Facets of Gender Expression and Discrimination in Relation to Alcohol and Other Drug Use Severity Among Sexual Minority Women and Gender Diverse Individuals Assigned Female at Birth

Abstract

Purpose:

Sexual minority women and gender diverse individuals assigned female at birth (SMW+) consistently report more alcohol and other drug (AOD) use severity than heterosexual women, with greater disparities reported among bisexual plus (bi+) SMW (including bisexual, pansexual, queer, and those with attractions to more than one gender regardless of identity). Furthermore, emerging evidence suggests that SMW with masculine gender expression (e.g., SMW with masculine gender appearance) disproportionately experience problematic AOD use compared to those with feminine gender expression. The minority stress model, which has predominantly been investigated in relation to internalized homonegativity and sexuality-based discrimination, may also account for these AOD use disparities. This study examined gender expression, related discrimination, and AOD use severity among SMW+.

Methods:

In a 2020 sample of SMW+ (n = 236), we investigated AOD use severity in relation to gender expression (appearance, emotional expression, and gender roles) and gender expression-based discrimination after controlling for internalized homonegativity and sexuality-based discrimination through an online survey.

Results:

Masculine gender roles were associated with AOD use severity, whereas masculine appearance and emotional expression were not. In multivariable models, gender identity was inconsistently associated with alcohol use severity, sexuality-based discrimination was consistently associated with alcohol use severity and inconsistently associated with other drug use severity, and gender expression-based discrimination was associated with neither.

Conclusion:

This study emphasizes the importance of examining intersecting aspects of minority identity among SMW+, including facets of gender expression, in relation to AOD use severity.

Introduction

Evidence indicates that there are significant disparities in problematic alcohol and other drug (AOD) use among sexual minority women and gender diverse individuals assigned female at birth (SMW+) compared to heterosexual women as well as across subgroups of SMW+. For instance, relative to heterosexual women, SMW have higher prevalence of alcohol,^1,2 tobacco,^3,4 marijuana,^5,6 and illicit drug use.^7,8 Furthermore, differences in AOD use and related outcomes (e.g., alcohol use disorder [AUD]) vary across subgroups of SMW, such as between gay/lesbian (or monosexual) women and bisexual plus (bi+; i.e., including bisexual, pansexual, queer, and those with attractions to more than one gender regardless of identity)^9–11 women, with bi+ SMW being more likely than monosexual SMW to meet criteria for AUD, use marijuana, use illicit drugs, and misuse opioids in the past year.¹²

The minority stress model,^13,14 which emphasizes negative effects of stressors associated with one or more stigmatized identities on health and engagement in care, is often used to explain SMW's elevated AOD use severity relative to heterosexual women. Specifically, the minority stress model posits that stigma, prejudice, and discrimination create a stressful and hostile social context, which increases the likelihood of mental and behavioral health problems, including problematic AOD use, among sexual minority individuals.^14,15 The model describes specific minority stress processes (e.g., expectations of rejection, internalized stigma), which can infringe on productive coping strategies and perpetuate the negative impact of minority stress.¹⁴ Consistent with the minority stress model, sexuality-based stress is associated with AOD use severity and related outcomes among SMW.^16,17 Studies have also found that internalized stigma is associated with AOD use among SMW.^18,19

AOD risk varies among SMW; yet, studies tend to aggregate data across SMW subgroups.²⁰ Consistent with Crenshaw's intersectionality framework,²¹ collapsing individuals who share a particular stigmatized identity into a single group may limit understanding of how experiences with discrimination and privilege differentially affect SMW+ across sexual identity, gender identity, and gender expression.²¹ Specifically, the experiences of intersecting stigmas related not only to sexuality but also to gender identity and gender expression, may help explain why some SMW+ subgroups are at a heightened risk for problematic AOD use compared to others.

Masculine gender expression and related discrimination may be associated with even greater AOD use severity among SMW+. Relative to heterosexual individuals, sexual minority individuals are more likely to dress in gendered style different from the traditional style associated with their sex assigned at birth,^22,23 signaling a “violation” of normative expectations for appropriate gendered behavior.^24,25 These violations may also signal a sexual minority status, increasing the likelihood of minority stress and victimization^26,27 and subsequent negative implications for AOD use severity.² For example, among a sample of young SMW, those who were categorized as “butch” (i.e., masculine presenting) reported more AOD use than SMW categorized as “femme” (i.e., feminine presenting).²⁸

Importantly, in this same analysis, emotional distress and gay-related stress (external stressors and internalized homonegativity) accounted for higher levels of AOD use among SMW who were categorized as “butch.” In another study, gender-based stressors (e.g., sexism) were associated with problematic AOD use among SMW who identified as masculine-presenting compared to SMW who identified as feminine-presenting.²⁹ Together, this evidence suggests that stress related to more masculine gender expression among SMW+ may uniquely contribute to AOD use severity, beyond sexuality-based stress.

Feminine (“femme”) gender expression might also catalyze increased AOD use among SMW+. Some evidence indicates that “femme” gender expression among SMW is associated with unique proximal stressors, such as perceived lack of authenticity within SMWs' communities, and distal stressors, including femme-related discrimination, which may also lead to increased AOD use.^30,31 Specifically, feminine gender expression may be at odds with valued norms within SMWs' communities, such as the assumption that masculine expression is the primary authentic SMW esthetic.^30,31 In addition, feminine gender expression has been associated with femme-related discrimination in the context of masculine privileging more broadly.³⁰ Although gender expression-related stressors that contribute to AOD use among SMW have been identified, little work has examined nuances of the relationship between self-reported gender expression and AOD use severity in conjunction with sexuality-based stress, particularly accounting for racial and ethnic differences.

Although disparities in AOD use severity are even more pronounced among Black and Latinx SMW compared to White SMW,³² evidence indicates differential relationships between gender expression and minority stress by race and ethnicity among SMW.³³ For example, Everett et al. found that masculinity among White SMW was associated with more victimization compared to Black or Latinx SMW. Their results also indicate that masculinity may be particularly threatening for Latinx SMW (compared to White SMW),³³ who are marginalized based on their race, nationality, gender, and sexuality.³⁴ As minority stress and discrimination differ across racial and ethnic groups, it is critical to account for race and ethnicity when examining intersectional aspects of identity in relation to AOD use severity among SMW+.²¹

As described above, evidence suggests that (1) sexuality-based discrimination and internalized homonegativity are associated with increased AOD use and (2) SMW who present as masculine report greater discrimination and greater AOD use severity. However, little is known about the ways in which components of gender expression beyond masculine/feminine presentation may influence AOD use severity. In addition, the degree to which gender expression-based discrimination is associated with AOD use severity, beyond the impact of sexuality-based discrimination and gender expression, as well as race, ethnicity, and sexual identity, has not been assessed among SMW+. Therefore, we aim to assess relationships between multiple facets of gender expression, including emotional expression and gender roles, and AOD use in SMW+, in the context of sexuality-based discrimination, internalized homonegativity, race, ethnicity, and sexual identity. We also examine the association of discrimination attributed to gender expression with AOD use.

Methods

Sample

This study was approved by the Fenway Health Institutional Review Board before data collection. All participants provided informed consent before beginning the survey. A total of 236 participants completed an online survey on Research Electronic Data Capture (REDCap). Study eligibility criteria included (1) being assigned female at birth, (2) identifying as a SMW (i.e., lesbian, bisexual, queer, or pansexual) or a woman who has sex with women, and (3) being 18 years of age or older.

Procedures

Participants were recruited through online platforms (e.g., Facebook). Self-report data were collected online from January 2020 to March 2020. Several steps were taken to ensure validity of data provided in the online survey.³⁵ First, the CAPTCHA module was used to block spambots.³⁶ In addition, survey responders were not included in the final data set if they (1) completed in under 10 minutes, (2) provided a nonsensical email address or if it was an iteration of a specific pattern seen across multiple surveys, or (3) failed to respond appropriately to repeat questions or items from the Infrequency Scale (e.g., “I eat cement occasionally”), indicative of nonauthentic responding. A total of 628 responses were obtained, ∼60% of which were excluded based on these criteria. Those who met eligibility and validity criteria and completed the survey received a $10 electronic-gift card. The email addresses collected for compensation purposes were stored separately from the anonymous survey data.

Measures

Demographics

Detailed demographic information, including age, race, ethnicity, education, and income, were collected (Table 1).

Table 1.

Descriptive Statistics for All Study Variables (N = 236)

Variable	Mean/n (%)	SD	Range
Demographics
Age	28.68	6.69	18–62
Racea
American Indian/Alaskan Native	4 (1.7)
Asian	18 (7.6)
Native Hawaiian/Pacific Islander	4 (1.7)
Black	15 (6.4)
White	197 (83.5)
Other	5 (2.1)
Ethnicity
Hispanic/Latinx	29 (12.3)
Education
High school/GED or less	18 (7.6)
Some college/AA/technical school	44 (18.6)
College degree	60 (25.4)
Some graduate school	33 (14.0)
Advanced degree	81 (34.3)
Income
$20,000 or less	72 (30.5)
$20,001–$40,000	63 (26.7)
$40,001–$60,000	41 (17.4)
Over $60,000	60 (25.4)
Gender identity
Cisgender female	190 (80.5)
Nonbinary/gender nonconforming/gender queer	39 (16.5)
Other	7 (3.0)
Sexual identity
Monosexualb	68 (28.5)
Bi+c	171 (71.5)
Internalized Homophobia Scale (mean score; 1 = strongly disagree–7 = strongly agree; higher scores indicate greater internalized homonegativity)	1.98	0.95	1.00–4.44
Heterosexist Harassment, Rejection, and Discrimination Scale (mean score; 1 = never–6 = almost all the time; higher scores indicate more frequent discrimination)	1.93	0.88	1.00–4.64
Gender Expression Measure among SMW Scale (mean score; 1 = strongly disagree–6 = strongly agree; higher scores indicate greater masculinity)
Appearance	3.10	1.05	1.00–5.71
Emotional expression	2.73	1.06	1.00–5.50
Gender roles	3.63	0.91	1.25–6.00
Everyday Discrimination Scale–gender expression (mean score; 0 = none–6 = almost every day; higher scores indicate greater discrimination)	1.43	0.96	0.00–3.78
Alcohol Use Disorders Identification Test (response options 0–4 with higher scores indicating greater alcohol use severity; sum range 0–40)	6.70	6.95	0.00–30.00
Drug Abuse Screening Test (response options 0–1 with higher scores indicating greater drug use severity; sum range 0–10)	1.57	2.01	0.00–10.00

^a Sample size and percentages do not sum to 236 and 100%, respectively, due to participants identifying as one or more racial identities.

^b Exclusively homosexual/gay/lesbian.

^c Inclusive of bisexual, queer, pansexual, and other.

AA, Associates of Arts degree; GED, general education diploma; SD, standard deviation; SMW, sexual minority women.

Sexual identity

Sexual identity was assessed using a single item: “Do you identify as”: with the following response options: exclusively heterosexual, mostly heterosexual, bisexual, mostly homosexual/gay/lesbian, exclusively homosexual/gay/lesbian, pansexual, queer, asexual, questioning, and other. Sexual identity was recoded to denote those who identified as monosexual (i.e., exclusively homosexual/gay/lesbian) or bisexual, an umbrella term that refers to individuals with bisexual identity and/or attractions to more than one gender.^10,11,37 Consistent with recent conceptualizations,^9–11 we use the term “bi+” to categorize individuals who reported the following sexual identities: mostly heterosexual, bisexual, mostly homosexual/gay/lesbian, pansexual, queer, asexual, questioning, and other.

Gender identity

Although an initial eligibility screener required endorsement of being a sexual minority woman and being assigned female at birth, gender identity was subsequently assessed in the web-based survey using a single item: “Which of the following best describes you?” with the following response options: male, female, transgender male, transgender female, nonbinary/gender nonconforming/gender queer, and other. Gender identity was recoded to distinguish those who identified as cisgender women from those who identified as nonbinary, gender nonconforming, gender queer, or other. Those who identified as transgender male (n = 11) and transgender female (n = 3) were excluded, given inconsistency with the inclusion criteria, and related screener questions, of identifying as a SMW and being assigned female at birth.

Sexuality-based minority stress

Internalized homonegativity

Internalized homonegativity was assessed using the Internalized Homophobia Scale.^38,39 The Internalized Homophobia Scale is a 9-item measure of personal acceptance and endorsement of sexuality-based stigma. Participants respond to items (e.g., “I wish I weren't lesbian/bisexual”; α = 0.92) on a Likert-type scale ranging from 1 = strongly disagree to 7 = strongly agree. A mean score was calculated, with higher scores indicating greater internalized homonegativity.

Sexuality-based discrimination

Sexuality-based discrimination was measured using the Heterosexist Harassment, Rejection, and Discrimination Scale (HHRDS).⁴⁰ The HHRDS is a 14-item measure of harassment, workplace or school discrimination, and rejection by family, friends, and others. Participants respond to items (e.g., “How many times have you been verbally insulted because of your sexuality”; α = 0.95) on a Likert-type scale ranging from 1 = never to 6 = almost all the time. A mean score was calculated, with higher scores indicating more frequent discrimination.

Gender expression

Gender expression was measured using the Gender Expression Measure among Sexual Minority Women Scale (GEM-SMW).⁴¹ The GEM-SMW has 15 items that assess the degree to which respondents express themselves as traditionally masculine or feminine along three subscales appearance (e.g., sample item: “I never wear makeup”), emotional expression (e.g., sample item: “I cry easily”), and gender roles (e.g., sample items: “I relate to straight men as ‘one of the guys’” and “I like to hold doors open for a date or let my date pass through doors first”). Participants respond to items on a Likert-type scale ranging from 1 = strongly disagree to 6 = strongly agree. Mean scores were calculated for each participant; higher scores indicated greater masculinity, whereas lower scores represented greater femininity; α = 0.75.

Gender expression-based discrimination

Gender expression-based discrimination was assessed using an adapted version of the Everyday Discrimination Scale,⁴² a 9-item measure of chronic and routine unfair treatment in everyday life. In the adapted version, participants responded to the following prompt: “In your day-to-day life how often have any of the following things happened to you because of your gender expression or presentation?” (e.g., “People act as if they think you are dishonest”; α = 0.91). Response options are presented on a Likert-type scale ranging from 0 = none to 6 = almost every day. A mean score was calculated with higher scores indicating greater discrimination.

Alcohol use severity

Alcohol use severity was assessed using the Alcohol Use Disorders Identification Test (AUDIT),⁴³ a 10-item screening measure that assesses alcohol consumption (i.e., frequency and quantity) and alcohol-related consequences (e.g., “unable to remember what happened the night before because you had been drinking”; α = 0.91) on a Likert-type scale. AUDIT scores were derived by summing the scores (0–4) for each of the 10 items, with higher scores indicating greater alcohol use severity.

Other drug use severity

Other drug use severity was assessed using the Drug Abuse Screening Test (DAST).⁴⁴ The DAST has 10 items that assess each participant's nonmedical use of any drug (e.g., cannabis, stimulants, hallucinogens, and narcotics) and drug use-related consequences (e.g., “feel bad or guilty about your drug use”; α = 0.77) on a Likert-type scale. Alcohol use was excluded from the DAST responses. DAST scores were derived by summing the scores (0–1) for each of the 10 items, with higher scores indicating greater drug use severity.

Data analysis plan

First, descriptive analyses were conducted to characterize the sample based on sexual identity, gender identity, gender expression, and associated minority stressors. Next, bivariate correlations and bivariate linear regression models that were adjusted for covariates characterized the associations of sexual identity, internalized homonegativity, sexuality-based discrimination, gender expression, and gender expression-based discrimination with alcohol use severity and other drug use severity. Bivariate linear regression models were conducted separately with alcohol use severity and other drug use severity as dependent variables, assessing relationships with (1) sexual minority identity (monosexual vs. bi+); (2) internalized homonegativity; (3) sexuality-based discrimination; (4) gender expression, including appearance, emotional expression, and gender roles; and (5) gender expression-based discrimination.

All models adjusted for age, White race, Hispanic/Latinx ethnicity, high school education attainment, and gender identity (cisgender vs. not, including nonbinary, gender nonconforming, gender queer, or other). Models 2–5 also adjusted for sexual identity.

To examine whether gender expression and gender expression-based discrimination were associated with AOD use severity over and above the associations of internalized homonegativity and sexuality-based discrimination, we added these two variables to the covariate-adjusted multivariate linear regression models. We sequentially added the following independent variables: (1) sexual identity (monosexual vs. bi+), (2) internalized homonegativity and sexuality-based discrimination, (3) gender expression, including appearance, emotional expression, and gender role subscales, and (4) gender expression-based discrimination to the two separate models predicting alcohol use severity and other drug use severity. The effect size estimate for the final model was quantified using R² and models were compared based on ΔR². Finally, to assess the likelihood of multicollinearity, we evaluated bivariate correlations (Supplementary Table S1) and variance inflation factors (VIFs) for the final models.

Results

Descriptive statistics

Participants' ages ranged from 18 to 62 years (M_age = 28.68, standard deviation = 6.69). The sample was predominantly White (83.5%) and non-Hispanic (85.6%), with most SMW reporting a college education and an income above $20,000. The majority identified as cisgender female (80.5%) and bi+ (71.5%; see Table 1).

Overall, sexual identity (monosexual vs. bi+) was positively associated with gender identity (r = 0.20), internalized homonegativity (r = 0.19), sexuality-based discrimination (r = 0.15), and the appearance (r = 0.19) and gender roles (r = 0.25) aspects of gender expression. Minority-related stressors (i.e., internalized homonegativity, sexuality-based discrimination, and gender expression-based discrimination) were all positively correlated (r = 0.39–0.64).

Alcohol use severity was positively associated with gender identity (r = 0.13), internalized homonegativity (r = 0.37), sexuality-based discrimination (r = 0.46), the emotional expression (r = 0.27) and gender roles (r = 0.26) aspects of gender expression, and gender expression-based discrimination (r = 0.39). Other drug use severity was positively associated with internalized homonegativity (r = 0.17), sexuality-based discrimination (r = 0.30), the emotional expression (r = 0.19) and gender roles (r = 0.21) aspects of gender expression, gender expression-based discrimination (r = 0.29), and alcohol use severity (r = 0.55). All bivariate correlations are shown in Supplementary Table S1.

Adjusted bivariate linear regression models

Alcohol use severity

Sexual identity (monosexual vs. bi+) was not associated with alcohol use severity. Both internalized homonegativity (b = 2.78, standard error [SE] = 0.47, p < 0.001) and sexuality-based discrimination (b = 3.51, SE = 0.45, p < 0.001) were positively associated with alcohol use severity. Two aspects of gender expression, more masculine emotional expression (i.e., emotional suppression, b = 1.47, SE = 0.40, p ≤ 0.001) and adherence to more masculine gender roles (b = 1.36, SE = 0.49, p = 0.01), were positively associated with alcohol use severity. However, the appearance subscale was not significantly associated with alcohol use severity. Finally, gender expression-based discrimination was positively associated with alcohol use severity (b = 2.70, SE = 0.42, p < 0.001; Table 2). Across all models, Hispanic ethnicity, greater than high school education, and non-cisgender identity were associated with greater alcohol use severity.

Table 2.

Adjusted Bivariate Linear Regression Models

	Alcohol use severity			Other drug use severity
	b (SE)	t-Test	p	b (SE)	t-Test	p
Model 1
Age	−0.09 (0.07)	−0.1.27	0.21	−0.01 (0.02)	−0.65	0.52
Race (White ref)	−0.32 (1.21)	−0.26	0.79	0.72 (0.37)	1.96	0.05
Ethnicity (non-Hispanic ref)	5.83 (1.32)	4.43	<0.001	0.89 (0.40)	2.23	0.03
Education (≤high school ref)	5.57 (1.71)	3.25	0.001	0.67 (0.52)	1.29	0.20
Gender identity (cisgender women ref)	2.54 (1.23)	2.07	0.04	−0.14 (0.37)	−0.38	0.71
Sexual identity (monosexual ref)	0.51 (0.97)	0.52	0.60	0.05 (0.29)	0.18	0.86
Model 2
Age	−0.04 (0.06)	−0.66	0.51	−0.01 (0.02)	−0.27	0.79
Race (White ref)	1.15 (1.15)	1.00	0.32	0.98 (0.37)	2.68	0.01
Ethnicity (non-Hispanic ref)	4.69 (1.24)	3.78	<0.001	0.69 (0.39)	1.74	0.08
Education (≤high school ref)	6.50 (1.61)	4.05	<0.001	0.83 (0.51)	1.63	0.10
Gender identity (cisgender women ref)	2.54 (1.23)	2.07	0.04	−0.43 (0.37)	−1.15	0.25
Sexual identity (monosexual ref)	−0.45 (0.92)	−0.49	0.62	−0.12 (0.29)	−0.41	0.68
Internalized homonegativity	2.78 (0.47)	5.94	<0.001	0.50 (0.15)	3.33	0.001
Model 3
Age	−0.08 (0.06)	−1.37	0.17	−0.01 (0.02)	−0.65	0.52
Race (White ref)	0.83 (1.09)	0.76	0.45	0.96 (0.35)	2.74	0.01
Ethnicity (non-Hispanic ref)	3.88 (1.20)	3.23	0.001	0.47 (0.39)	1.23	0.22
Education (≤high school ref)	5.95 (1.53)	3.89	<0.001	0.75 (0.49)	1.52	0.13
Gender identity (cisgender women ref)	2.65 (1.10)	2.42	0.02	−0.12 (0.35)	−0.33	0.74
Sexual identity (monosexual ref)	−0.62 (0.88)	−0.71	0.48	−0.19 (0.28)	−0.67	0.51
Sexuality-based discrimination	3.51 (0.45)	7.71	<0.001	0.75 (0.15)	5.09	<0.001
Model 4
Age	−0.10 (0.06)	−1.55	0.12	−0.01 (0.02)	−0.94	0.35
Race (White ref)	0.22 (1.15)	0.19	0.85	0.83 (0.36)	2.32	0.02
Ethnicity (non-Hispanic ref)	5.19 (1.26)	4.13	<0.001	0.73 (0.39)	1.87	0.06
Education (≤high school ref)	5.81 (1.62)	3.58	<0.001	0.72 (0.50)	1.44	0.15
Gender identity (cisgender women ref)	1.47 (0.40)	3.65	<0.001	0.03 (0.38)	0.09	0.93
Sexual identity (monosexual ref)	−0.46 (0.97)	−0.47	0.64	−0.26 (0.30)	−0.87	0.38
Gender expression
Appearance	0.06 (0.43)	0.13	0.89	0.13 (0.13)	0.95	0.34
Emotional expression	1.47 (0.40)	3.65	<0.001	0.27 (0.13)	2.19	0.03
Gender roles	1.36 (0.49)	2.75	0.01	0.39 (0.15)	2.53	0.01
Model 5
Age	−0.09 (0.06)	−1.46	0.15	−0.01 (0.02)	−0.74	0.46
Race (White ref)	0.13 (1.11)	0.12	0.91	0.82 (0.35)	2.33	0.02
Ethnicity (non-Hispanic ref)	4.65 (1.23)	3.79	<0.001	0.62 (0.39)	1.61	0.11
Education (≤high school ref)	5.44 (1.58)	3.45	<0.001	0.64 (0.50)	1.29	0.20
Gender identity (cisgender women ref)	3.30 (1.14)	2.90	0.004	0.03 (0.36)	0.09	0.93
Sexual identity (monosexual ref)	0.35 (0.90)	0.39	0.69	0.02 (0.28)	0.06	0.95
Gender expression-based discrimination	2.70 (0.42)	6.45	<0.001	0.61 (0.13)	4.63	<0.001

SE, standard error.

Other drug use severity

Sexual identity was not associated with other drug use severity. Both internalized homonegativity (b = 0.50, SE = 0.15, p = 0.001) and sexuality-based discrimination (b = 0.75, SE = 0.15, p < 0.001) were positively associated with other drug use severity. Two aspects of gender expression, more masculine emotional expression (i.e., emotional suppression; b = 0.27, SE = 0.13, p = 0.03) and adherence to more masculine gender roles (b = 0.39, SE = 0.15, p = 0.01), were positively associated with other drug use severity. Finally, gender expression-based discrimination was positively associated with other drug use severity (b = 0.61, SE = 0.13, p < 0.001; Table 2). Ethnicity was the only covariate significantly associated with greater drug use severity in model 1, whereas in all subsequent models, non-White race was the only covariate associated with greater drug use severity.

Adjusted multivariable linear regression models

Alcohol use severity

In model 1 (R² = 0.12, p < 0.001), which examined associations between demographics and alcohol use severity, Hispanic ethnicity (b = 5.83, SE = 1.32, p < 0.001), high school education attainment (b = 5.57, SE = 1.71, p = 0.001), and non-cisgender identity (b = 2.54, SE = 1.23, p = 0.04) were positively associated with alcohol use severity. In model 2 (ΔR² = 0.19, F [2, 228] = 31.73, p ≤ 0.001), sexuality-based discrimination (b = 2.85, SE = 0.58, p < 0.001) was positively associated with alcohol use severity, but internalized homonegativity was not.

In model 3 (ΔR² = 0.03, F [3, 225] = 3.03, p = 0.03), adherence to more masculine gender roles was the only facet of gender expression that was positively associated with alcohol use severity (b = 1.10, SE = 0.46, p = 0.02). In model 4 (R² = 0.36; ΔR² = 0.01, F [1, 224] = 4.98, p = 0.03), gender expression-based discrimination was not associated with alcohol use severity, but sexuality-based discrimination and adherence to more masculine gender roles remained significant. Across models, ethnicity and education were significantly associated with alcohol use severity. In addition, sexuality-based discrimination remained significantly associated with alcohol use severity in models 3 and 4. Sexual identity was not associated with alcohol use severity in any of the models (Table 3).

Table 3.

Adjusted Multivariable Linear Regression Models

	Alcohol use severity			Other drug use severity
	b (SE)	t-Test	p	b (SE)	t-Test	p
Model 1
Age	−0.09 (0.07)	−0.1.27	0.21	−0.01 (0.02)	−0.65	0.52
Race (White ref)	−0.32 (1.21)	−0.26	0.79	0.72 (0.37)	1.96	0.05
Ethnicity (non-Hispanic ref)	5.83 (1.32)	4.43	<0.001	0.89 (0.40)	2.23	0.03
Education (≤high school ref)	5.57 (1.71)	3.25	0.001	0.67 (0.52)	1.29	0.20
Gender identity (cisgender woman ref)	2.54 (1.23)	2.07	0.04	−0.14 (0.37)	−0.38	0.71
Sexual identity (monosexual ref)	0.51 (0.97)	0.52	0.60	0.05 (0.29)	0.18	0.86
Model 2
Age	−0.07 (0.06)	−1.10	0.27	−0.01 (0.02)	−0.59	0.56
Race (White ref)	1.16 (1.10)	1.06	0.29	0.98 (0.36)	2.76	0.01
Ethnicity (non-Hispanic ref)	3.82 (1.20)	3.19	0.002	0.47 (0.39)	1.21	0.23
Education (≤high school ref)	6.22 (1.53)	4.07	<0.001	0.76 (0.50)	1.54	0.13
Gender identity (cisgender women ref)	2.02 (1.14)	1.77	0.08	−0.16 (0.37)	−0.42	0.67
Sexual identity (monosexual ref)	−0.77 (0.88)	−0.88	0.38	−0.20 (0.29)	−0.69	0.49
Internalized homonegativity	1.35 (0.57)	1.81	0.07	0.06 (0.18)	0.34	0.73
Sexuality-based discrimination	2.85 (0.58)	4.95	<0.001	0.71 (0.19)	3.77	<0.001
Model 3
Age	−0.08 (0.06)	−1.38	0.17	−0.02 (0.02)	−0.88	0.38
Race (White ref)	1.26 (1.08)	1.17	0.25	1.00 (0.35)	2.83	0.01
Ethnicity (non-Hispanic ref)	3.63 (1.19)	3.07	0.002	0.42 (0.39)	1.08	0.28
Education (≤high school ref)	6.20 (1.51)	4.12	<0.001	0.77 (0.49)	1.57	0.12
Gender identity (cisgender women ref)	2.30 (1.18)	1.95	0.05	0.002 (0.38)	0.004	1.0
Sexual identity (monosexual ref)	−1.24 (0.91)	−1.36	0.17	−0.40 (0.30)	−1.35	0.18
Internalized homonegativity	0.78 (0.58)	1.35	0.18	0.02 (0.19)	0.12	0.91
Sexuality-based discrimination	2.55 (0.58)	4.41	<0.001	0.60 (0.19)	3.21	0.002
Gender expression
Appearance	−0.02 (0.40)	−0.06	0.95	0.10 (0.13)	0.78	0.44
Emotional expression	0.63 (0.40)	1.55	0.12	0.12 (0.13)	0.95	0.35
Gender roles	1.10 (0.46)	2.41	0.02	0.33 (0.15)	2.22	0.03
Model 4
Age	−0.09 (0.06)	−1.43	0.15	−0.02 (0.02)	−0.93	0.36
Race (White ref)	1.17 (1.08)	1.09	0.28	0.97 (0.35)	2.77	0.01
Ethnicity (non-Hispanic ref)	3.60 (1.18)	3.06	0.003	0.41 (0.38)	1.06	0.29
Education (≤high school ref)	6.07 (1.50)	4.05	<0.001	0.73 (0.49)	1.50	0.14
Gender identity (cisgender women ref)	2.57 (1.18)	2.19	0.03	0.08 (0.38)	0.20	0.84
Sexual identity (monosexual ref)	−1.04 (0.91)	−1.14	0.26	−0.34 (0.30)	−1.16	0.25
Internalized homonegativity	0.75 (0.58)	1.29	0.20	0.01 (0.19)	0.06	0.95
Sexuality-based discrimination	1.87 (0.67)	2.78	0.01	0.42 (0.22)	1.91	0.06
Gender expression
Appearance	−0.02 (0.39)	−0.06	0.96	0.10 (0.13)	0.79	0.43
Emotional expression	0.58 (0.40)	1.45	0.15	0.11 (0.13)	0.86	0.39
Gender roles	1.06 (0.45)	2.34	0.02	0.32 (0.15)	2.15	0.03
Gender expression-based discrimination	1.02 (0.52)	1.98	0.05	0.28 (0.17)	1.64	0.10

Other drug use severity

In model 1 (R² = 0.05, p = 0.03), Hispanic ethnicity (b = 0.89, SE = 0.40, p = 0.03) was positively associated with other drug use severity. In model 2 (ΔR² = 0.19, F [2, 228] = 12.68, p ≤ 0.001), sexuality-based discrimination was positively associated with other drug use severity (b = 0.71, SE = 0.19, p < 0.001). In model 3 (ΔR² = 0.03, F [3, 225] = 2.47, p = 0.063), adherence to more masculine gender roles was the only facet of gender expression that was positively associated with other drug use severity (b = 0.33, SE = 0.15, p = 0.03). In the final model 4 (R² = 0.19; ΔR² = 0.01, F [1, 224] = 3.36, p = 0.07), gender expression-based discrimination was not associated with other drug use severity (b = 0.28, SE = 0.17, p = 0.10).

Hispanic ethnicity was the only covariate associated with other drug use severity in model one and non-White race was the only covariate associated with other drug use severity in models 2–4. Sexuality-based discrimination remained significant when gender expression was added in model 3, but was no longer significant when gender expression-based discrimination was added in model 4. In the final model (model 4), adherence to more masculine gender roles was the only facet of gender expression significantly associated with other drug use severity. Furthermore, neither sexual identity nor internalized homonegativity was associated with other drug use severity (Table 3).

Discussion

In this sample of SMW and gender diverse individuals assigned female at birth (SMW+), we identified that aspects of gender expression were differentially associated with AOD use severity, indicating that conceptualizing gender expression beyond masculine versus feminine appearance is important for understanding the minority stress experiences of SMW+ in relation to AOD use. Specifically, we found that gendered appearance was not associated with alcohol use severity in adjusted and unadjusted models; however, gendered emotional expression (or suppression of emotion, which is traditionally associated with masculinity) and engagement in behaviors linked to more masculine gender roles were both associated with AOD use severity in bivariate models, consistent with existing literature indicating more masculine presenting SMW report higher levels of AOD use.^28,29

In adjusted multivariable models that first included internalized homonegativity and sexuality-based discrimination and subsequently added three facets of gender expression, only expression of masculine gender roles was significantly associated with alcohol use severity and other drug use severity. Future studies should examine drinking norms among SMW+ as potentially influencing the association of AOD use and gender roles.⁴⁵

Although the change in R² was relatively small with the introduction of facets of gender expression, this finding indicates that more work is needed to examine the clinical relevance of considering facets of gender expression beyond single-item assessments of masculinity and femininity,⁴⁶ butch/femme,²⁸ and gender nonconformity.²² Although gender expression is often conceptualized as being synonymous with appearance,³¹ these results indicate that gendered emotional expression and behaviors consistent with gender roles are distinct and critical aspects of gender expression that are chronically underaddressed in the literature and are related to AOD use severity among SMW+ individuals.

Emerging research has begun to focus on clinical interventions to address AOD use among SMW+ individuals.^47–49 Future clinical interventions to address AOD use among SMW+ individuals may benefit from examination of ways in which behavioral enactment of masculine gender norms and gendered limitations in emotional expression (e.g., alcohol consumption to adhere to perceived masculine norms or withholding emotional experiences) may perpetuate AOD use. Providing strategies for challenging and restructuring perceived masculine gender norms associated with AOD use in conjunction with strategies to mitigate the negative impacts of minority stress may benefit SMW+ individuals seeking treatment for AOD use.

Inconsistent with existing literature, in unadjusted and adjusted models, we did not identify differences in AOD use severity between monosexual and bi+ SMW.^12,50 Given the expansive literature indicating significant differences in AOD use and minority stress among monosexual versus bi+ individuals,^12,51,52 we subsequently included sexual identity (monosexual vs. bi+ individuals) as a covariate along with age, race, ethnicity, education, and gender. However, the variance associated with sexual identity did not account for differences in AOD use severity. The lack of association between sexual identity and AOD use severity may be attributable to the heterogeneity and inclusivity of the bi+ category (e.g., inclusive of bisexual, queer, pansexual, and other).

Across models, we identified demographic patterns both consistent and inconsistent with extant literature. Hispanic ethnicity and greater than a high school education were associated with greater alcohol use severity, adding nuance to the existing literature.^53,54 Identifying as non-cisgender (i.e., nonbinary/gender nonconforming/gender queer, or other) was inconsistently associated with alcohol use severity and not associated with other drug use severity; there is limited research examining AOD use between cisgender SMW and nonbinary/gender nonconforming/gender queer, and other-identified individuals assigned female at birth. Furthermore, across models assessing relationships between internalized homonegativity, sexuality-based discrimination, gender expression, and gender expression-based discrimination with other drug use severity, non-White race was associated with greater severity of other drug use.⁵⁵ Together, these results indicate a need to investigate intersectional demographic differences among SMW+ in relation to AOD use severity, and minority stress related to sexual identity and gender expression.

Limitations

Although this study uniquely emphasizes the need for future investigation of facets of gender expression in relation to AOD use severity among SMW+, there are several limitations. First, this study involved a cross-sectional web-based convenience sample, limiting generalizability of our findings. Second, the relatively small sample (N = 236) was predominantly White and college educated, limiting racial and education comparisons. In addition, measures used to assess gender expression captured masculine and feminine expression in the same scale, which may be reductive, as the absence of one is not necessarily indicative of the presence of the other. Future investigation would benefit from assessing masculine and feminine gender expression separately.⁵⁶

Finally, while the included variables were conceptually related, indicating the possibility of multicollinearity, the bivariate correlations were all <0.70 (Supplementary Table S1 r = −0.25 to 0.64) and the VIFs for the final models (Model 4 in Table 3) did not indicate likely multicollinearity (1.01–1.75). Notably, variables that we anticipated would be most highly correlated (e.g., facets of gender expression: appearance, emotional expression, and gender roles) were only moderately correlated (r = 0.20–0.28), suggesting that these subscales are indeed capturing related, but discrete components of gender expression that are each worthy of investigation.

Conclusion

This study identified that masculine gender roles, but not masculine appearance or gendered emotional expression, were associated with higher AOD use severity among SMW+, above the associations with internalized homonegativity and sexuality-based discrimination. This finding indicates the importance of examining facets of gender expression beyond appearance in relation to disparities in AOD use severity and minority stress among SMW+. Additional work is needed to better understand how facets of gender expression are differentially associated with discrimination, minority stress, and AOD use severity among SMW+. Furthermore, future research should utilize larger representative samples and qualitative interviews to investigate how these relationships interact with race, ethnicity, education, and gender identity. Finally, it is critical that clinicians remain cognizant of the ways in which aspects of identity and expression may differentially contribute to problematic AOD use in the context of internalized homonegativity, sexuality-based discrimination, and gender expression-based discrimination among SMW+.

Authors' Contributions

A.W.B. and J.C.M. led the development of study design. A.W.B and J.D.F. completed the statistical analyses and A.M.S. and K.R.G. offered feedback regarding data analysis and interpretation. A.W.B., J.D.F., and A.M.S. drafted the article, and K.R.G., J.C.M., and J.R.S. lent specific content expertise to provide critical feedback on revising the work. All authors reviewed and approved the final version of this article.

Disclaimer

The content of this work is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The NIH had no role in study design, data collection and analysis, decision to publish, or preparation of the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

The study was supported by a Fenway Institute Grant (PI: Batchelder). A.W.B.'s time was supported by NIDA (K23DA043418 and the Fenway Institute). Both J.D.F.'s and A.M.S.'s time were supported by research training grants (2T32AT000051 and 5T32MH116140, respectively), and J.R.S. was supported by NIAAA (K01AA028239).

Supplementary Material

Supplementary Table S1