Questionnaire, APT = Alcohol Purchase Task, IGT = Iowa Gambling Task, SE = Standard error, CI = confidence Interval

^~p < .10

^*p< .05

^**p < .01

^***p < .001

Table 4.

Hedge’s g effect sizes for differences across conditions on all outcomes of interest.

	Hedge’s g	95% CI
Stress Outcomes
VAS Stress	1.32	[−0.24, 2.88]
IHP-R	0.24	[−0.75, 1.22]
GRSS	0.45	[−0.55, 1.45]
Salivary Cortisol	−0.43	[−1.44, 0.58]
Craving Outcomes
VAS Alcohol	0.54	[−0.53, 1.61]
AUQ	0.36	[−0.64, 1.35]
APT Outcomes
Intensity	0.28	[−0.72, 1.28]
Breakpoint	0.17	[−0.81, 1.16]
Omax	0.09	[−0.89, 1.07]
Pmax	0.46	[−0.55, 1.46]
Elasticity	−0.71	[−1.72, 0.31]
Risky Decision Making
IGT Net Total	0.16	[−0.83, 1.14]
IGT Total Money Won	0.03	[−0.95, 1.01]

Note: Effect sizes calculated as Neutral (group 1) vs. Minority Stress (group 2); Findings significant at the p < .05 level are bolded; Hedge’s g cutoffs are ≥ .2 = small effect, ≥ .5 = medium effect, ≥ .8 = large effect; VAS = visual analog scale, IHP-R = Internalized Homophobia scale, GRSS = Gay Related Rejection Sensitivity Scale, AUQ = Alcohol Urge Questionnaire, APT = Alcohol Purchase Task, IGT = Iowa Gambling Task

Post Hoc Analyses

To further assess feasibility of the paradigm, an additional post hoc test was conducted to determine if changes in VAS Stress scores were related to changes in alcohol craving. The MIXED command was used, in which condition was the repeated factor and change in VAS Stress was included in the model as a time-varying predictor. Findings indicated greater changes in VAS Stress scores were related to greater changes in VAS Alcohol Craving scores (F(1, 12.8) = 5.49, b = 0.49, SE = 0.21, p = .04).

Discussion

To our knowledge, this is the first study to adapt and test an experimental personalized minority stress induction paradigm. As minority stress can potentially be experienced in unique ways compared to other forms of stressors (Meyer, 2003), it was important to first test this paradigm in a small sample to determine safety of use and other feasibility measures. Findings indicated the personalized minority stress induction paradigm is feasible, acceptable, and appropriate for use in experimental minority stress research with SMW who may experience lower levels of minority stress. This was confirmed through a variety of assessments: independent ratings of scripts, VAS Imagery scores, and participant ratings. Other aspects of the study further provided evidence of the paradigm’s feasibility. Namely, the study had very little drop out (n = 1), indicating study procedures were not too burdensome. Furthermore, participants did not report lingering feelings of stress, indicating the paradigm was well-tolerated. These findings support the paradigm can be safely used in future experimental research with cisgender SMW.

Further, this paradigm effectively elicited a large increase in stress compared to the neutral condition. Therefore, this induction can be added to the growing repertoire of effective stress induction paradigms for minority stress, such as those supported by Mereish and Miranda Jr. (2019) and Seager van Dyk et al. (2023), in which image-, film- and media-based minority stress induction paradigms elicited negative emotions in SM participants. This further indicates the personalized minority stress can be used in experimental research to elicit minority stress, bolstering feasibility findings. It is worth noting there was variability in the amount of subjective stress response to the paradigm, with some participants reporting very little change in stress scores post-induction (Figure 2A). Future work should consider explaining these sources of variability, which could include characteristics of the stressors used (e.g., proximal vs. distal stressor used) as well as individual differences (e.g., resilience factors and coping resources). Further, it is important to note, while not significant, a trend was present such that participants who experienced the neutral induction first reported smaller changes in VAS Stress scores. This could indicate the presence of expectancy effects, and highlights the importance of counterbalancing conditions.

The induction did not correspond with significant changes in indexes of minority stress. Unexpected, mean changes in internalized heterosexism scores across conditions indicated that scores decreased across both conditions, with greater decreases in the Neutral condition (M_{minority stress} = −0.05, M_neutral = −0.20). These decreases were minor, however, and may suggest the IHP-R scale is less resistant to change. In general, there is a paucity of minority stress measures specifically designed to asses state experiences of minority stress. Future work would benefit from the development of more diverse minority stress scales, for use in a variety of research contexts.

When examining alcohol outcomes, parallel plots suggested the minority of participants reported higher VAS Alcohol difference and AUQ scores in the Minority Stress condition as compared to the Neutral condition. These findings are in line with those observed in Mereish and Miranda Jr. (2019), and support the personalized minority stress induction paradigm’s use in future alcohol research. However, findings were less clear regarding alcohol demand and risky decision-making. Parallel plots suggested the majority of participants reported increase in P_max for the Minority Stress condition compared to the Neutral condition, but not for intensity, breakpoint, and O_max. This means that after exposure to minority stress, participants reported greater peak expenditure at higher price points (P_max) compared to the neutral induction. In contrast, the personalized minority stress induction paradigm did not impact risky decision-making, as determined by IGT outcomes. However, participants tended to score lower and win less money at their first visit, suggesting practice effects were present. This finding has also been observed in other studies (Aram et al., 2019), suggesting the IGT is likely not suitable for within-subjects designs with multiple study visits in a short period. Future work should seek to use other measures of decision-making that have been implicated in alcohol use.

Limitations and Future Directions

The current study had several limitations. First, the final sample included eight participants, resulting in underpowered inferential tests. It is possible that models for craving and demand outcomes could be statistically and clinically significant in a larger sample, however the current sample did not allow for sufficient testing of these outcomes. While pilot studies typically include smaller samples and the findings provide evidence of the paradigm’s safety and feasibility, effects need to be examined in a larger sample. Future work should also consider reducing the study enrollment requirements to increase generalizability. Participants were excluded for a variety of medical and psychological reasons, partly to guarantee participant safety. However, given the well-established relationship between minority stress and health outcomes (Meyer, 2003) this may have excluded individuals for which minority stress is especially salient (and could explain the lower levels of minority stress reported in the sample). Future studies should consider the evaluation of concurrent mental health conditions such as depression or anxiety within the present paradigm. Additionally, findings from the current study cannot necessarily be applied sexual minority individuals with intersecting minoritized identities (e.g., diverse gender and racial/ethnic identities), as it is likely that potency of effects may differ. In order to utilize the paradigm in a larger, more diverse sample, considerations do need to be made for how to conduct research using the paradigm in a way that maximizes study benefits while reducing risk. Community engagement would greatly improve experimental research using the minority stress paradigm going forward, as it would address institutional concerns and improve the overall study design.

Another potential limitation in the current study is possible effects of regional differences in attitudes towards SGM identity. Compared to many other U.S. regions and states, the Pacific Northwest has significantly more policies protecting and supporting SGM individuals (Movement Advancement Project, 2020). This could potentially explain the high prevalence of proximal stressors endorsed in the scrips developed for the inductions, as individuals living in states with greater equity likely experience less discrimination and victimization events. It is possible stress response to experimental minority stress induction may be more severe in less affirming states, in large part due to higher chronic levels of minority stress. Therefore, in addition to testing the validity of the paradigm in diverse locations, researchers should also consider how local sociocultural attitudes might impact overall safety of the paradigm’s use. (This is another avenue in which community engagement would benefit minority stress induction research.)

Last, while the current study design can speak to the causal effect of minority stress on mechanism of alcohol use, this does not always directly correlate to actual use. Future work with the personalized minority stress induction paradigm should consider self-administration methods to observe the role of minority stress in alcohol consumption. While self-administration would require additional ethical oversight, findings from this research could help inform intervention development and public health policy.

Implications

The personalized minority stress induction provides an additional resource for researchers wanting to conduct experimental minority stress research beyond those already developed (Mereish & Miranda Jr., 2019; Seager van Dyk et al., 2023). While all the minority stress paradigms were developed using minority stress experiences, the personalized minority stress induction paradigm includes situations each participant personally experienced, ensuring the minority stress is salient and relevant to the individual. Furthermore, paradigms developed using media sources likely over-represent distal minority stressors, making it difficult to examine the role of proximal stressors in health outcomes. In the current study, 50% of stressors were proximal, suggesting the personalized minority stress induction paradigm targets unique minority stress experiences.

The personalized minority stress induction paradigm can aid in the development of more complex models explaining alcohol use in SMW, but could also be applied more broadly. Namely, the personalized minority stress induction could be used to understand causal effects of stress on other substance use and health outcomes and within other SGM subgroups. This research could provide direct evidence that minority stress explicitly contributes to SGM health disparities, as seen in the connection between events and subsequent behaviors, bolstering many preexisting theories and frameworks. The paradigm could also be used to explore potential differential effects from types of minority stress outcomes (e.g., discrimination vs. expectations of rejection). Teasing out these effects could identify important and tailored intervention points.

Conclusions

There is preliminary evidence to suggest the personalized minority stress induction paradigm is feasible and increases subjective stress in cisgender SMW. Findings also suggest it is possible that the paradigm could target minority stress specifically and could be applied to examine alcohol outcomes in SMW, although these findings need to be confirmed in a larger sample. The paradigm can be used in future work to further expand the theoretical understanding of minority stress’ impact on substance use disparities and broader health outcomes.