Mental health impact of multiple sexually minoritized and gender expansive stressors among LGBTQ+ young adults: a latent class analysis

Aims In the United States, lesbian, gay, bisexual, transgender, queer, intersex, asexual and other sexually minoritized and gender expansive (LGBTQ+) young adults are at increased risk for experiencing mental health inequities, including anxiety, depression and psychological distress-related challenges associated with their sexual and gender identities. LGBTQ+ young adults may have unique experiences of sexual and gender minority-related vulnerability because of LGBTQ+-related minority stress and stressors, such as heterosexism, family rejection, identity concealment and internalized homophobia. Identifying and understanding specific LGBTQ+-related minority stress experiences and their complex roles in contributing to mental health burden among LGBTQ+ young adults could inform public health efforts to eliminate mental health inequities experienced by LGBTQ+ young adults. Therefore, this study sought to form empirically based risk profiles (i.e., latent classes) of LGBTQ+ young adults based on their experiences with familial heterosexist experiences, LGBTQ+-related family rejection, internalized LGBTQ+-phobia and LGBTQ+ identity concealment, and then identify associations of derived classes with psychological distress. Methods We recruited and enrolled participants using nonprobability, cross-sectional online survey data collected between May and August 2020 (N = 482). We used a three-step latent class analysis (LCA) approach to identify unique classes of response patterns to LGBTQ+-related minority stressor subscale items (i.e., familial heterosexist experiences, LGBTQ+-related family rejection, internalized LGBTQ+-phobia and LGBTQ+ identity concealment), and multinomial logistic regression to characterize the associations between the derived classes and psychological distress. Results Five distinct latent classes emerged from the LCA: (1) low minority stress, (2) LGBTQ+ identity concealment, (3) family rejection, (4) moderate minority stress and (5) high minority stress. Participants who were classified in the high and moderate minority stress classes were more likely to suffer from moderate and severe psychological distress compared to those classified in the low minority stress class. Additionally, relative to those in the low minority stress class, participants who were classified in the LGBTQ+ identity concealment group were more likely to suffer from severe psychological distress. Conclusion Familial heterosexist experiences, LGBTQ+-related family rejection, internalized LGBTQ+-phobia and LGBTQ+ identity concealment are four constructs that have been extensively examined as predictors for mental health outcomes among LGBTQ+ persons, and our study is among the first to reveal nuanced gradients of these stressors. Additionally, we found that more severe endorsement of minority stress was associated with greater psychological distress. Given our study results and the previously established negative mental health impacts of minority stressors among LGBTQ+ young adults, findings from our study can inform research, practice, and policy reform and development that could prevent and reduce mental health inequities among LGBTQ+ young adults.

LGBTQ+ university students experience elevated rates of minority stressors and psychological distress relative to non-LGBTQ+ students, with gender expansive university students experiencing unique psychological distress relative to their cisgender counterparts (Hunt et al., 2021;Woodford et al., 2018;Ploderl and Tremblay, 2015).Identifying and understanding LGBTQ+related minority stressor-specific experiences and their complex roles in contributing to mental health burden among LGBTQ+ young adults and university students could inform public health efforts to eliminate psychological inequities, such as depression, anxiety and psychological distress, among these populations.Psychological distress is more widely defined as emotional suffering, which can include depressive symptomology (e.g., unhappiness) and anxiety (e.g., feeling tense) symptoms, and physical suffering (e.g., insomnia, headaches and lack of energy) (Belay et al., 2021;Horwitz, 2002).
To better understand perceived oppression and stressors among LGBTQ+ young adults, the current study is guided by the Minority Stress Theory (Brooks, 1981;Meyer, 2003;Testa et al., 2015).The Minority Stress Theory emphasizes the role of externalized (e.g., discrimination-related occurrences due to LGBTQ+ identity) and internalized (e.g., negative personal feelings related to LGBTQ+ identity) minority stressors and their impact on mental health among LGBTQ+ people.Previous studies have examined these constructs as independent variable-level predictors on various mental health outcomes, such as depression, anxiety and psychological distress (Chodzen et al., 2019;Dyar et al., 2020;Inderbinen et al., 2021;McLean, 2021;Newcomb and Mustanski, 2010;Paceley et al., 2020;Pachankis et al., 2020;Puckett et al., 2018).However, as these constructs do not exist in a vacuum, and instead, interact with each other, it is important to examine their intersections through use of non-linear modelling techniques, such as latent class analysis (LCA; Masyn, 2013;Weller et al., 2020).Compared to the variable-centred general linear modeling, LCA, a person-centred model, could help elucidate the nuances of multiple LGBTQ+-related minority stress experiences, and how these experiences relate to psychological distress (Collins and Lanza, 2009;Richman and Lattanner, 2014).In other words, LCA can reveal hidden or unobservable groups of LGBTQ+ young adults based across their unique levels and combinations of multiple LGBTQ+-related minority stressors.
This study aimed to form empirically based risk profiles (i.e., latent classes) of LGBTQ+ young adults based on four minority stressors: familial heterosexist experiences, LGBTQ+-related family rejection, internalized LGBTQ+-phobia and LGBTQ+ identity concealment.Then, we aimed to identify which classes of LGBTQ+ young adults were at greater risk for experiencing psychological distress.We hypothesized that multiple latent classes would emerge from the data, and classes with higher minority stress would be associated with greater psychological distress.Findings from our study may inform research, practice, policy reform and development that could be used to prevent mental health challenges driven by minority stress among LGBTQ+ young adults.

Study design and sample
A nonprobability cross-sectional online survey was conducted between May 27 and August 14 2020.The aim of the parent study was to explore mental health and minority stress among LGBTQ+ university students.The University of Maryland provided institutional review board approval prior to study commencement.Additional study information can be found elsewhere (Salerno et al., 2023).
An electronic recruitment flyer with a link to an online selfadministered Qualtrics survey was distributed through popular social media platforms (i.e., Facebook, LinkedIn and Twitter) and email campaigning.Email campaigning included the university listservs of historically Black colleges and universities, Hispanic serving institutions and LGBTQ+ student centres across the U.S. Upon opening the survey online, participants completed a selfadministered electronic informed consent process.Participant eligibility criteria included: (1) being a full-time student attending a U.S. tertiary institution, (2) identifying as LGBTQ+ and (3) being age 18 years or older.Participants were incentivized with the option to be entered into a raffle for a $50 Amazon electronic gift card.

Measures
Socio-demographic characteristics Gender identity.Participants indicated whether they identified as a cisgender woman, a cisgender man, nonbinary, a transgender woman, transfeminine, a transgender man, transmasculine, genderqueer, two-spirit, gender fluid, agender or another gender.Gender was recoded and categorized as cisgender man (referent), cisgender woman, non-binary, or genderqueer (included two-spirit, gender fluid, agender or other), transgender man or transmasculine, and transgender woman or transfeminine.
Age. Participants indicated their age in years.
Social isolation.Social isolation was measured using the threeitem short loneliness scale (Hughes et al., 2004).Item responses were coded on a 3-point Likert-type scale consisting of 'hardly ever' (1), 'some of the time' (2) and 'often' (3).We calculated the mean score of items to assess social isolation ( = 0.758).
Race and ethnicity.Participants indicated their race (select all that apply) as American Indian or Alaskan Native; Native Hawaiian or other Pacific Islander; Asian; Black or African American; White; or another race not listed.Ethnicity was collected with the following yes/no question: 'Are you Hispanic or Latino?' Race and ethnicity were recoded and categorized as non-Hispanic White (referent); non-Hispanic Asian American Indian, non-Hispanic Alaskan Native, non-Hispanic Native Hawaiian or other Pacific Islander; non-Hispanic Black or African American; Latino or Hispanic, and multiracial or another race not listed.
LGBTQ+-related minority stress latent class indicator variables The LGBTQ+-related minority stress survey items can be found in Appendix 1.For the LCA, items were assessed individually.
Familial heterosexist experiences.An adapted version of seven items from the Daily Heterosexist Experiences Questionnaire (DHEQ) 'Family of Origin' subscale (Balsam et al., 2013) was used to assess past-year experiences of heterosexism perpetrated by family members.To capture presence of past year familial heterosexist experiences, students were asked to indicate if they experienced these stressors in the past year (yes = 1; no = 0).For descriptive analysis, a composite score was calculated by summing responses across the seven items ( = 0.729).
LGBTQ+-related family rejection.An adapted version of 10 items from the 'family rejection' subscale of the Sexual Minority Adolescent Sexual Minority Stress Inventory (Schrager et al., 2018) was used to measure past year LGBTQ+-related family rejection.To capture presence of past year LGBTQ+-related family rejection, students were asked to indicate if they experienced these stressors in the past year (yes = 1; no = 0).For descriptive analysis, a composite score was calculated by summing responses across the 10 items ( = 0.821).
Internalized LGBTQ+-phobia.An adapted version of seven items from the LGBT Minority Stress Measure (LMSM; Outland, 2016) was used to measure past year internalized LGBTQ+-phobia.To capture presence of past year internalized LGBTQ+-phobia, students were asked to indicate if they experienced these stressors in the past year (yes = 1; no = 0).A composite score was calculated by summing responses across the seven items ( = 0.801).
LGBTQ+ identity concealment.LGBTQ+ identity concealment within the past year was measured using an adapted version of three items from the LMSM (Outland, 2016) and four items from the DHEQ (Balsam et al., 2013).To capture presence of past year LGBTQ+ identity concealment, students were asked to indicate whether they experienced these stressors in the past year (yes = 1; no = 0).A composite score was calculated by summing responses across the seven items ( = 0.768).
Psychological distress.The previously validated 10-item Kessler-10 (K10) was used to measure current nonspecific psychological distress (Kessler et al., 2002).This 10-item scale provided measures of depression and anxiety within the past 30 days.Item responses were coded on a 5-point Likert-type scale from 'none of the time' (1) to 'all of the time' (5).There was strong internal consistency for psychological distress in the current sample ( = 0.801).Participants were classified as having 'healthy' (referent), 'mild' , 'moderate' or 'severe' psychological distress (Andrews and Slade, 2001;Slade et al., 2011).
Analytic framework.Using a person-centred approach, we used LCA to form empirically based risk profiles of LGBTQ+ young adults based on their response patterns to LGBTQ+ minority stress subscale items (i.e., familial heterosexist experiences, LGBTQ+-related family rejection, internalized LGBTQ+-phobia and LGBTQ+ identity concealment) (Fergusson et al., 2005).
A total of 31 binary minority stress variable items were assessed in the LCA.Using the poLCA package on the R environment (Lewis and Linzer, 2011;R Core Team, 2013), we conducted LCA with two to six classes.Due to sample size limitations, we opted not to include covariates in the LCA and instead utilize covariates in the multinomial logistic regression.We chose to use 30 repetitions to estimate the LCA model and used random matrices of classconditional response probabilities as the starting values.We set the LCA to run a maximum of 3,000 iterations.We used the following fit statistics to assess which model solution best fit our data: class sizes, intra-class correlations, average posterior probabilities, consistent Akaike's information criterion (cAIC), Bayes information criterion (BIC), Akaike's Bayes information criterion (aBIC), Lo-Mendell-Rubin likelihood ratio test (LMR), bootstrap likelihood ratio test (LRT) and entropy.Participants with missing data across the 31 minority stress items were removed from analyses (1.8% missing).
To test class differences between LCA class assignment and psychological distress, chi-square tests of association were used.To test for class differences between LCA class assignment and minority stress items, we used analysis of variance.To test for multivariable (adjusted for socio-demographic characteristics) associations between latent class assignment and psychological distress, we used multinomial logistic regression.Alpha was set to 0.05, and all bivariate and multivariable statistical models were conducted using the nnet package using R Statistical Software (R Core Team, 2013;Ripley et al., 2016).

Socio-demographic findings
A descriptive summary of total sample (N = 482) and class-specific socio-demographic characteristics are described in Table 1.Participants reported a mean age of 22 years, and most participants identified as cisgender women (54%), non-Hispanic (85%), non-Hispanic White (70%) and single (51%).

Latent class analysis
Latent class analysis model fit indices are reported in Table 2.All five classes in the five-class model demonstrated adequate sample sizes that met the suggested 10% of the total sample threshold (n = 119, n = 133, n = 109, n = 61, n = 60) (Sinha et al., 2021).The five-class model produced the lowest AIC, BIC and aBIC compared to all models and detected significant LMR compared to the four-class model, suggesting that this is the strongest model in-terms of cAIC, BIC, and aBIC (four-class model not further considered).The entropy of the five-class model was above the 0.80 recommended cut-off (Nylund-Gibson and Choi, 2018;Weller et al., 2020), suggesting composition of classes with strong separation (Nylund-Gibson and Choi, 2018).We then examined the five-class model for interpretability and discovered a meaningful pattern (Fig. 1).Therefore, the five-class model was determined to be conceptually interpretable, with strong model fit, and was selected as the final model for further analysis.
Because the five-class model is the most parsimonious model, participants were categorized into five distinct minority stress classes: low minority stress (n = 119), LGBTQ+ identity concealment only (n = 133), family rejection only (n = 109), moderate minority stress (n = 61) and high minority stress (n = 60).A descriptive summary of minority stressor means stratified by class membership is reported in Table 3 and visualized    (using composite scores).There were statistically significant differences in class membership and age (p < 0.001) and mean social isolation score (p < 0.001) based on minority stress latent class assignment.
The low minority stress class was characterized by participants having low conditional probability (i.e., Pr ∼ 0.1) of responding 'yes' to most items across all minority stress subscales.LGBTQ related family rejection LGBTQ identity concealment

Low
LGBTQ identity concealment Family rejection Moderate High  The LGBTQ+ identify concealment class was characterized by participants having moderately high conditional probability (i.e., 0.5 < Pr < 1) of responding 'yes' to the LGBTQ+ identity concealment items and moderately low conditional probability (i.e., Pr < 0.3) of responding 'yes' to all other items.
The family rejection only class was characterized by participants having a moderately low conditional probability (i.e., 0.1 < Pr < 0.3) of responding 'yes' to most items, with the exception of the LGBTQ+-related family rejection subscales in which participants were characterized by moderate conditional probability (i.e., 0.3 > Pr > 0.7) of responding 'yes' to the majority of items.
The moderate minority stress class was characterized by participants having a high (i.e., 0.5 > Pr > 1) conditional probability of responding 'yes' to the majority of items in the LGBTQ+-related family rejection and LGBTQ+ identity concealment subscales, a moderate probability of responding 'yes' to the majority familial heterosexist experiences items (i.e., Pr ∼ 0.4) and low conditional probability of responding 'yes' to the majority of items in the internalized LGBTQ+-phobia subscale (i.e., Pr ∼ 0.1).
The high minority stress class was characterized by participants having a high (i.e., 0.5 > Pr > 1) conditional probability of responding 'yes' to most items across all subscales.

Bivariate latent class group differences in psychological distress
Frequencies of participants' psychological distress levels stratified by class membership and bivariate associations between

Multivariable associations between latent class membership and psychological distress
Table 5 describes the results of the multinomial logistic regression analysis testing the multivariable associations between latent class membership and psychological distress (adjusting for sociodemographic characteristics).

Discussion
This study identified unique groups of LGBTQ+ young adults based on their differential experiences of LGBTQ+-related minority stress across five classes: low, LGBTQ+ identity concealment, family rejection only, moderate and high minority stress.Our hypothesis was partially correct; latent class membership was associated with severity of psychological distress, such that those in the moderate minority stress and high minority stress groups were consistently at increased risk for moderate and severe psychological distress compared to the low minority stress group, and the LGBTQ+ identity concealment group was at increased risk for severe psychological distress compared to the low minority stress group.However, our hypothesis was also partially incorrect: Asian, American Indian, or Native Hawaiian and Pacific Islander (compared to non-Hispanic White) LGBTQ+ university students were less likely to suffer from moderate psychological distress, and multiracial or another race identifying LGBTQ+ university students were more likely to suffer from severe psychological distress.
Our study reveals the salient impact of multiple LGBTQ+-related minority stress on psychological distress among LGBTQ+ young adults.Our study is among the first to demonstrate that nuanced gradients of minority stress were associated with greater likelihood of psychological distress among LGBTQ+ young adults.Findings around the high and moderate minority stress groups are consistent with existing literature documenting the negative effects of minority stress on LGBTQ+ young adults' mental health (Price-Feeney et al., 2020;Newcomb and Mustanski, 2010;Pachankis et al., 2020;Ryan et al., 2009;Testa et al., 2015) and suggest an additive or perhaps compacting or intersecting relationship, in which more minority stress correlates with greater magnitude of psychological distress, supporting our hypothesis.Yet, findings on the LGBTQ+ identity concealment group reveal that this stressor may have a particularly strong impact on mental health among young adults (Pachankis et al., 2020), even when other stressors such as family rejection, familial heterosexist experiences and internalized LGBTQ+-phobia are at lower levels.Lastly, the fact that the LGBTQ+-related family rejection only group did not demonstrate significance for any elevated level of psychological distress counters previous evidence documenting the salient impact of family rejection on the mental health of LGBTQ+ youth (Gattamorta et al., 2022;Klein and Golub, 2016;Mitrani et al., 2017;Ryan et al., 2009).This unexpected finding could relate to analytical, measurement or sample differences compared to past studies and calls for more nuanced conceptualization of LGBTQ+-related family rejection scales and more application of complex and non-linear models of minority stress.
We found that Asian, American Indian, and Native Hawaiian or Pacific Islander (compared to non-Hispanic White) identifying LGBTQ+ university students were less likely to experience moderate psychological distress.Our findings suggest the possibility that these populations are resilient in resolving psychological distress relative to their non-Hispanic White counterparts or perhaps are less likely to recognize their psychological distress due to the stigma surrounding mental health in these communities (Misra et al., 2021;Ng, 1997).We also discovered that LGBTQ+ university students who identified as multiracial or other another race were more likely to experience severe psychological distress compared to non-Hispanic White LGBTQ+ university students.This may be related to their unique intersectional experiences of discrimination and racism (Bowleg et al., 2003;Salerno et al., 2023) along with multiple LGBTQ+-related minority stress experiences.It is imperative for researchers to investigate the intersectionality of race, ethnicity and sexually minoritized and gender expansive identities to address mental health concerns among these populations.Future research is needed to understand how racial and ethnic identities create risk or resilience and how to leverage these identities to prevent and address LGBTQ+-related minority stressors and negative mental health outcomes among LGBTQ+ young people.
As this LCA has indicated, psychological health inequities are driven by LGBTQ+-related minority stressors.As such, it is important for treatment to address co-occurring, additive and compounding LGBTQ+-related minority stress.Recent research identified 44 individual-, interpersonal-, structural-and multi-level interventions developed to reduce sexual minority stressors and/or bolster coping resources and strategies (Chaudoir et al., 2017).For instance, Puckett and Levitt's general guidelines include aiming to understand LGBTQ+ clients' minority stress in the context of the oppressive U.S. system, not overattributing mental health symptoms to internalized stigma, and helping LGBTQ+ clients recognize when minority stress affects their mental health (Puckett and Levitt, 2015).Future minority stress and psychosocial distress reduction programs could examine how to tailor psychosocial and multicomponent strategies based on the intersections of multiple LGBTQ+-related minority stressors, such as those identified among participants in the moderate and high minority stress classes.Intersectionality framing of mental health and minority stress interventions for LGBTQ+ young adults could be beneficial in this regard (Huang et al., 2020).Interdisciplinary, public health and mental health scientists and practitioners are needed urgently to improve understanding of how to adapt existing culturally relevant resources to prevent mental illness driven by multiple LGBTQ+-related minority stress among LGBTQ+ young adults.
This study had several limitations.This study used a nonprobability sampling strategy, which limits our ability to generalize findings to broader populations of LGBTQ+ young adults.As a cross-sectional study, responses were subject to recall bias, and we were unable to test causality and temporality among constructs; such limitations are important to consider in the context of intervention development.Due to correlations between variables, sample size limitations, and the existing complexity of this analysis, we were unable to utilize an intersectional perspective to investigate the significance of multiply marginalized identities within LGBTQ+ young adults, which includes sexual identity.This should be addressed in future studies with additional resources and greater sample sizes.Lastly, our survey collected data surrounding LGBTQ+ young adults' experiences of minority stress during the COVID-19 pandemic, which may be a factor that influences our study findings, as participants may have been under greater stress during the pandemic, as seen in other studies (Wang et al., 2020).However, our results add to emerging literature, which indicates that LGBTQ+ people are experiencing mental health disparities and treatment access inequities since the start of the COVID-19 pandemic (Algarin et al., 2022;Kamal et al., 2021;Salerno and Boekeloo, 2022).Despite limitations, this study provides important public health implications to consider for the mental health of LGBTQ+ young adults.
Our study findings have important public health implications for LGBTQ+ young adults.First, mental health services could address intersecting and multidimensional LGBTQ+-related minority stress in their practices and policies to mitigate poor mental health among LGBTQ+ young adults (Huang et al., 2020).
Recommendations for mental health services include increasing access to LGBTQ+-affirming mental health care (Austin et al., 2018;Burton et al., 2019;Chaudoir et al., 2017;Cohen et al., 2018;Hughto et al., 2019;Pachankis, 2015;Ryan, 2009) and dissemination of resources useful for the prevention of LGBTQ+-related minority stress and psychological distress (Cohen et al., 2018;Diamond and Shpigel, 2014;Ryan, 2009;SAMHSA, 2014).Despite significant advancement in the development of mental health and minority stress reduction programs for LGBTQ+ youth (IOM, 2011;Romanelli and Hudson, 2017), there is a severe dearth of competent and affirming mental health services and providers equipped with the tools, resources and skills to meet the needs of LGBTQ+ young adults in the U.S. (Williams and Fish, 2020).This is highly concerning given that LGBTQ+ young adults are already less likely to use mental health services compared to their heterosexual and cisgender counterparts (Bourdon et al., 2020;Dunbar et al., 2017;Filice and Meyer, 2018;Progovac et al., 2018).Indeed, significant barriers in access to treatment among LGBTQ+ young adults have been identified (Chaudoir et al., 2017;IOM, 2011;Romanelli and Hudson, 2017).
To advance the prevention of severe mental health burdens driven by LGBTQ+-related minority stress, it is imperative to increase the wide implementation and dissemination of LGBTQ+ affirmative practice and to prepare a mental health workforce that is able to address the unique identity-related concerns of LGBTQ+ young adults.Familial heterosexist experiences, LGBTQ+-related family rejection, LGBTQ+ identity concealment and internalized LGBTQ+-phobia are four constructs that have been extensively examined under the minority stress theory to serve as predictors for mental health outcomes among LGBTQ+ persons.Given our study results and the previously established negative mental health impacts of minority stressors among LGBTQ+ young adults, findings from our study may inform research, practice and policies that could prevent and eliminate mental health inequities among LGBTQ+ young adults.Prevention interventions tailored for LGBTQ+ young adults suffering from psychological distress must address distinct classes and gradients of LGBTQ+-related minority stress.Supplementary material.The supplementary material for this article can be found at https://doi.org/10.1017/S2045796024000118. 1975, as revised in 2013.The study was approved by the University of Maryland, College Park Institutional Review Board.

Figure 1 .
Figure 1.Composite minority stress indicator variables stratified by latent class group.

Table 1 .
Sample socio-demographic and background characteristics stratified by latent class, N = 482 a All race/ethnicity categories other than Hispanic/Latino refer to non-Hispanic/Latinx participants.b Another race included Arab.c Another sexual identity includes questioning or heterosexual/straight.**indicates significance at the p < 0.01 level; ***indicates significance at the p < 0.001 level.

Table 2 .
Results of the latent class analysis enumeration and model fit indices for two to six classes = consistent Akaike's information criterion, BIC = Bayes information criterion, aBIC = Akaike's Bayes information criterion, BSLRT = bootstrap likelihood ratio test; LMR-LRT = Lo-Mendell-Rubin likelihood ratio test. cAIC

Table 3 .
Latent class analysis minority stressor indicator variables as composite scores, stratified by latent class, N = 482 ***indicates significance at the p < 0.001 level.

Table 5 .
Multivariable multinomial logistic regression testing the associations between latent class group and psychological distress All race/ethnicity categories other than Hispanic/Latino refer to non-Hispanic/Latinx participants.The bold values indicate the variable name/group. a