Associations Between Latent Classes of Trauma Exposure and Minority Stressors and Substance Use Among Cisgender Sexual Minority Women

Abstract

Psychosocial stressors (e.g., minority stressors, trauma exposure) profoundly impact sexual minority women’s (SMW’s) risk of alcohol and other drug (AOD) use. However, research has not examined whether there are distinct typologies (i.e., patterns) of psychosocial stressors and whether these vary based on sociodemographic characteristics or are differentially associated with AOD outcomes (e.g., alcohol dependence) among SMW. This study aimed to identify latent classes of SMW reporting distinct typologies of psychosocial stressors and examine predictors and outcomes of latent classes of psychosocial stressors among SMW. Participants included a community sample of 602 SMW (M_age = 39.9, SD = 14.0; 74.0% lesbian; 37.4% White, 36.6% Black, 22.3% Latinx; 26.6% annual income ≤$14,999). Latent class analysis was used to identify typologies of psychosocial stressors. Regression analyses were employed to examine sociodemographic predictors and AOD outcomes of class membership. Three classes of psychosocial stressors emerged. Participants in Class 1 were likely to report relatively low adversity. SMW in Class 2, who reported childhood physical abuse (CPA), severe childhood sexual abuse, and adult physical assault, were vulnerable to discrimination and stigma consciousness. A distinct subgroup of SMW (Class 3) was at heightened risk of CPA, adult sexual assault (ASA), and stigma consciousness. Older SMW, Black SMW, and SMW with lower social support were more likely to be in classes characterized by higher adversity. Older SMW were at disproportionate risk of CPA and ASA. Different combinations of psychosocial stressors were uniquely associated with AOD outcomes. Findings underscore the importance of considering within-group heterogeneity in SMW’s differential risk of psychosocial stressors and AOD outcomes. Routine screening of psychosocial stressors across several dimensions, brief interventions targeting AOD outcomes, and policies mitigating structural drivers of SMW’s increased risk of trauma and minority stressors may be especially important for older SMW, Black SMW, and SMW who lack social support.

Cisgender sexual minority women (SMW; e.g., lesbian, bisexual) are at heightened risk of negative alcohol and other drug (AOD) outcomes, such as alcohol use disorder, cannabis use, and illicit polydrug use (use of at least two illicit substances) (Evans-Polce et al., 2020; Hughes et al., 2020; McCabe, Engstrom, et al., 2022; Schuler & Collins, 2020). For instance, studies using data from national surveys have shown that SMW are more likely than heterosexual women to report heavy episodic drinking (HED), and cannabis, illicit polydrug, and heroin use (Evans-Polce et al., 2020; McCabe, Engstrom, et al., 2022). These risks vary across sociodemographic characteristics (Hughes et al., 2020; Schuler & Collins, 2020). For example, bisexual women have elevated rates of negative AOD outcomes compared to lesbian/gay women (Schuler & Collins, 2020). Within-group differences in AOD outcomes may result from SMW’s differential exposure to psychosocial stressors, such as trauma and minority stressors (Cerezo & Ramirez, 2020; Dyar et al., 2021; McCabe et al., 2022). Psychosocial stressor exposure varies in severity, timing, context, and type (Keyes et al., 2011). To date, no research has examined whether patterns—or typologies—of psychosocial stressors across these stress-exposure dimensions emerge among SMW and whether typologies are differentially associated with SMW’s sociodemographic characteristics and AOD outcomes.

Research findings highlight that SMW are at heightened risk of trauma across the lifespan, including childhood sexual abuse (CSA), childhood physical abuse (CPA), and intimate partner violence (IPV) (McCabe, Hughes, et al., 2022; Porsch et al., 2022; Scheer et al., 2022). For instance, one study using population-based data demonstrated that SMW were 8 to 14 times more likely than heterosexual women to experience recent IPV (Scheer et al., 2022). Another recent study using nationally representative data found that 30.6% of bisexual and 17.8% of lesbian women reported at least two types of CSA compared to 9.8% of heterosexual women (McCabe, Hughes, et al., 2022). Societal stigma toward same-sex sexuality may help explain SMW’s disproportionate risk of trauma relative to heterosexual women (McCabe, Hughes, et al., 2022; McLaughlin et al., 2012; Scheer et al., 2022). That is, in addition to general stressors, sexual minority individuals experience unique and chronic sexual minority-related stressors (i.e., minority stress; Brooks, 1981; Meyer, 2003). Minority stressors range from the interpersonal to the structural and are associated with subjective processes (e.g., internalized stigma beliefs) and AOD outcomes (Brooks, 1981; Hughes et al., 2020; Meyer, 2003). There is a strong need to develop integrative models that examine the confluence of trauma and minority stressors (i.e., psychosocial stressors; Evans-Polce et al., 2020; Fitzpatrick et al., 2020) as it remains unknown how stressors intersect to create typologies and may be differentially associated with AOD use among SMW.

Psychosocial stressors may synergistically erode SMW’s health by spiking SMW’s stress response systems and overwhelming coping resources (i.e., allostatic load; Juster et al., 2019). Persistent and chronic activation of the stress response system leads to cumulative “wear and tear” that can become maladaptive and increase risk of negative AOD outcomes (Koob & Le Moal, 2001). The association between psychosocial stressors and negative AOD outcomes among SMW is well established (Dyar et al., 2021; McCabe, Hughes, et al., 2022; Scheer et al., 2021). Repeated exposure to psychosocial stressors can have lasting effects on stress reactivity by sensitizing victimized individuals to future stressors (Nurius et al., 2013). Yet, research has yet to examine simultaneously the overlapping effects of multiple stressors among SMW.

Modeling Psychosocial Stressors With Latent Class Analysis

Most extant research investigating psychosocial stressors and health outcomes among SMW have relied on variable-centered approaches (Cerezo & Ramirez, 2020; McCabe, Hughes, et al., 2022; McCabe et al., 2019; Sutter et al., 2019). These approaches have typically been used to assess the presence of each stressor separately, to differentiate the impact of minority stressors and trauma among SMW by controlling for a particular stressor, or to categorize sources of psychosocial stressors (e.g., anti-gay vs. non-identity-based assault; Cerezo & Ramirez, 2020; McCabe, Hughes, et al., 2022; Sutter et al., 2019). Examining the behavioral health impact of psychosocial stressors using this sort of methodological approach may fail to capture complex, multivariate patterns of stressors across various dimensions, and resultant risk for AOD use.

Compared to variable-centered approaches, person-centered approaches (e.g., latent class analysis [LCA]) can provide a broader understanding of how psychosocial stressors cluster among SMW. Person-centered approaches assume heterogeneity in psychosocial stressors (Lanza & Rhoades, 2013) and thus are well suited for studying how SMW group into homogenous, mutually exclusive subsets of shared psychosocial stressor classes (Lanza & Rhoades, 2013). Previous research has used LCA to understand victimization and perpetration typologies among sexual and gender minority populations (Kassing et al., 2021; Layland et al., 2022; Scheer et al., 2021; Sutter et al., 2019; Talley et al., 2022). Fewer studies have used person-centered methods to uncover homogenous subgroups of sexual minority people characterized by shared minority stressors (Layland et al., 2022; Puckett et al., 2017; Shramko et al., 2018; Talley et al., 2022). No studies have used LCA to examine complexity of trauma and minority stressors across different dimensions or associations between classes of stressors and AOD use among SMW.

Prevention scholars have called for intersectional approaches (Bowleg et al., 2003) to identify disparities in psychosocial stressors among sexual and gender minority people, including SMW (Calabrese et al., 2015; Cerezo & Ramirez, 2020). Yet, information about underlying subgroups of SMW across sociodemographic characteristics who have a high probability of endorsing multiple psychosocial stressors is lacking. Prior studies using variable-centered approaches have documented that younger age, bisexual identity, racial/ ethnic minority identity, lower income, and lower social support are associated with SMW’s elevated exposure to minority stressors and trauma (Calabrese et al., 2015; Feinstein & Dyar, 2017; Puckett et al., 2017). Identifying social determinants of latent classes of psychosocial stressors can help determine specific clinical and public health efforts to reach vulnerable subgroups of SMW.

This study’s aims were to: (a) uncover classes of SMW reporting distinct patterns of psychosocial stressors, (b) examine associations between sociodemographic characteristics and classes of psychosocial stressors, and (c) identify associations between class membership and SMW’s AOD outcomes. Findings may advance disparity-focused, within-group research seeking to assess unique configurations of psychosocial stressors and associated behavioral health outcomes.

Methods

Sample and Procedures

Data are from Wave 3 of the Chicago Health and Life Experiences of Women (CHLEW) study (see Hughes et al., 2021 for more details), a community-based, 22-year, five-wave study of cisgender SMW aged 18 and older. Wave 3 data were collected from 2010 to 2012 (N = 353 women from the original cohort were retained [response rate = 79%]). In Wave 3 a supplemental sample of younger (18–25 years) SMW, bisexual women, and additional Black and Latinx SMW (N = 373) was recruited using a modified version of respondent-driven sampling (Martin et al., 2015). We focused on Wave 3 data for these analyses because it includes a larger (N = 726) and more diverse sample than other waves. Four SMW who were believed to have misrepresented their sexual identity were excluded (Scheer et al., 2022). Participants who identified as mostly heterosexual, only heterosexual, or other sexual identity were also excluded from analyses (n = 26; Scheer et al., 2022). Those with missing data for psychosocial stressors (n = 4, 0.6% for physical IPV to n = 44, 6.1% for sexual IPV) were also excluded from analyses, leaving an analytic sample of 602. The Institutional Review Board of the University of Illinois at Chicago approved all study procedures.

Measures

Alcohol Use Outcomes.

HED was defined as consuming either ≥6 drinks/day in the past 12 months, or typically consuming ≥4 drinks on drinking days (1 = at least one episode of consuming ≥6 drinks on one or more days in the past year or consuming ≥4 drinks on drinking days, 0 = neither). Participants who endorsed three or more (out of five) DSM-IV alcohol dependence symptoms (e.g., continued use despite consequences) in the past 12 months were coded as meeting criteria for past-year DSM-IV alcohol dependence (Cronbach’s α = .74; Hasin et al., 2003). Consistent with prior research (Glass et al., 2015; Wilsnack et al., 2008), alcohol problem recognition was determined by asking participants whether they had ever wondered if they were developing a drinking problem (1 = yes, 0 = no).

Drug Use Outcomes.

Cannabis use was assessed using questions about participants’ use of marijuana, pot, hashish, or tetrahydrocannabinol (THC) in the past 12 months (1 = any use, 0 = no use; Liautaud et al., 2021). Heroin or illegal methadone use was determined by participants’ responses to questions asking whether they had ever used heroin or illegal methadone (1 = any use, 0 = no use). Similar to prior research (Houang et al., 2022), illicit polydrug use was defined as lifetime use of two or more of the following substances: stimulants, cocaine/crack, heroin/illegal methadone, hallucinogens, or club drugs.

Psychosocial Stressors.

Trauma was assessed using several measures of adversity during childhood and adult. CSA was assessed using Gail Wyatt’s CSA inventory (Wyatt, 1985) of eight types of sexual experiences occurring before age 18. Consistent with prior research (Hughes et al., 2014; Wilsnack et al., 2012; Wyatt, 1985), CSA in this study was defined as: (a) any unwanted intrafamilial sexual activity before age 18, or that involved a family member five or more years older than the participant; and/or (b) any unwanted extrafamilial sexual activity before age 18, or before age 13 and involved another person five or more years older. Given that dichotomizing CSA (any vs. no exposure) could evade the possibility of examining dose-response relationships between CSA severity and AOD outcomes (Dovran et al., 2019) and result in loss of statistical power (MacCallum et al., 2002), we assessed CSA severity using an ordered categorical approach. Although there is no gold standard for measuring CSA severity (Boroughs et al., 2015), it can be characterized across dimensions (e.g., chronicity, age at first abuse; Young et al., 2011). Similar to prior research (Boroughs et al., 2015), we used intrusion level (i.e., penetration) to demarcate CSA severity: 3 = high-risk contact (e.g., having vaginal/anal penetration), 2 = low-risk contact (e.g., fondling), 1 = noncontact (e.g., being asked or forced to expose private parts), 0 = no CSA.

Childhood neglect was determined by asking participants whether their basic needs (food, clothing, shelter, love) were neglected as a usual discipline method; 1 = yes, 0 = no. CPA was assessed with the question, “Do you feel that you were physically abused by your parents or other family members when you were growing up?”; 1 = yes, 0 = no.

Adult sexual assault (ASA) was assessed by asking participants, “Since you were 18 years old was there a time when someone forced you to have intercourse that you really did not want? This might have happened with partners, lovers, or friends, as well as with more distant persons and strangers.”; 1 = yes, 0 = no (Caceres et al., 2019; López et al., 2023; Veldhuis et al., 2022; Wilsnack et al., 2012). Adult sexual IPV was assessed by asking, “Since you were 18 years old was there a time when someone you were dating, an ex-partner, a recent partner, or a current partner forced you to have intercourse that you really did not want?” and whether they had “experienced any other kind of sexual assault” by someone they were dating, an ex-partner, a recent partner, or a current partner; 1 = any adult sexual IPV, 0 = no (Caceres et al., 2019; Veldhuis et al., 2022). Adult physical assault was determined by asking, “Not counting unwanted sexual experiences, has anyone other than your partner attacked you with a gun, knife or some other weapon, whether you reported it or not?” and whether they had been attacked by someone other than a partner “without a weapon but with the intent to kill or seriously injure you?”; 1 = any adult physical assault, 0 = no (Caceres et al., 2019; Veldhuis et al., 2022). Similar to prior research (Decker et al., 2015), participants were also asked two questions from the Conflict Tactics Scale (Straus et al., 1996) including whether a partner ever “threw something at you, pushed you, or hit you?” or “threatened to kill you, with a weapon or in some other way?”; 1 = any adult physical IPV, 0 = no (Caceres et al., 2019; Veldhuis et al., 2022).

Minority stressors included discrimination, stigma consciousness, and internalized stigma. Similar to prior research (Kidd et al., 2022), past-year discrimination was assessed using the six-item Experiences of Discrimination Scale (Krieger et al., 2005) based on sex/gender, race/ethnicity, and/or sexual identity. This scale assessed frequency of experiencing discrimination in six different situations, including discrimination affecting access to health care treatment. Following prior studies (Glass et al., 2020; Krieger et al., 2005), we calculated the number of situations in which any discrimination was experienced based on sex/gender, race/ethnicity, and/or sexual identity. Items were summed (0–6). Cronbach’s alpha (α = .69). Given that discrimination was positively distributed in the current study, and as recommended in contemporary epidemiological research (Gillespie & Anderson, 2018), tertiles were calculated: “high” (3 = 2+ forms of discrimination), “medium” (2 = 1 form of discrimination), and “low” (1 = no discrimination).

Stigma consciousness (i.e., the extent to which participants are influenced by and perceive stereotypes) was assessed using the Stigma Consciousness Questionnaire (Pinel, 1999). Response options ranged from 0 (strongly disagree) to 6 (strongly agree). Standardized scores were used because the scale for bisexual women had one more item than the scale for lesbian women. Each participant was assigned the standardized mean scale score that corresponded with their reported sexual identity. Final scale scores ranged from −2.71 to 3.25; higher scores indicated greater stigma consciousness (Cronbach’s α = .81). Consistent with our scoring approach for discrimination, responses were divided into tertiles and labeled “high” (3 = ≥0.441), “medium” (2 = −0.552 to 0.440), and “low” (1 = ≤ −0.553).

Internalized stigma was assessed with 10 questions adapted from Herek et al. (1997). Responses ranged from 1 (strongly agree) to 5 (strongly disagree). Scores ranged from 10 to 38 (SD = 5.37); higher total scores indicated greater internalized stigma (Cronbach’s α = .81). Given that internalized stigma was positively distributed and had high dispersion in the current study, and consistent with our scoring approach for discrimination and stigma consciousness, responses were divided into tertiles and labeled “high” (3 = ≥15), “medium” (2 = 12–14), and “low” (1 = ≤11).

Sociodemographic Characteristics.

Participants reported their sexual identity using the following response options: “only lesbian/gay,” “mostly lesbian/gay,” “bisexual,” “mostly heterosexual,” “only heterosexual/straight,” and “other.” Consistent with prior research (López et al., 2023) and because responses to key questions differed little between SMW who identified as mostly lesbian/gay versus only lesbian/gay (López et al., 2023), we created a dummy variable to represent those who identified as “only lesbian/gay” or “mostly lesbian/gay” (0; “lesbian”) or “bisexual” (1). We measured age continuously. Race/ethnicity was assessed using responses two questions assessing Hispanic or Latinx origin or descent and race. Participants who identified as both Latinx and any other group were categorized as Latinx (LaVeist-Ramos et al., 2012); participants who identified with more than one race or as a race other than White, Black, or Latinx were excluded from the current analyses. That is, given small cell sizes, SMW who identified as Asian/Pacific Islander (n = 10), Native American/Alaskan Native (n = 13), or who reported that they did not know their race/ethnicity (n = 2) were excluded. Income was assessed with the following question: “For the last tax year, which represents your total household income from all sources?” We dichotomized responses (0 = above lowest quartile [>$14,999]; 1 = under lowest quartile [≤$14,999]). Participants reported their relationship status (in a relationship vs. not in a relationship) and geographical location (medium city or larger vs. town, village, or small city). Social support was assessed using the 12-item Multidimensional Scale of Perceived Social Support (MSPSS; Zimet et al., 1988); we used total social support scores (Cronbach’s α = .91), consistent with other SMW-focused research (Caceres et al., 2019; Ehlke et al., 2020; Puckett et al., 2015). Prior research has shown MSPSS to have strong internal consistency for the overall scale (Cronbach’s α = .88; Zimet et al., 1988).

Statistical Analysis

We used descriptive statistics to describe the sample’s sociodemographic characteristics, psychosocial stressors, and AOD outcomes. We then used LCA to identify unique patterns of psychosocial stressors among participants. Specifically, we used the three-step latent class analytic approach (Bakk et al., 2013), which prevents measurement bias related to class membership by correcting for classification error (Bakk et al., 2013).

All psychosocial stressors were treated as ordered categorical variables in the LCA models. This included minority stressor continuous variables that were ordered into categorical tertiles to avoid making untenable normality assumptions that would otherwise be required for continuous measures in LCA. We fit 1- to 6-class models with the 10 psychosocial stressors as indicators and considered models with residual correlations. A central assumption of LCA is local or conditional independence, which specifies that conditional on the latent variable, observed indicators are independent. We evaluated the local independence assumption by examining residual correlations among all item pairs (Vermunt & Magidson, 2016). Residual correlations were included for LCA models that violated the local independence assumption. The choice of LCA model was based on specified a priori criteria to identify the most optimally fitting LCA model: relative fit, including low Log Likelihood, Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC), sample-size-adjusted BIC (aBIC), entropy above 0.60; class size; and interpretability (Lanza & Rhoades, 2013). Average posterior probabilities were used to examine class homogeneity (Nylund et al., 2007).

After selecting the best-fitting LCA model, we employed multinomial logistic regression to examine sociodemographic factors as class predictors, while accounting for classification error in class assignment (Bakk et al., 2013). We used multinomial logistic regression models regressing AOD outcomes onto class membership. For all models, we used the bias-adjusted maximum likelihood approach. Regression models controlled for sociodemographic factors.

Sample sizes depended on the amount of missing data for each model. Missing data for sociodemographic characteristics ranged from 0% for age to 4.2% for income. By design of the analytic study sample, there were no missing data across our indicators of psychosocial stressors. Missing data across outcome variables ranged from 0% for illicit polydrug use to 0.3% for HED. Only SMW who reported lifetime cannabis use were included in the regression model examining past-year cannabis use. Missing data were handled using pairwise deletion.

Descriptive statistics were conducted in SPSS version 27. All other analyses were conducted using Latent GOLD software, version 6.0 (Vermunt & Magidson, 2016). We performed a post hoc adjustment of p values using Benjamini–Hochberg procedures (Benjamini & Hochberg, 1995). We also conducted a post hoc sensitivity analysis comparing our LCA results to a variable-centered approach by regressing AOD outcomes on psychosocial stressors.

Results

Sample Characteristics

As shown in Table 1, participants were, on average, 39.9 years of age (SD = 14.0). Most (74%) identified as lesbian; 26.0% identified as bisexual. Relatively equal percentages identified as White (37.4%) and Black (36.6%); fewer (22.3%) identified as Latinx. Less than half (41.9%) reported past-year HED; fewer reported lifetime alcohol problem recognition (38.3%), past-year cannabis use (33.9%), lifetime heroin/illegal methadone use (11.8%), and illicit polydrug use (26.6%). About 10% met criteria for past-year DSM-IV alcohol dependence (10.8%).

Table 1.

Sociodemographic Characteristics and Presence of Alcohol and Other Drug Use and Negative Outcomes (Chicago Health and Life Experiences of Women [CHLEW]; N = 601).

	n (%)
Sociodemographic characteristics
Age, M [SD], (range: 18–82)	39.98 [14.04]
Sexual identity
Lesbian	445 (74.0)
Bisexual	156 (26.0)
Race/ethnicity
White	225 (37.4)
Black	220 (36.6)
Latinx	134 (22.3)
Incomea
Annual income over $14,999	416 (69.2)
Annual income $14,999 or less	160 (26.6)
Relationship status
Not in a relationship	231 (38.4)
In a relationship	367 (61.1)
Geographical location
Medium-sized city or larger	572 (95.2)
Town, village, or small city	28 (4.7)
Perceived social support, M [SD], (range: 12–84)	66.24 [12.50]
Presence of alcohol and other drug outcomes
Heavy episodic drinkingb	252 (41.9)
DSM-IV alcohol dependencec	65 (10.8)
Alcohol problem recognitiond	230 (38.3)
Cannabis usee	204 (33.9)
Heroin or illegal methadone usef	71 (11.8)
Illicit polydrug useg	214 (35.6)

Note. Percentages may not equal 100 due to missing data.

^aLowest quartile was used to calculate low-income status (1 = ≤$14,999 annually).

^bPast-year heavy episodic drinking (1 = ≥6 drinks on one or more days in the past year or ≥4 drinks on a typical drinking day; 0 = <6 drinks on one or more days in the past year and <4 drinks on a typical drinking day).

^cPast-year DSM-IV alcohol dependence (1 = ≥3 DSM-IV symptoms of alcohol dependence; 0 = <3 DSM-IV symptoms of alcohol dependence; Hasin et al., 2003).

^dLifetime alcohol problem recognition (1 = wondered if developing a drinking problem; 0 = never wondered if developing a drinking problem).

^ePast-year cannabis use among lifetime users only (1 = cannabis use; 0 = no cannabis use).

^fLifetime heroin or illegal methadone use (1 = heroin or illegal methadone use; 0 = no heroin or illegal methadone use).

^gLifetime illicit polydrug use (1 = ≥2 illicit substances [stimulants, cocaine/crack, heroin/illegal methadone, hallucinogens, club drugs]).

Latent Class Identification of Psychosocial Stressors

Allowing for local dependencies, the three-class model of psychosocial stressors (i.e., with the addition of residual associations) had the lowest AIC, BIC, and aBIC (see Supplemental Table 1). Entropy was relatively high in the two-class model allowing for local dependencies (Bakk et al., 2013). The three-class model of stressors allowing for local dependencies had adequate class separation (Nylund et al., 2007). Although the 2-class solution was the most parsimonious, the 3-class solution suggested a well-defined latent class characterized by high probability of ASA, which was not present in the 2-class solution. Some researchers have found that sexual abuse is a stronger predictor of substance use than physical abuse among women in the general population (Lown et al., 2011; Moustafa et al., 2021). Other studies have shown that CPA and childhood neglect have distinct effects on AOD outcomes among women (Wilsnack et al., 2018). As such, the 3-class solution was deemed optimal to distinguish variation in SMW’s AOD use.

Probabilities of psychosocial stressor indicators across latent classes are shown in Supplemental Figure 1 and in Table 2. Class 1 (Low Likelihood of Psychosocial Stressors) included slightly more than half of the sample (n = 336; 55.9%; see Table 2). SMW who belonged to Class 1 were likely to report low adversity given below-average probabilities of psychosocial stressors; many probabilities were close to zero (e.g., childhood neglect). Although lower than in Classes 2 and 3, SMW in Class 1 had a moderate probability of CPA. Class 1 was also characterized by the highest probability of low levels of minority stressors.

Table 2.

Descriptive Statistics and Probabilities of Indicators (N = 601).

		Latent Classes of Psychosocial Stressors
		Class 1b	Class 2c	Class 3d
10 Psychosocial Stressor Indicators	n (%)	Conditional Item Probabilities
Childhood physical abuse	348 (57.8)	0.44	0.84	0.55
Childhood neglect	59 (9.8)	0.01	0.24	0.14
Childhood sexual abuse severity
No CSA	266 (44.2)	0.60	0.22	0.33
Noncontact childhood sexual abuse	70 (11.6)	0.13	0.10	0.12
Low-risk contact childhood sexual abuse	87 (14.5)	0.11	0.18	0.17
High-risk contact childhood sexual abuse	179 (29.7)	0.16	0.50	0.38
Adult sexual assault	180 (29.9)	0.08	0.30	0.99
Sexual intimate partner violence	76 (10.9)	0.01	0.01	0.62
Adult physical assault	201 (33.4)	0.14	0.59	0.48
Physical intimate partner violence	177 (29.4)	0.21	0.39	0.39
Discriminationa
Low discrimination	349 (50.2)	0.55	0.41	0.50
Medium discrimination	157 (22.6)	0.23	0.25	0.24
High discrimination	184 (26.5)	0.22	0.34	0.26
Stigma consciousnessa
Low stigma consciousness	222 (31.9)	0.40	0.25	0.29
Medium stigma consciousness	208 (29.9)	0.31	0.30	0.31
High stigma consciousness	246 (35.4)	0.29	0.45	0.41
Internalized stigmaa
Low internalized stigma	274 (39.4)	0.41	0.38	0.34
Medium internalized stigma	157 (22.6)	0.23	0.23	0.23
High internalized stigma	256 (36.8)	0.36	0.39	0.43

^aTertiles were used to calculate discrimination, stigma consciousness, and internalized stigma.

^bClass 1: “Low Likelihood of Psychosocial Stressors”; n = 336; 55.9%.

^cClass 2: “High Likelihood of Childhood Physical and Sexual Abuse and Adult Physical Abuse and Moderate Likelihood of Minority Stressors”; n = 175; 29.1%.

^dClass 3: “Moderate Likelihood of Childhood Physical Abuse, High Likelihood of Adult Sexual Assault and Sexual IPV, and Moderate Likelihood of Minority Stressors”; n = 90; 15.0%.

Class 2 (High Likelihood of CSA, CPA and Adult Physical Abuse, and Moderate Likelihood of Minority Stress) included 175 (29.1%) participants. It was characterized by higher adversity than Class 1, as demonstrated by elevated probabilities of experiencing some but not all types of trauma. Participants in this class were likely to report CPA, have high-risk CSA, and likely to report adult physical assault and physical IPV. This class was distinguished by moderate probabilities of high levels of discrimination, stigma consciousness, and internalized stigma.

Class 3 (Moderate Likelihood of CPA, High Likelihood of ASA and Sexual IPV, and Moderate Likelihood of Minority Stressors) included 90 (15.0%) of participants. This was the only class characterized by elevated probabilities of ASA. This class was marked by above-average probability of CPA, although lower than Class 2. SMW in Class 3 had lower probabilities of childhood neglect and sexual abuse and adult physical assault, and a similar probability of adult physical IPV compared to Class 2. SMW in Class 3 had low probabilities of discrimination but moderate probabilities of high stigma consciousness and internalized stigma.

Sociodemographic Characteristics and AOD Outcomes as Correlates of Class Membership

Older SMW were more likely than younger SMW to be in Class 3 than Class 1 (see Table 3). Black SMW were more than three times as likely as White SMW to be in Class 2 as in Class 1. SMW in Class 2 reported lower levels of social support than SMW in Class 1. Latent classes did not vary by sexual identity, income, relationship status, or geographical location.

Table 3.

Multinomial Logistic Regression Model Assessing Sociodemographic Characteristics Associated With Latent Classes of Psychosocial Stressors Among Sexual Minority Women.

Sociodemographic characteristics	Class 2a			Class 3b
	aOR [95% CI]		FDR-Adjusted p-Value	aOR [95% CI]		FDR-Adjusted p-Value
Age	1.02 [0.99, 1.05]		.135	1.04 [1.02, 1.06]		.002
Sexual identity
Lesbian	Ref			Ref
Bisexual	0.75 [0.35, 1.60]		.264	1.34 [0.61, 2.96]		.264
Race/ethnicity
White	Ref			Ref
Black	3.07 [1.32, 7.13]		.014	0.81 [0.36, 1.79]		.319
Latinx	1.24 [0.51, 3.06]		.183	0.79 [0.38, 1.67]		.343
Income
Annual income over $14,999	Ref			Ref
Annual income $14,999 or less	0.60 [0.29, 1.27]		.225	1.18 [0.53, 2.60]		.224
Relationship status
Not in a relationship	Ref			Ref
In a relationship	0.71 [0.37, 1.37]		.224	0.70 [0.36, 1.34]		.237
Geographical location
Medium-sized city or larger	Ref			Ref
Town, village, or small city	1.97 [0.61, 6.33]		.127	0.35 [0.04, 3.28]		.177
Social support	0.95 [0.92, 0.98]		<.001	0.98 [0.95, 1.01]		.135

Note. All models used the bias-adjusted maximum likelihood approach and pairwise deletion. FDR-adjusted refers to the Benjamini-Hochberg procedure used to correct for the false discovery rate. Boldface type indicates a significant aOR. Omitted (reference) category is Class 1 (“Low Likelihood of Psychosocial Stressors”; n = 336; 55.9%).

aOR = adjusted odds ratio; CI = confidence interval; Ref = reference group.

^aClass 2: “High Likelihood of Childhood Physical and Sexual Abuse and Adult Physical Abuse and Moderate Likelihood of Minority Stressors”; n = 175; 29.1%.

^bClass 3: “Moderate Likelihood of Childhood Physical Abuse, High Likelihood of Adult Sexual Assault and Sexual IPV, and Moderate Likelihood of Minority Stressors”; n = 90; 15.0%.

After adjusting for sociodemographic characteristics, SMW in Classes 2 and 3 were more likely than SMW in Class 1 to recognize problematic drinking and to report heroin or illegal methadone use and illicit polydrug use (see Table 4). SMW in Class 3 were more likely to meet criteria for DSM-IV alcohol dependence. SMW’s HED or cannabis use did not vary by classes.

Table 4.

Multinomial Logistic Regression Models Assessing Psychosocial Stressors Latent Classes as Predictors of Alcohol and Other Drug Use and Negative Outcomes Among Sexual Minority Women.

	Heavy Episodic Drinkinga		DSM-IV Alcohol Dependenceb		Alcohol Problem Recognitionc		Cannabis Used		Heroin or Illegal Methadone Usee		Illicit Polydrug Usef
		FDR-Adjusted		FDR-Adjusted		FDR-Adjusted		FDR-Adjusted		FDR-Adjusted		FDR-Adjusted
	aOR [95% CI]	p-Value	aOR [95% CI]	p-Value	aOR [95% CI]	p-Value	aOR [95% CI]	p-Value	aOR [95% CI]	p-Value	aOR [95% CI]	p-Value
Sample size	n = 545		n = 545		n = 545		n = 4468		n = 546		n = 546
Psychosocial stressors class
Class 1g	Ref		Ref		Ref		Ref		Ref		Ref
Class 2h	1.00 [0.45, 2.24]	.999	3.13 [0.82, 11.89]	.132	3.49 [1.49, 8.19]	.020	0.66 [0.26, 1.71]	.600	15.14 [7.09, 32.34]	<.001	6.23 [2.30, 16.88]	<.001
Class 3i	1.76 [0.39, 1.76]	.950	4.72 [1.29, 15.00]	.028	3.27 [1.60, 6.67]	.011	1.72 [0.73, 4.05]	.600	4.59 [1.93, 10.93]	<.001	2.69 [1.13, 6.41]	.063
Soclodemographic characterlstlcs
Age	0.94 [0.93, 0.96]	<.001	0.96 [0.94, 0.99]	<.001	1.01 [0.99, 1.02]	.286	0.93 [0.91, 0.95]	<.001	1.07 [1.05, 1.10]	<.001	1.01 [0.99, 1.02]	.425
Sexual identlty
Lesblan	Ref		Ref		Ref		Ref		Ref		Ref
Bisexual	1.068 [0.70, 1.66]	.950	1.23 [0.65, 2.32]	.578	1.19 [0.76, 1.88]	.563	0.86 [0.53, 1.41]	.630	2.22 [1.10, 4.51]	.045	1.38 [0.81, 2.33]	.329
Race/ethnlcity
White	Ref		Ref		Ref		Ref		Ref		Ref
Black	1.04 [0.66, 1.62]	.967	3.14 [1.43, 6.91]	.027	0.62 [0.39, 0.98]	.100	1.23 [0.74, 2.02]	.540	0.94 [0.46, 1.93]	.860	0.42 [0.24, 0.73]	.011
Latinx	1.28 [0.79, 2.07]	.950	2.26 [0.94, 5.45]	.115	0.58 [0.34, 0.97]	.100	0.85 [0.49, 1.48]	.590	2.14 [0.97, 4.75]	.087	0.48 [0.26, 0.88]	.060
Income
Annual income over $14,999	Ref		Ref		Ref		Ref		Ref		Ref
Annual income $14,999 or less	0.94 [0.60, 1.45]	.950	1.38 [0.74, 2.56]	.387	0.89 [0.56, 1.39]	.667	1.35 [0.82, 2.23]	.600	4.60 [2.31, 9.15]	<.001	0.90 [0.53, 1.53]	.700
Relationship status
Not in a relationship	Ref		Ref		Ref		Ref		Ref		Ref
In a relationship	0.70 [0.47, 1.03]	.345	0.53 [0.29, 0.96]	.087	1.06 [0.71, 1.60]	.750	0.89 [0.57, 1.39]	.630	1.23 [0.66, 2.28]	.567	0.68 [0.42, 1.08]	.200
Geographical location
Medium-sized city or larger	Ref		Ref		Ref		Ref		Ref		Ref
Town, village, or small city	0.71 [0.29, 1.75]	.950	1.19 [0.30, 4.68]	.800	1.73 [0.75, 4.02]	.286	0.62 [0.20, 2.02]	.600	0.27 [0.06, 1.21]	.110	0.48 [0.15, 1.44]	.317
Social support	0.99 [0.98, 1.01]	.950	0.98 [0.96, 1.00]	.115	0.99 [0.97, 1.00]	.220	0.62 [0.20, 2.02]	.600	1.07 [1.04, 1.09]	<.001	1.00 [0.99, 1.02]	.700

Note. FDR-adjusted refers to the Benjamani–Hochberg procedure used to correct for the false discovery rate. aOR = adjusted odds ratio; CI = confidence interval; Ref = reference group. Boldface type indicates a significant aOR.

^aPast-year heavy episodic drinking (1 = ≥6 drinks on one or more days in the past year or ≥4 drinks on a typical drinking day; 0 = <6 drinks on one or more days in the past year and <4 drinks on a typical drinking day).

^bPast-year DSM-IV alcohol dependence (1 = ≥3 DSM-IV symptoms of alcohol dependence; 0 = <3 DSM-IV symptoms of alcohol dependence; Hasin et al., 2003).

^cLifetime alcohol problem recognition (1 = wondered if developing a drinking problem; 0 = never wondered if developing a drinking problem).

^dPast-year cannabis use (1 = cannabis use; 0 = no cannabis use). Sexual minority women who reported no lifetime cannabis use were excluded in model examining psychosocial stressors latent classes as predictors of past-year cannabis use.

^eLifetime heroin or illegal methadone use (1 = heroin or illegal methadone use; 0 = no heroin or illegal methadone use).

^fLifetime illicit polydrug use (1 = ≥2 illicit substances; 0 = <2 illicit substances).

^gClass 1: “Low Likelihood of Psychosocial Stressors”; n = 336; 55.9%.

^hClass 2: “High Likelihood of Childhood Physical and Sexual Abuse and Adult Physical Abuse and Moderate Likelihood of Minority Stressors”; n = 175; 29.1%.

ⁱClass 3: “Moderate Likelihood of Childhood Physical Abuse, High Likelihood of Adult Sexual Assault and Sexual IPV, and Moderate Likelihood of Minority Stressors”; n = 90; 15.0%.

Our post hoc sensitivity analysis comparing LCA results to a variable-centered approach while adjusting for sociodemographic characteristics indicated that SMW who experienced various forms of trauma, compared to those who did not, were more likely to report alcohol problem recognition, heroin or illegal methadone use, and illicit polydrug use (see Supplemental Table 2). SMW who experienced adult physical assault in general and in dating relationships as well as ASA in dating relationships and elevated levels of discrimination were uniquely at risk for DSM-IV alcohol dependence. SMW who reported sexual IPV compared to those who did not were more likely to report past-year cannabis use.

Discussion

The current study extends prior research (Feinstein & Dyar, 2017; McCabe, Hughes, et al., 2022; Puckett et al., 2017; Shramko et al., 2018) by documenting unique patterns of psychosocial stressors among SMW and heterogeneity in SMW’s risk of endorsing psychosocial stressors. In addition, building on prior findings (Kassing et al., 2021; Keyes et al., 2011; Layland et al., 2022; Talley et al., 2022), these results are among the first to show that exposure to multiple types of psychosocial stressors is associated with AOD outcomes among SMW.

We examined 10 distinct experiences of trauma exposure and minority stressors. Two latent classes (Classes 2 and 3) exhibited higher adversity relative to Class 1, which was characterized by low (but not zero) probabilities of experiencing psychosocial stressors (i.e., lower adversity). Even though SMW in Class 1 were less likely to report psychosocial stressors than SMW in other classes, they had a moderate probability of reporting CPA, reinforcing evidence that SMW are vulnerable to this form of childhood maltreatment (McCabe, Hughes, et al., 2022).

SMW in Class 2, who reported CPA, severe CSA, and ASA, were distinctively vulnerable to discrimination and stigma consciousness (Pinel, 1999). Additional research is needed to examine mechanisms underlying this co-occurrence, such as exteroception (e.g., perception of threat; Adhia et al., 2021). Our findings also noted a distinct subgroup of SMW (Class 3) at heightened risk of CPA, ASA, and stigma consciousness. Classes 2 and 3 build on prior research in a sample of sexual minority Latinx youth that revealed two latent profiles characterized by discrimination and victimization and related to poor mental health (Shramko et al., 2018).

This study’s post hoc results expand variable-centered research that has primarily focused on sexual revictimization among CSA-exposed SMW (Sutton et al., 2021). Our person-centered results support and add to prior research (Layland et al., 2022) by showing the combined burden of multiple types of adversity on SMW’s AOD outcomes. Our findings also highlight the importance of addressing the intersection of childhood and adult trauma, discrimination, and stigma consciousness, rather than focusing on separate, additive experiences across type.

Latent classes of psychosocial stressors varied among SMW by age, race/ethnicity, and social support. Aligned with prior findings (Calabrese et al., 2015; Ehlke et al., 2020), older SMW, Black SMW, and SMW with lower social support were more likely than younger SMW, White SMW, and SMW with higher social support, respectively, to be in classes characterized by high probabilities of childhood and adult maltreatment and minority stressors. Older SMW may be uniquely burdened by CPA, ASA, and stigma consciousness. Furthermore, Black SMW appear to be especially vulnerable to polyvictimization and minority stressors, potentially rooted in intersectional oppression (e.g., racism, heterosexism, sexism; Bowleg et al., 2003; Calabrese et al., 2015). Our findings are consistent with prior research noting that SMW who lack strong social support networks may be at heightened risk of psychosocial stressors (Ehlke et al., 2020).

Findings also highlight that SMW in classes marked by moderate to high probabilities of trauma exposure and minority stressors were between 2 and 15 times as likely to report AOD use and negative AOD outcomes as SMW in the class marked by lower probabilities of trauma and minority stressors AOD use. Even after accounting for sociodemographic differences, SMW who reported high adversity were more likely than those with low adversity to meet criteria for DSM-IV alcohol dependence, recognize problematic drinking, and to report heroin or illegal methadone use and illicit polydrug use. These findings provide preliminary evidence of the stress-sensitizing effects of early and cumulative adversity and benefits of utilizing LCA to formulate a life course, context-sensitive, and cumulative-effect approach to conceptualizing SMW’s vulnerability (Nurius et al., 2013). Notably, SMW’s likelihood of reporting HED and cannabis use did not vary across latent classes.

Our variable-centered findings demonstrated links between various psychosocial stressors and AOD outcomes and are consistent with our heterogeneous stressor latent classes that predicted within-group variation in AOD outcomes among SMW. Unlike our LCA results, this variable-centered approach identified past-year cannabis use among SMW who reported ASA in dating relationships. However, differentiating AOD outcomes based on isolated psychosocial stressors, rather than considering multiple stressors simultaneously, ignores the potential synergistic effect of trauma exposure and minority stressors on SMW’s AOD use.

Implications and Future Directions

An integrated public health approach is also needed to prevent structural drivers of SMW’s repeated exposure to trauma and minority stressors (Cardona et al., 2021; Decker et al., 2018). For instance, policies that reduce inequity, address structural disadvantage, and establish norms that address attitudes toward violence against SMW should be implemented (Decker et al., 2018; Flentje et al., 2022; Scheer et al., 2022). Social marketing approaches that challenge heterosexist norms, increase public awareness of SMW’s increased risk of trauma exposure and minority stressors, and mobilize the broader sexual and gender minority community as important safety nets, supports, and allies in violence prevention represent other important structural initiatives to reducing rates of psychosocial stressors facing SMW (Decker et al., 2018; Orchowski et al., 2020; Woodford et al., 2018). Prior research has also highlighted the importance of addressing SMW’s repeated exposure to trauma, minority stressors, and insufficient social safety by developing and implementing macro-level interventions, including increasing anti-discrimination protections via state-wide and federal policies (Diamond & Alley, 2022; Puckett et al., 2020).

Results from this study highlight the need for universal screening for lifetime adversity, including trauma and minority stressors, and AOD use and related problems among SMW. In addition, clinicians might consider employing functional analysis to identify proximal antecedents (e.g., negative emotionality, impulsivity, emotion dysregulation, dissociation, posttraumatic stress disorder symptoms) to negative AOD outcomes among SMW who report multiple psychosocial stressors (Boness & Witkiewitz, 2022; Dworkin et al., 2021; Fitzpatrick et al., 2020). Clinicians should also assess distress resulting from these psychosocial stressors to advance personalized delivery of substance use treatment for SMW. Depending on the source of SMW’s distress, clinicians might prioritize: (1) cognitive restructuring skills to help SMW modify AOD outcome expectancies related to internalized stigma; or (2) distress tolerance skills to help SMW cope with AOD craving following exposure to trauma-related stimuli (Boness & Witkiewitz, 2022). This modularized approach could increase the effectiveness of AOD treatment for SMW (Boness & Witkiewitz, 2022). Moreover, given that intimate relationships and social support protect against SMW’s negative health behaviors and outcomes (Caceres et al., 2019; Veldhuis et al., 2019, 2020), clinicians and intervention researchers should consider bolstering SMW’s access to strong support networks and enhance skills among those supporting SMW. For instance, trauma-informed social environments, including romantic or dating partners, SMW community members, caregivers, and mentors, may support SMW in coping with psychosocial stressors in ways that promote recovery and resilience (Matlin et al., 2019).

Future studies can build on these results by identifying risk pathways, such as neurobiological mechanisms and affective states, to inform etiology and maintenance models of AOD outcomes among SMW. For example, recent research using a sample of college drinkers found that drinking-to-cope motives increase drinking urges immediately following exposure to trauma cues (Zaso & Read, 2020), highlighting the need for future research to examine momentary affective processes driving trauma- and minority stress-related AOD use among SMW. Future studies should examine whether substance use motives or mental health needs influence SMW’s likelihood of engaging in HED and using cannabis following psychosocial stressors. Additional research is also needed to examine the connections between SMW’s risk of psychosocial stressors and broader sociopolitical context (Veldhuis et al., 2018).

Limitations

It is unknown whether our results generalize to SMW who identify as other than lesbian or bisexual or who live outside of Chicago Metropolitan Area in Illinois. Furthermore, data were collected in 2010 to 2012. Newer studies should determine whether findings can be replicated perhaps particularly due to narrowing gender differences in alcohol use (White, 2020) and given the shifting landscape of both supports for sexual and gender minority people (e.g., Obergefell, improved societal attitudes toward sexual and gender minority individuals; Aksoy et al., 2020; Drabble et al., 2020, 2021) and heightened structural stigma against sexual and gender minority people (e.g., anti-LGBT legislation, rise in hate crimes; Coston, 2020; Fredrick et al., 2022; Horne et al., 2022). Data were cross-sectional, and not all measures used consistent timeframes. However, findings advance foundational knowledge that can be replicated in longitudinal research. We used retrospective self-report measures and several measures were single-item indicators. In addition, this study did not include measures of motives for alcohol or other drug use. LCA results are limited by the latent class indicators included in the parent study. Future studies should consider including other factors, such as resilience, among SMW (Layland et al., 2022). Furthermore, gender-diverse SMW report unique AOD-related risks. Future research is needed to examine whether our findings generalize to SMW across gender identity and expression (Batchelder et al., 2022).

Conclusions

Three distinct classes of psychosocial stressors emerged among SMW and these varied by age, race/ethnicity, and social support. SMW who were in the higher adversity classes had heightened risks for negative AOD outcomes, such as alcohol dependence, heroin or illegal methadone use, and illicit polydrug use. Results highlight the need for universal screening for lifetime trauma exposure, minority stressors, and AOD use and related problems among SMW. Future studies should test modifiable mechanisms (e.g., coping strategies) underlying these associations to inform brief screening protocols and evidence-based adapted interventions for at-risk SMW. Moreover, public health prevention and intervention efforts are needed to mitigate structural drivers of trauma exposure and minority stressors, which are disproportionately shouldered by SMW and contribute to AOD use and negative outcomes in this population.

Biographies

Jillian R. Scheer, PhD, is a licensed counseling psychologist, the Cobb-Jones Professor of Clinical Psychology, and an Assistant Professor in the Department of Psychology at Syracuse University. Dr. Scheer’s research focuses on understanding the synergistic effect of violence exposure and minority stressors on sexual minority women’s hazardous drinking and co-occurring mental health outcomes and developing tailored interventions for this population.

Melanie M. Wall, PhD, is the director of Mental Health Data Science in the Psychiatry Department of the New York State Psychiatric Institute (NYSPI)/Columbia University Irving Medical Center (CUIMC), where she oversees a team of biostatisticians collaborating on predominately National Institutes of Health (NIH)-funded research projects related to psychiatry. She has worked extensively with modeling complex multilevel and multimodal data on a wide array of psychosocial public health and psychiatric research questions in both clinical studies and large epidemiologic studies.

Cindy B. Veldhuis, PhD, is an Assistant Professor in the Department of Medical Social Sciences at Northwestern Feinberg School of Medicine and in the Institute for Sexual and Gender Minority Health and Wellbeing at Northwestern University. She is a K99/R00 Pathway to Independence awardee from the National Institutes of Health. Broadly, her research focuses on LGBTQIA+ women’s relationships, health, and health behaviors as well as violence, trauma, mental health, and the impacts of sociopolitical events on wellbeing.

Jessie V. Ford, PhD, is a Public Health Sociologist and Assistant Professor in the Department of Sociomedical Sciences at Columbia University’s Mailman School of Public Health. Dr. Ford’s research explores how expectations and inequalities around gender and sexuality shape sexual health, violence, and pleasure. At present, she is working on a series of research projects examining the high rates of sexual assault associated with hazardous drinking among bisexual women.

Cory J. Cascalheira, BA, is a doctoral student of counseling psychology at New Mexico State University and a research project coordinator at Syracuse University. Cascalheira’s research aims to (1) examine stigma-based individual, interpersonal, and structural mechanisms conferring risk for SGM health disparities (e.g., substance misuse); and (2) apply computational methods (i.e., artificial intelligence) to expand the scientific understanding of theory-driven constructs (e.g., minority stress) pertinent to SGM wellbeing.

Emily C. Helminen, MS, is a doctoral candidate in School Psychology at Syracuse University. Helminen’s research focuses on (1) examining how compassion-based interventions can improve mental and behavioral health among sexual and gender minority populations, and (2) understanding the physiological and cognitive mechanisms by which these interventions are effective.

Thomas J. Shaw, BA, is doctoral student at Virginia Polytechnic Institute and State University. Shaw’s research interests pertain to (1) better understanding the pathways and manifestation of bidirectional intimate partner violence (IPV) in young adult couples; and (2) creating a more nuanced understanding of how risky behaviors such as alcohol use, heavy episodic drinking, and heavy cannabis are associated with IPV perpetration and victimization.

Virinca Jaipuriyar, BA, is a Research Assistant at the Minority Stress & Trauma Lab. Jaipuriyar’s research interests concern intervention-focused research from a social, cognitive, and neurobiological and psychobiological perspective in psychological symptoms relating to interpersonal stressors, anxiety, and substance use.

Michelle J. Zaso, PhD, is a postdoctoral associate at the University at Buffalo—State University of New York. She received her doctorate in clinical psychology from Syracuse University. Dr. Zaso’s major research interests include etiologies of alcohol use across development, with particular attention to the role of alcohol-promoting stressful and traumatic experiences.

Tonda L. Hughes, PhD, RN, FAAN, is the Henrik H. Bendixen Professor of International Nursing, and Associate Dean for Global Health in the School of Nursing at Columbia University in New York City. She is also Executive Director of the Center for Sexual and Gender Minority Health Research in the School of Nursing and holds an interdisciplinary appointment as Professor in the College of Medicine, Department of Psychiatry at Columbia. She has a distinguished research career focusing on women’s mental health and substance use and is recognized internationally as a leading expert on sexual minority women’s health.