LGBT Identity,Untreated Depression,and Unmet Need for Mental Health Services by Sexual Minority Women and Trans-Identified People

Abstract

Background:

Previous studies have found that transgender, lesbian, and bisexual people report poorer mental health relative to heterosexuals. However, available research provides little information about mental health service access among the highest need groups within these communities: bisexual women and transgender people. This study compared past year unmet need for mental health care and untreated depression between four groups: heterosexual cisgender (i.e., not transgender) women, cisgender lesbians, cisgender bisexual women, and transgender people.

Materials and Methods:

This was a cross-sectional Internet survey. We used targeted sampling to recruit 704 sexual and gender minority people and heterosexual cisgendered adult women across Ontario, Canada. To ensure adequate representation of vulnerable groups, we oversampled racialized and low socioeconomic status (SES) women.

Results:

Trans participants were 2.4 times (95% confidence intervals [CI] = 1.6–3.8, p < 0.01) and bisexual people 1.8 times (95% CI = 1.1–2.9, p = 0.02) as likely to report an unmet need for mental healthcare as cisgender heterosexual women. Trans participants were also 1.6 times (95% CI = 1.0–27, p = 0.04) more likely to report untreated depression. These differences were not seen after adjustment for social context factors such as discrimination and social support.

Conclusion:

We conclude that there are higher rates of unmet need and untreated depression in trans and bisexual participants that are partly explained by differences in social factors, including experiences of discrimination, lower levels of social support, and systemic exclusion from healthcare. Our findings suggest that the mental health system in Ontario is not currently meeting the needs of many sexual and gender minority people.

Introduction

There is a growing body of evidence demonstrating that sexual minority people, including gay men, lesbians, and bisexual people, report poorer mental health outcomes relative to heterosexuals.^1
–3 Gender minority (i.e., transgender) people are also at substantially increased risk for depression,^4,5 completed suicide, suicide attempts, and other forms of self-harm,^6,7 relative to rates reported for the general population. For example, a recent study of more than 400 transgender people from across the province of Ontario found that 61.2% of male-to-female spectrum transgender people and 66.4% of female-to-male-spectrum transgender people reported current depression.^4,5 In this same study, past year suicidal ideation was reported by 36% of transgender people, and past year suicide attempt by 10%.⁸ Experiences of discrimination and social exclusion are important contributors to mental health disparities associated with minority sexual orientations and gender identities.^9

–13

There are well-established differences in both prevalence of mental health problems and patterns of mental health service utilization between men and women in the general population. The prevalence of depression is between 1.5- and 3-fold higher among women than among men,^14,15 and women are more likely to seek mental health services than men.^16,17 The increased risk for depression among women appears to be amplified in the context of sexual minority status: data from the Canadian Community Health Survey (CCHS, combined years 2003 and 2005) indicate that lifetime mood disorder diagnoses are significantly more common among sexual minority women than among heterosexual women (11.4% in lesbians vs. 25.2% in bisexual women vs. 7.7% among heterosexual women).¹⁸ These Canadian data are consistent with other international studies. For example, in a representative study of the Dutch population, lifetime prevalence of major depression among women with same-sex sexual partners was 44.2%, significantly higher than the rate of 20% reported among heterosexual women.¹⁹ The few studies that have examined bisexual women independently of lesbian-identified women have found that bisexual women have the poorest mental health outcomes of all sexual orientation groups.^20
–22 These data indicate that the burden of depression among sexual minority women is higher than the burden of depression among heterosexual women.

Considering the high rates of need, high rates of mental health service utilization would be anticipated for sexual minority women and transgender people. Indeed, available data suggest that sexual minority people, and particularly lesbians and gay men, are both more likely than heterosexuals to consult mental health professionals and to report barriers to mental health service access associated with their sexual orientation.^{9,18,23

–26} For example, in an American study of 67 sexual minority people with severe and persistent mental illness, 21.7% of sexual minority women reported being dissatisfied with their mental health services; only 7.2% of heterosexual women reported dissatisfaction.²⁷ However, very little research has examined rates and predictors of mental health service access among bisexual women and transgender people (the highest need groups) in particular. The limited available research on bisexual people's mental health services experiences indicates that widely held social beliefs about bisexuality (such as the belief that bisexuals cannot be monogamous and that bisexuality is not a stable sexual identity) may both create mental distress for bisexual people,²⁸ and impact access to culturally competent services.²⁹ Transgender people who seek mental healthcare for reasons other than gender reassignment may also face barriers, including refusal of treatment, poor quality of care, inappropriate attribution of their mental health concerns to their gender identity, or refusal to address them by their preferred pronoun.^30

–33

In summary, available research provides little information about mental health service access among the highest need groups within the sexual and gender minority communities: bisexual women and transgender people. A lack of access to inclusive mental health services may be a contributing factor in the elevated levels of depression, suicidality, and other mental health concerns among these populations. We aim to address this research gap by comparing past year unmet need for mental healthcare and untreated depression between four groups: heterosexual cisgender (i.e., not transgender) women, cisgender lesbians, cisgender bisexual women, and transgender people. We also examine variables associated with unmet need for mental healthcare and untreated depression in these groups.

Materials and Methods

This article is based on data from a cross-sectional Internet survey that was implemented as part of a mixed methods community-based study. The research was guided by a 10-member community advisory committee composed of self-identified lesbian, gay, bisexual, transgender (LGBT) individuals with experience of mental health service use and mental health professionals with expertise in serving the LGBT community. The committee met approximately twice annually for the duration of the study, and service user members of the advisory committee received an honorarium for their participation.

We used targeted convenience sampling to recruit self-identified sexual and gender minority people and heterosexual, cisgendered (nontransgender) adult women from across Ontario, with selection to ensure that we met target numbers for nine mutually exclusive groups that had different intersections of sociodemographic factors, including lower socioeconomic status, racialized identities, sexual and gender orientation, and disability status (Table 1). The group definitions were developed in collaboration with our community advisory group. Examples include (1) racialized, lower socioeconomic status (SES), LBGT women and (2) racialized, higher SES, cisgendered, heterosexual (CIS-HET) women, and so on. Consecutive participants were enrolled in each group until the target sample size for the group was reached. This stratified sampling strategy was used to ensure we had a diverse sample of women and to allow us to conduct future analyses that could examine the impact of intersecting marginalized identities (e.g., LGBT women living in poverty).

Table 1.

Matrix of Groups for Targeted Convenience Sampling Strategy

Racialized		Lower SES		LGBT		Group
Yes	No	Yes	No	Yes	No
X		X		X		1
X		X			X	2
X			X	X		3
X			X		X	4
	X	X		X		5
	X	X			X	6
	X		X	X		7
	X		X		X	8
Any of the above and identifies as living with a disability						9

LGBT, lesbian, gay, bisexual, transgender; SES, socioeconomic status.

In this article, we report on our findings as they relate to comparisons between various sexual and gender identity groups. The study was reviewed and approved by the Research Ethics Board of the Centre for Addiction and Mental Health in Toronto.

Participants

Inclusion criteria included Ontario residency, 18 years or older, and sufficiently fluent in English to understand the consent form and questionnaires. All women and all trans-identified (TRANS) participants were eligible. Participants who self-identified as cisgender (nontrans) men were excluded regardless of their sexual orientation. Sexual minority men were also excluded from the current study because a separate research partnership is examining mental health services use in the population of gay and bisexual men.

Recruitment

Our primary method of recruitment was through electronic posting of flyers advertising a study of “women's experiences with depression treatment in Ontario” through women's online communities, health and social service agencies known to serve LGBT people, and university campus LGBT and women's centers. To supplement electronic recruitment, hard copies of study flyers were mailed to key health and social service agencies. A research coordinator visited key community agencies to promote the project and facilitate recruitment of persons belonging to our target identity groups. All participants had the option of participating anonymously. Participants who chose to provide contact information received $10 compensation via mail in the form of a gift card. Those who participated anonymously had the option of donating their compensation to a registered charitable organization.

Data collection

Data were collected via an Internet-based survey that was programmed by an experienced programmer. Participants without private access to a computer could request a paper and pencil version of the survey. Key community agencies provided space for our research coordinator to bring in a laptop to facilitate participation.

Survey instrument

Most of the standard items in our survey were based on questions used in the CCHS.^34,35 Population-specific questions (e.g., gender identity, sexual orientation) were developed in consultation with our community advisory committee. These items had been piloted in our previous research.^36,37

The usability and technical functionality of the survey were tested extensively before recruitment by project staff, students, and members of the community advisory committee. The survey was posted on a website designed solely for the purpose of this research project. Eligibility criteria were integrated into the beginning of the questionnaire and ineligible participants were not permitted to complete the survey. To reduce the risk of multiple entries per participant, participants who partially completed the survey were registered via a login and password so they could continue completing the survey at a later time. Responses were automatically captured and transformed into a database for analysis using SPSS. The survey was launched on June 9, 2011, and data were collected until July 20, 2012.

Independent variables: sexual orientation and gender identity

Participants were asked to self-report their self-identified sexual orientation and gender identity in two “check all that apply,” 10-item questions that included an open probe for those whose orientations were not captured by the options provided (Appendix 1). These items were developed by our team and used in other published research³⁷ with continued refinement after input from our community advisory committee. On the basis of responses to those two questions, we developed a four-category self-identified gender identity and sexual orientation variable with the following exclusive categories (1) trans (2) bisexual or pansexual (BI/PAN), (3) lesbian, gay, or queer (LGQ), and (4) CIS-HET. To accomplish this, the following algorithm was applied: individuals who endorsed any gender-minority response item (e.g., trans woman, trans man, woman of trans experience, man of trans experience, genderqueer, two-spirit, intersex, other) were assigned to the trans category; individuals who endorsed any bisexual response item (bisexual, pansexual) and were not trans were assigned to the bisexual category; individuals who endorsed any LGQ response item [gay, two-spirit (in relation to sexual orientation), homosexual, asexual, queer, other], and were not BI or TRANS were assigned to the LGQ category; and the remaining individuals were assigned to the CIS-HET category.

Sociodemographic variables

We created a binary variable (RACIALIZED/NOT RACIALIZED) based on participant responses to a check all that apply 14-category item that asked about racial, ethnic, or cultural identity, using categories of ethnicity from Statistics Canada.³⁸ Participants who endorsed only “white (e.g., European background),” were assigned to the NOT RACIALIZED category, while other responses [(Aboriginal/First Nations, Arab, black African, black Caribbean, Chinese, Filipino, Japanese, Korean, Latin American, and Southeast Asian (e.g., Vietnamese, Cambodian, Malaysian, and Laotian), South Asian (e.g., East Indian, Pakistani and Sri Lankan), West Asian (e.g., Iranian and Afghan)] were assigned to the RACIALIZED category. Responses for participants who used the open probe category “I Identify as ____” were reviewed by members of the research team and assigned by consensus to the most appropriate category.

We created a binary socioeconomic status variable (LOW SES/NOT LOW SES) by assigning participants to the “LOW SES” category if they endorsed at least two of the following six characteristics: (1) not employed or on disability, (2) if employed, in a temporary or contract position, (3) sole source of income was from the Canada Pension Plan, disability benefits, social assistance, or employment insurance, (4) the household income divided by the number of individuals in the household is less than the 2007 low-income cutoff value, (5) to the self-reported percentage of household income spent on housing is ≥30%, or (6) the calculated percentage of household income spent on housing is ≥30%.

Need and use variables

We created a binary variable for current major depressive episode based on participants' responses to the well-validated PHQ-9.³⁹ Specifically, PHQ-9 was positive for a major depressive episode only if the respondent endorsed a 2-week history of either depression or anhedonia, and at least five depressive symptoms for more than half the days in the previous 2 weeks. This cutoff has a sensitivity of 88% and a specificity of 88% for major depression.³⁹

Global physical and mental health status was measured using validated global physical and mental health status self-report single-item scales.⁴⁰

Items related to mental health services and barriers to use were adapted from the CCHS 1.2: Mental Health and Well-Being.³⁴

Social context variables

We measured social support using the 24-item Social Provisions Scale.⁴¹ Reliability and validity have been well established.⁴² Scores range from 0 to 94, with higher scores indicating greater levels of perceived social support.

We measured perceived discrimination based on participants' responses to the Perceived Discrimination Scale (PDS).⁴³ The PDS queries whether possible experiences of discrimination have occurred (e.g., do you think you have ever been unfairly fired or denied a promotion), as well as the frequency of microaggressions—unfair treatment in everyday life (e.g., in your day to day life, how often do people act as if they are better than you?). To simplify the interpretation of the regression results, we created two binary variables that identified participants who endorsed having experienced at least three episodes of discrimination or at least three types of microaggressions either fairly or very often. We used these cutoffs because they have been significantly correlated with negative outcomes in previous research.^22,36

In collaboration with the community advisory committee, we derived a variable that identified individuals who reported a barrier to mental healthcare that was related to the concept of systemic exclusion. Our concept of systemic exclusion was informed by the literature on “cultural safety,” as it applies to the systemic exclusion of Indigenous people from the healthcare system.⁴⁴ Systemic exclusion was assigned the value “1” if participants had (1) reported an unmet need for mental healthcare AND (2) had endorsed any one of the following sentences: (a) “[I was] afraid to ask for help or of what others would think,” (b) “[I] had a bad experience with a mental health professional in the past,” (c) “[I have] a lack of trust in the medical system,” (d) “[I] could not find a provider who was knowledgeable about issues of sexual orientation,” (e) “I could not find a provider who was knowledgeable about issues of gender identity,” (f) “I could not find a provider who was knowledgeable about racial, cultural, or ethnic issues that are important to me,” (g) “I could not find a provider who was knowledgeable about issues related to social class/income level that are important to me,” (h) “I could not find a provider who was knowledgeable about some other aspects of my identity(ies) that are important to me.” Participants who reported an unmet need but did not endorse any of the abovementioned barriers to care were assigned the value = “0” for the systemic exclusion variable. Participants who did not report an unmet need were excluded from the analyses that included systemic exclusion.

Dependent variables

Measures for our dependent variables were adapted from items used in the CCHS 1.2: Mental Health and Well-Being.³⁴

We created a binary unmet need variable based on participants' responses to the question “During the past 12 months, was there ever a time when you felt that you needed help from any kind of support for your emotions or mental health but you didn't receive it?”

To measure self-reported untreated depression, we first created a single binary variable for respondents who endorsed either a 12-month history of depression (In the past 12 months, have you had two consecutive weeks or more when you felt consistently depressed or down, most of the day, nearly every day?) or anhedonia (In the past 12 months, have you had two consecutive weeks or more when you were much less interested in most things or much less able to enjoy the things you used to enjoy most of the time?). We created a binary variable for self-reported 12-month depression treatment based on responses to the question “In the past 12 months have you been treated for depression?” Respondents who endorsed 12-month depression or anhedonia and who reported that they had not been treated for depression were considered to have untreated depression.

Sample size

Our a priori sample size (n = 700) was calculated to meet the requirements of the qualitative portion of our mixed method study, which required 40 participants with a current episode of major depression. Because nonprobability web-based sampling in the LGBT population can result in samples that are overrepresented by nonracialized individuals of higher socioeconomic status,⁴⁵ our community advisory committee requested that we used a panelized sampling approach to enable us to answer questions related to the impact of multiple intersections of identity categories (i.e., income, sexual and gender identity, and racialization). Consequently, the recruitment was continued until we had achieved a minimum of 100 participants in each of the following subgroups: (1) racialized, low SES, LGBT; (2) not racialized, low SES, LGBT; and (3) racialized, not low SES, LBGT; and a minimum of 50 participants in each of the following subgroups: (4) racialized, low SES, not LGBT; (5) racialized, not low SES, not LGBT; (6) not racialized, low SES, not LGBT; (7) not racialized, not low SES, LGBT; and (8) not racialized, not low SES, not LGBT. A power analysis for a logistic regression demonstrated that with our projected sample size, we would be able to demonstrate an odds ratio of 1.5 with statistical significance for our two outcomes (assuming alpha of 0.5 and beta of 0.20).⁴⁶

Analysis

We described respondents' demographic characteristics by gender and sexual orientation group. Next, we conducted bivariate analyses to describe rates of outcomes and identify potential associations between gender and sexual orientation category and our outcomes. For these analyses, we used one-way ANOVA tests for continuous outcomes and either chi-square or Fisher's exact tests for binary outcomes. Finally, we used multivariable logistic regression to assess the independent effects of sexual orientation or gender identity on unmet need and untreated depression and to understand how potential confounding barriers influence the relationship between our main exposure variables and the outcomes. We analyzed four separate models for each outcome. For unmet need, our first model included only the sexual orientation and gender identity variable; our second model added sociodemographic variables (age, lower SES, and racialized identity); our third model added mental health need and use variables (12-month depression or anhedonia, and 12-month treatment for depression); and our fourth model added social context variables (discriminatory events, microaggression, events and social support). Systemic exclusion was not included for the unmet need analysis because this information was only available for those who had endorsed an unmet need. The analysis of untreated depression was limited to those participants who had endorsed an unmet need for mental healthcare. For this analysis, we ran the first two models as described above, a third model added the social context variables described above plus a systemic exclusion variable. For all logistic regression models, we calculated odds ratios, p-values, standard errors, and 95% confidence intervals (CIs). All analyses were performed using SPSS 21.

We included all available data regardless of whether participants completed the survey or not. Because this could have led to bias if completers were markedly different than noncompleters, we compared these two groups on age, sexual/gender identity, racialization, and SES and found no significant differences.

Results

In total, 3963 unique IP addresses visited the website and 1071 unique visitors consented to participate; of these, 62 individuals did not provide sufficient information to determine eligibility and 305 individuals were excluded because they were younger than 18 years (n = 16), not living in Ontario (n = 69), self-identified only as men (n = 40), or belonged to an overrepresented risk group (n = 180). This left 704 eligible participants whose data were included in the analysis. Six hundred twenty-three participants completed the entire survey for a completeness rate of 88.5%. Given these numbers, the minimal response rate was 19% (704/3963–304), assuming that each IP address represents an eligible individual. However, if we apply the eligibility rate of those who gave consent (66%) to those with unknown eligibility (AAPOR Response Rate 4),⁴⁷ the response rate increases to 30% (704/[3963 × 0.66]−304). We included all available data in the analysis whether or not participants completed the entire questionnaire.

Baseline characteristics

Baseline characteristics by gender and sexual orientation group are shown in Table 2. There were 192 TRANS individuals, 114 BI/PAN women, 153 LGQ-identified women, and 245 CIS-HET women in the sample. The TRANS participants were most likely to have a high school education or less, have less than full-time employment, be in the lowest income category, be assigned low SES status, report greater than three lifetime discriminatory or microaggression events, report poor physical and emotional health, and report a 12-month history of depression or anhedonia. They also reported the lowest level of perceived social support. In contrast, women in the CIS-HET group reported the highest level of perceived social support, and the lowest rates of lifetime discriminatory events. Overall, 68.3% of the total sample reported an unmet need for mental healthcare in the past 12 months, and 33.1% of the sample reported a 12-month history of anhedonia or depression without any treatment.

Table 2.

Baseline Characteristics of Sample (N = 704) by Gender and Sexual Identity Group

Variable	TRANS (n = 192)	BI/PAN (n = 114)	LGQ (n = 153)	CIS-HET (n-245)	F, χ² statistics and p-values	Total mean (SD) or N (%)
Age ( n = 697) (Range:19–77)	36.1 (11.3)	33.8 (9.8)	36.9 (11.4)	41.8 (11.7)	F = 16.8, df 3, <0.001	37.9 (11.6)
Sociodemographic factors
Geographic region (n = 676), n (%)
Eastern Ontario	15 (8.1)	11 (9.9)	19 (13.2)	22 (9.3)	Nonsignificant (ns)	67 (9.9)
Central Ontario	27 (14.6)	21 (18.9)	26 (18.1)	60 (25.4)	^χ2 = 8.18, df 3, p = 0.042	134 (19.8)
Toronto	110 (59.5)	60 (54.1)	84 (58.3)	88 (37.3)	^χ2 = 26.5, df 3, <0.001	342 (50.6)
Southwestern Ontario	27 (14.6)	11 (9.9)	13 (9.0%)	39 (16.5)	ns	90 (13.3)
Northern Ontario	^a	8 (7.2)	^b	27 (11.4)	^χ2 = 19.4, df 3, <0.001	43 (6.4)
Education completed (n = 693)^χ n (%)
High school or less	76 (39.8)	26 (23.0)	45 (30.4)	62 (25.7)	² = 13.4, df 3, 0.004	209 (29.7)
College/bachelor's degree	84 (44.0)	55 (48.7)	59 (39.9)	137 (56.8)	^χ2 = 12.7, df 3, 0.005	335 (47.6)
Graduate or professional	31 (16.2)	32 (28.3)	44 (29.7)	42 (17.4)	^χ2 = 14.6, df 3, 0.002	149 (21.2)
Employment (n = 701) n (%)
Full-time paid/self-employed	60 (31.4)	53 (46.5)	63 (41.4)	117 (48.0)	^χ2 = 13.3, df 3, 0.004	293 (41.8)
Part-time paid	56 (29.3)	23 (20.2)	37 (24.3)	50 (20.5)	Ns	166 (23.7)
No paid employment	44 (23.0)	25 (21.9)	37 (24.3)	42 (17.2)	Ns	148 (21.1)
Disability	29 (15.2)	13 (11.4)	14 (9.2)	30 (12.3)	Ns	86 (12.3)
Retired	^b	0 (0.0)	^b	5 (2.0)	Ns	8 (2.7)
Combined household income (CAD) (n = 692) n (%)
<20,000	84 (44.4)	42 (37.2)	61 (40.4)	70 (29.3)	^χ2 = 11.3, df 3, 0.01	257 (37.1)
20,000–59,000	69 (36.5)	47 (41.6)	54 (35.8)	97 (40.6)	Ns	267 (3.6)
>59,000	36 (19.0)	24 (21.2)	36 (23.8)	72 (30.1)	^χ2 = 7.8, df 3, 0.049	168 (24.3)
Total perceived social support (n = 518)	72.6 (12.7)	74.1 (12.5)	75.2 (11.9)	76.5 (11.8)	F = 3.39, df 3, <0.018	74.9 (12.2)
% Racialized (n = 704)	75 (39.1)	35 (30.7)	72 (47.1)	99 (40.4)	Ns	281 (39.9)
Aboriginal/First Nations	25 (13.0)	11 (9.6)	11 (7.2)	31 (12.7)	Ns	78 (11.1)
Arab	5 (2.6)	^b	^b	^b	Ns	9 (1.3)
Black African	5 (2.6)	6 (5.3)	9 (5.9)	11 (4.5)	Ns	31 (4.4)
Black Caribbean	8 (4.2)	8 (7.0)	13 (8.5)	26 (10.6)	Ns	55 (7.8)
Chinese	6 (3.1)	^b	7 (4.6)	^a	Ns	21 (3.0)
Filipino	^b	0 (0.0)	^b	^b	Ns	7 (1.0)
Japanese	0 (0.0)	0 (0.0)	0 (0.0)	^b	Ns	^b
Korean	^b	0 (0.0)	0 (0.0)	0 (0.0)	Ns	^b
Latin America	8 (4.2%)	^b	^b	6 (2.4)	^χ2 = 9.4, df 3, 0.024	30 (4.3)
Southeast Asian	^b	^b	^b	^b	Ns	9 (1.3)
South Asian	11 (5.7)	^b	16 (10.5)	^a	^χ2 = 12.1, df 3, 0.007	38 (5.4)
West Asian	^b	0 (0.0)	^b	^b	Ns	^b
Other	16 (8.3)	5 (4.4)	11 (7.2)	19 (7.8)	Ns	51 (7.2)
% Low SES (n = 704)	107 (55.7)	42 (36.8)	67 (43.8)	113 (46.1)	^χ2 = 11.3, df 3, 0.010	329 (46.7)
Lifetime discrimination n (%)
% 3 or more discriminatory events (n = 704)	112 (58.3)	55 (48.2)	70 (45.8)	86 (35.1)	^χ2 = 23.7, df 3, <0.001	323 (45.9)
% 3 or more microaggressions (n = 704)	36 (18.8)	11 (9.6)	19 (12.4)	24 (9.8)	^χ2 = 9.1, df 3, 0.028	90 (12.8)
Health morbidity (need factors) n (%)
Self-reported general health poor or fair (n = 683)	60 (32.4)	25 (22.3)	25 (17.0)	61 (25.5)	^χ2 = 10.9, df 3, 0.012	171 (25.0)
Self-reported mental health poor or fair (n = 681)	102 (55.1)	57 (50.9)	57 (38.8)	96 (40.5)	^χ2 = 13.3, df 3, 0.004	312 (45.8)
12-month depression or anhedonia (n = 704)	162 (84.4)	87 (76.3)	106 (69.3)	178 (72.7)	^χ2 = 12.5, df 3, 0.006	533 (75.7)
PHQ-9 positive (n = 668)	49 (27.4)	24 (22.2)	30 (20.8)	55 (23.2%)	Ns	158 (23.7)
Unmet need (n = 652)	136 (78.2)	76 (72.4)	96 (67.1)	137 (59.6)	^χ2 = 16.8df 3, 0.001	445 (68.3)
Untreated depression (n = 668)	73 (38.0)	39 (34.2)	53 (34.6)	68 (27.8)	ns	233 (33.1)

Bold text indicates statistically significant differences between gender and sexual identity groups.

Small cells (n < 10) were suppressed to protect participant confidentiality.

Small cells (n < 5) were suppressed to protect participant confidentiality.

PHQ, patient health questionnaire; BI/PAN, bisexual or pansexual; CIS-HET, cisgendered heterosexual; LGQ, lesbian, gay, or queer; TRANS, trans identified.

The CIS-HET women in our study differed from a population-based sample of Canadian heterosexual women taken from the CCHS Cycle 2.1. Our participants were younger (41.8 years vs. 45.8 years), more likely to have less than a high school education (25.7% vs. 18.9%), and less likely to have paid employment (48.0% vs. 55.0%).²⁰

Unmet need for mental healthcare

Results of our logistic regressions for unmet need are shown in Table 3. In the unadjusted model, TRANS participants were 2.4 times as likely to report an unmet need for mental healthcare than CIS-HET women (p < 0.01). BI/PAN women were 1.8 times as likely to report an unmet need for mental health relative to CIS-HET women (p = 0.02), and there was no significant difference in rates of unmet need between LGQ and CIS-HET women. The associations between TRANS and BI/PAN identity and unmet need were mitigated after adjusting for socioeconomic variables and age (Model 2), and in Model 3, psychiatric morbidity and treatment variables, but still remained significant for TRANS participants after these adjustments. Self-reported treatment for depression was not significantly associated with self-reported unmet need for mental healthcare. In Model 4, we see that the experience of a high number of lifetime discriminatory events and lower levels of perceived social support is independently associated with the report of unmet need for mental healthcare. With these additions to the model, the association between TRANS identity and self-reported unmet need was no longer statistically significant.

Table 3.

Logistic Regression Models Predicting Past 12-Month Unmet Need for Mental Healthcare

	Model 1 (n = 652) sexual/gender identity		Model 2 (n = 648) sociodemographic characteristics		Model 3 (n = 634) need and use variables		Model 4 (n = 494) social context
Variables	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p
Trans	2.4 (1.6–3.8)	<0.01	2.1 (1.3–3.3)	0.02	1.9 (1.2–3.1)	0.01	1.3 (0.7–2.3)	0.37
Bi/Pan	1.8 (1.1–2.9)	0.02	1.5 (0.9–2.6)	0.12	1.5 (0.9–2.7)	0.15	1.2 (0.6–2.4)	0.59
LGQ	1.4 (0.9–2.1)	0.14	1.2 (0.8–1.9)	0.39	1.3 (0.8–2.0)	0.32	1.0 (0.6–1.8)	0.94
Base: Cis-Het
Age			0.97 (0.96–0.99)	<0.001	0.97 (0.96–0.99)	<0.001	0.97 (0.95–0.99)	<0.001
Lower SES			1.5 (1.1–2.1)	0.02	1.3 (0.9–1.9)	0.18	1.0 (0.6–1.5)	0.94
Racialized			1.1 (0.8–1.6)	0.61	1.2 (0.8–1.7)	0.43	0.8 (0.5–1.2)	0.30
Depression or anhedonia					3.0 (1.9–4.6)	<0.01	2.6 (1.5–4.3)	<0.01
Depression treatment					1.1 (0.8–1.6)	0.60	1.0 (0.6–1.6)	0.94
>3 Discriminatory events							2.2 (1.4–3.4)	<0.01
>3 Microaggression events							0.7 (0.3–1.4)	0.28
Social support							0.95 (0.93–0.97)	<0.001

Bold text indicates statistical significance p = 0.05.

CI, confidence intervals; OR, odds ratio.

Untreated depression

Results of our logistic regressions for untreated depression are shown in Table 4. In the unadjusted model, TRANS participants were 1.6 times more likely to report an untreated depression than CIS-HET women (p = 0.04), but there were no other significant differences between groups. The association for TRANS participants was no longer significant after adjusting for socioeconomic variables and age. In Model 3 we see that age, SES, and barriers to mental healthcare related to systemic exclusion were independently associated with the report of untreated depression.

Table 4.

Logistic Regression Models Predicting Untreated Depression Among Participants with Unmet Need

	Model 1 (n = 445)		Model 2 (n = 443)		Model 3 (n = 349)
	Sexual/gender identity		Sociodemographic characteristics		Social context
Variables	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p
Trans	1.6 (1.0–2.7)	0.04	1.6 (1.0–2.7)	0.07	1.7 (0.9–3.1)	0.12
Bi/Pan	1.5 (0.8–2.7)	0.17	1.2 (0.6–2.1)	0.65	1.3 (0.6–2.7)	0.47
LGQ	1.4 (0.8–2.5)	0.20	1.2 (0.7–2.2)	0.46	1.0 (0.5–1.9)	0.97
Base: Cis-Het	1.0
Age			0.97 (0.95–0.99)	<0.05	0.97 (0.95–0.99)	<0.05
Lower SES			0.5 (0.3–0.8)	<0.05	0.5 (0.3–0.8)	<0.05
Racialized			1.1 (0.8–1.7)	0.52	1.3 (0.8–2.0)	0.31
>3 Discriminatory events					1.0 (0.98–1.0)	0.88
>3 Microaggression events					1.5 (0.78–3.0)	1.00
Social support					1.0 (0.98–1.0)	0.80
Systemic exclusion					1.8 (1.0–3.0)	0.04

Bold text indicates statistical significance p = 0.05.

Our large sample of transgender participants is an improvement over most other research in this area. However, it is important to acknowledge that a diverse group of individuals were categorized as “trans” in this study. There may be important differences between trans women, trans men, and those who identify outside of binary genders (e.g., genderqueer) that are masked by the broad categorization used in this study. To address this concern, we examined differences between trans women (n = 86), trans men (n = 60), and other trans people (n = 46) in our study. We found people who identified outside the binary had higher levels of both unmet need and untreated depression although these differences did not meet statistical significance (UNMET NEED: trans women: 75.9%, trans men: 76.8%, other trans people: 84.6%; chi-square 1.24, p = 0.54. UNTREATED DEPRESSION: trans women: 30.2%, trans men: 43.3%, other trans people: 45.7%; chi-square 4.1, p = 0.13).

Discussion

In this study of a large sample of women and/or trans people with various sexual identities, we found high rates of self-perceived unmet need for mental healthcare and untreated depression among all participants; however, there was a consistent gradient in which trans participants reported the highest rates of unmet need and untreated depression, followed by bisexual/pansexual women; LGQ respondents reported rates of need and untreated depression that were similar to those of the cisgender heterosexual participants. Psychiatric morbidity and social factors, including experiences of discrimination and low levels of social support, were independently associated with unmet need. We also found that a measure of systemic exclusion from healthcare was significantly associated with untreated depression among participants with unmet need. We conclude that there are higher rates of unmet need and untreated depression in trans and bisexual participants that are partly explained by differences in social factors, including experiences of discrimination, lower levels of social support, and systemic exclusion from healthcare.

A primary strength of the study is the large sample size, which allowed for comparisons between groups within the LGBT acronym. In most other research on LGBT mental health, these groups are examined separately from one another, and our findings indicate that this approach may mask important differences between groups. In particular, our study indicates the importance of including TRANS participants in research of this nature, a group that is often excluded from analyses due to small sample sizes, and yet may experience the poorest access to mental health services. Our study is also strengthened by deliberate inclusion of participants with a broad spectrum of racialized identities, as well as a substantial proportion of participants of low socioeconomic status, thus extending our findings beyond the limited scope of most studies on LGBT mental health. Finally, our study is strengthened by use of standardized items to measure many of our outcome and predictor variables, allowing for comparison with other research in this field.

The primary limitation of this study is the convenience sampling design, although it is important to note that there are no population-based Canadian data sources that would enable us to address our research questions (particularly since there are no population-based surveys in Canada currently that assess gender identity). The response rate for our survey limits the generalizability of our findings, although across disciplines, response rates for Internet-based surveys have been found to be lower than those for mail or other surveys.⁴⁸ For example, a meta-analysis of response rates in a total of 68 Internet surveys yielded a mean response rate of 39.6%.⁴⁹ In this particular study, we would anticipate that some of those individuals who logged on to the study home page did so without an intention to participate; for example, representatives of organizations determining whether or not they would promote the survey to their clients. It is also likely that individuals viewed the survey before determining that they were not eligible to participate. These types of “views” would artificially lower the minimal response rate. Using the IP address count as the denominator for calculating the response rate is imprecise; individual IP addresses could represent more than one individual (e.g., when a firewall is in place), or a single individual could be assigned more than one IP address (e.g., logging in from two different settings, or having a dynamic IP assigned by a service provider). These influences could erroneously raise or lower the estimated response rate.

While our sample size was relatively large in comparison to existing studies, we may still have been limited in identifying some important differences between the groups. Odds ratios of less than 1.5 tended to be nonsignificant in our models. So, important but smaller differences between groups could have been missed.

Some of our measures did not utilize standardized scales or indices. For example, our measure of systemic exclusion was a composite measure of standard items developed in conjunction with the participation of our community advisory committee and made sense for our participants in our setting. There is a global literature on the measurement of social exclusion that could inform researchers in other settings.⁵⁰ Furthermore, we did not assess gender assigned at birth, and consequently, it is possible that we misclassified some individuals who had changed gender, but did not endorse any of the trans identities. This would bias our results toward the null hypothesis.

Our study adds to existing but limited evidence that factors related to social exclusion comprise barriers to mental health services and effective depression treatment, thus, contributing to mental health disparities associated with minority sexual orientations and gender identities.^9

–13 In our study, individuals with low income were more likely to report an unmet need for healthcare, but less likely to report a 12-month history of untreated depression. In Canada, which has universal healthcare coverage, physician-provided mental healthcare is provided without user fees or limits—this leads to higher rates of treatment in lower income groups who have higher rates of mental health morbidity.⁵¹ Other Canadian studies have identified that ongoing unmet need in the context of treated depression is often related to a perceived need for counseling.⁵² In Canada, counseling provided by non-physician providers is not covered by our universal health insurance. Since sexual and gender minorities are overrepresented among low SES groups,⁵³ economic insecurity may compound the difficulties that sexual and gender minority people face to access receptive, diversity-competent (racially competent, LGBT competent) mental health counseling services.

Social exclusion includes elements such as experiences of discrimination and exclusionary healthcare practices that are highly relevant to LGBT communities.⁵⁴ In this regard, our study findings on rates of perceived unmet need for mental healthcare and untreated depression are aligned with existing research that associates social exclusion with lower levels of mental health service access.^55,56 While the importance of the social determinants of health has been widely acknowledged in the realms of population and public health, the understanding of health as contextual has not been translated into the mental health system, which continues to rely on a predominantly biomedical approach to treatment.⁵⁷ If social exclusion is central to the experience of mental health services for LGBT people (and indeed, for other communities that experience marginalization), the predominance of the biomedical model may, in fact, be inequitable in that it is less likely to meet the needs of some groups than others.

These findings suggest a need to explore models of mental healthcare that can more adequately address the experience of social exclusion that may be fundamental to meeting the mental health service needs of LGBT people. Additional research is needed to confirm these findings using a population-based sample; as noted above, this requires improvements to many population-based data sources in most countries or states to enable identification of sexual, and in particular, gender minority people. Research is needed to further explore the mental health needs and experiences of the highest risk groups, in particular, that is, trans and bisexual identified individuals, as well as LGBT individuals with intersecting marginalized identities, such as those related to class and racialization. Finally, research is needed to determine which models of healthcare, as well as which specific mental health interventions, are effective for LGBT people specifically. The field of LGBT mental health research has to date largely focused on description, and intervention research for this population is sorely needed.

In sum, our finding of high rates of unmet need for mental healthcare and untreated depression, with elevated rates among transgender people and bisexual women in particular, suggests that the mental health system is not currently meeting the needs of many sexual and gender minority people. While this was also true for many of the cisgender heterosexual women who participated in this study, the striking mental health disparities associated with minority sexual and gender identities suggest a particularly urgent need to address access for these populations.

Footnotes

Acknowledgments

This project was funded by the Canadian Institutes for Health Research in partnership with Echo: Improving Women's Health in Ontario, grant number MPO-105685. The authors thank the following for their essential contributions: Punam Khosla, Loralee Gillis, members of the Pathways Project Advisory Committee, the study participants, organizations who facilitated recruitment, all the project staff and students, and other members of the Re:searching for LGBTQ Health research team.

Author Disclosure Statement

No competing financial interests exist.

References

Cochran

, Mays

. Physical health complaints among lesbians, gay men, and bisexual and homosexually experienced heterosexual individuals: Results from the California Quality of Life Survey. Am J Public Health, 2007; 97:2048–2055.

Diamant

, Wold

. Sexual orientation and variation in physical and mental health status among women. J Women's Health, 2003; 12:41–49.

Sandfort

, Bakker

, Schellevis

, et al. Sexual orientation and mental and physical health status: Findings from a Dutch population survey. Am J Public Health, 2006; 96:1119–1125.

Rotondi

, Bauer

, Scanlon

, et al. Prevalence of and risk and protective factors for depression in female-to-male transgender ontarians: Trans PULSE Project. Can J Commun Ment Health, 2011; 30:135–155.

Rotondi

, Bauer

, Travers

, el al. Depression in male-to-female transgender ontarians: Results from the Trans PULSE Project. Can J Commun Ment Health, 2011; 30:113–133.

Lawrence

. Transgender health concerns. In: Meyer

, Northridge

, eds. The health of sexual minorities: Public health perspectives on lesbian, gay, bisexual and transgender populations. New York: Springer, 2007:473–505.

Clements-Nolle

, Marx

, Guzman

, et al. HIV prevalence, risk behaviors, health care use, and mental health status of transgender persons: Implications for public health intervention. Am J Public Health, 2001; 91:915–921.

Bauer

, Pyne

, Francino

, et al. Suicidality among trans people in Ontario: Implications for social work and social justice/La suicidabilité parmi les personnes trans en Ontario: Implications en travail social et en justice sociale. Serv Soc, 2013; 59:35–62.

Grella

, Greenwell

, Mays

, et al. Influence of gender, sexual orientation, and need on treatment utilization for substance use and mental disorders: Findings from the California Quality of Life Survey. BMC Psychiatry, 2009; 9:52.

10.

Meyer

, Dietrich

, Schwartz

. Lifetime prevalence of mental disorders and suicide attempts in diverse lesbian, gay, and bisexual populations. Am J Public Health, 2008; 98:1004–1006.

11.

Warner

, McKeown

, Griffin

, et al. Rates and predictors of mental illness in gay men, lesbians and bisexual men and women: Results from a survey based in England and Wales. Br J Psychiatry, 2004; 185:479–485.

12.

Mays

, Cochran

. Mental health correlates of perceived discrimination among lesbian, gay, and bisexual adults in the United States. Am J Public Health, 2001; 91:1869–1876.

13.

Gevonden

, Selten

, Myin-Germeys

, et al. Sexual minority status and psychotic symptoms: Findings from the Netherlands Mental Health Survey and Incidence Studies (NEMESIS). Psychol Med, 2014; 44:421–433.

14.

Kessler

. Epidemiology of women and depression. J Affect Disord, 2003; 74:5–13.

15.

Romans

, Ross

. Gender and depression. In: Cairney

, Streiner

, Eds. Mental Disorder in Canada: An Epidemiological Perspective. Toronto: University of Toronto Press, 2010. 259–285.

16.

Kessler

, Brown

, Broman

. Sex differences in psychiatric help-seeking: Evidence from four large-scale surveys. J Health Soc Behav, 1981; 22:49–64.

17.

Rhodes

, Goering

, To

, Williams

. Gender and outpatient mental health service use. Soc Sci Med, 2002; 54:1–10.

18.

Tjepkema

. Health care use among gay, lesbian and bisexual Canadians. Health Rep, 2008; 19:53–64.

19.

Sandfort

, de Graaf

, Bijl

, et al. Same-sex sexual behavior and psychiatric disorders: Findings from the Netherlands Mental Health Survey and Incidence Study (NEMESIS). Arch Gen Psychiatry, 2001; 58:85–91.

20.

Steele

, Ross

, Dobinson

, Veldhuizen

, Tinmouth

. Women's sexual orientation and health: Results from a Canadian population-based survey. Women Health, 2009; 49:353–367.

21.

Conron

, Mimiaga

, Landers

. A population-based study of sexual orientation identity and gender differences in adult health. Am J Public Health, 2010; 100:1953–1960.

22.

Ross

, Siegel

, Dobinson

, et al. I don't want to turn totally invisible: Mental health, stressors and supports among bisexual women during the perinatal period. J GLBT Fam Stud, 2012; 8:137–154.

23.

Koh

, Ross

. Mental health issues: A comparison of lesbian, bisexual and heterosexual women. J Homosex, 2006; 51:33–57.

24.

Balsam

, Beauchaine

, Mickey

, et al. Mental health of lesbian, gay, bisexual, and heterosexual siblings: Effects of gender, sexual orientation, and family. J Abnorm Psychol, 2005; 114:471–476.

25.

Welch

, Collings

, Howden-Chapman

. Lesbians in New Zealand: Their mental health and satisfaction with mental health services. Aust N Z J Psychiatry, 2000; 34:256–263.

26.

McCann

, Sharek

. Survey of lesbian, gay, bisexual, and transgender people's experiences of mental health services in Ireland. Int J Ment Health Nurs, 2014; 23:118–127.

27.

Avery

, Hellman

, Sudderth

. Satisfaction with mental health services among sexual minorities with major mental illness. Am J Public Health, 2001; 91:990–991.

28.

Ross

, Dobinson

, Eady

. Perceived determinants of mental health for bisexual people: A qualitative examination. Am J Public Health, 2010; 100:496–502.

29.

Eady

, Dobinson

, Ross

. Bisexual people's experiences with mental health services: A qualitative investigation. Commun Ment Health J, 2011; 47:378–389.

30.

Namaste

. Invisible lives: The erasure of transsexual and transgendered people. Chicago: University of Chicago Press, 2000.

31.

Lombardi

. Substance use treatment experiences of transgender/transsexual men and women. J LGBT Health Res, 2007; 3:37–47.

32.

Sperber

, Landers

, Lawrence

. Access to health care for transgendered persons: Results of a needs assessment in Boston. Int J Transgenderism, 2005; 8:74–91.

33.

Bauer

, Hammond

, Travers

, et al. “I don't think this is theoretical; this is our lives”: How erasure impacts health care for transgender people. J Assoc Nurses AIDS Care, 2009; 20:348–361.

34.

Statistics Canada. Canadian Community Health Survey 1.2: Mental Health and Well-Being, 2003. Available at: www.statcan.gc.ca/imdb-bmdi/instrument/5015_Q1_V1-eng.pdf Accessed August 17, 2015 .

35.

Statistics Canada. Canadian Community Health Survey, 2008 Questionnaire. Available at: www.statcan.gc.ca/imdb-bmdi/instrument/3226_Q1_V5-eng.pdf Accessed August 17, 2015 .

36.

Steele

, Ross

, Epstein

, et al. Correlates of mental health service use among lesbian, gay, and bisexual mothers and prospective mothers. Women Health, 2008; 47:95–112.

37.

Simeonov

, Steele

, Anderson

, et al. Perceived satisfaction with mental health services in the lesbian, gay, bisexual, transgender and transsexual communities in Ontario, Canada: An internet-based survey. Can J Commun Ment Health, 2015; 34:31–44.

38.

Statistics Canada. 2006. Census Dictionary. Statistics Canada Catalogue no. 92-566-XWE. Ottawa, Ontario. Available at: www12.statcan.ca/english/census06/reference/dictionary/index.cfm Accessed August 17, 2015 .

39.

Kroenke

, Spitzer

, Williams

. The PHQ-9: Validity of a brief depression severity measure. J Gen Intern Med, 2001; 16:606–613.

40.

Idler

, Benyami

. Self-rated health and mortality: A review of twenty-seven community studies. J Health Soc Behav, 1997; 38:21–37.

41.

Cutrona

, Russell

. The provisions of social relationships and adaptation to stress. In: Jones

, Perlman

, Eds. Advances in Personal Relationships. Greenwich, Connecticut: JAI Press, 1987. 37–67.

42.

Russell

, Cutrona

, Rose

, et al. Social and emotional loneliness: An examination of Weiss's typology of loneliness. J Pers Social Psychol, 1984; 46:1313–1321.

43.

Forman

, Williams

, Jackson

. Race, place, and discrimination. In: Forman

, David

, James

, Gardner

, eds. Perspectives on Social Problems. Greenwich, Connecticut: JAI Press, 1997.

44.

Brascoupe

, Waters

. Cultural safety: Exploring the applicability of the concept of cultural safety to aboriginal health and community wellness. J Aborig Health, 2009; 5:6–41.

45.

Meyer

, Wilson

. Sampling lesbian, gay and bisexual populations. J Couns Psychol, 2009; 56:23–31.

46.

Demidenko

. Sample size determination for logistic regression revisited. Stat Med, 2007; 26:3385–3397.

47.

The American Association for Public Opinion Research. Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys. 7th edition. AAPOR, 2011. Available at: https://www.esomar.org/uploads/public/knowledge-and-standards/codes-and-guidelines/ESOMAR_Standard-Definitions-Final-Dispositions-of-Case-Codes-and-Outcome-Rates-for-Surveys.pdf. Accessed November 19, 2016 .

48.

Lozar Manfreda

, Bosnjak

, Berzelak

, et al. Web surveys versus other survey modes: A meta-analysis comparing response rates. Int J Market Research, 2008; 50:79–104.

49.

Cook

, Heath

, Thompson

. A meta-analysis of response rates in web- or internet-based surveys. Educ Psychol Meas, 2000; 60:821–836.

50.

Popay

, Escorel

, Hernández

, et al. Social Exclusion Knowledge Network. Understanding and Tackling Social Exclusion Final Report to the WHO Commission on Social Determinants of Health, 2008. Available at: www.who.int/social_determinants/knowledge_networks/final_reports/sekn_final%20report_042008.pdf?ua=1 Accessed on November 19, 2016 .

51.

Vasiliadis

, Tempier

, Lesage

, et al. General practice and mental health care: Determinants of outpatient service use. Can J Psychiatry, 2009; 54:468–476.

52.

Sunderland

, Findlay

. Perceived need for mental health care in Canada: Results from the 2012 Canadian Community Health Survey–Mental Health. Health Rep, 2013; 24:3–9.

53.

Lee Badgett

, Durso

, Schneebaum

. New Patterns of Poverty in the Lesbian, Gay, and Bisexual Community. The Williams Institute, UCLA School of Law. June 2013. Available at: http://williamsinstitute.law.ucla.edu/wp-content/uploads/LGB-Poverty-Update-Jun-2013.pdf Accessed August 17, 2015 .

54.

MacKinnon

. Poverty and Social Exclusion: Solving Complex Issues through Comprehensive Approaches. CCPA Review: Economic and Social Trends. Canadian Centre for Policy Alternatives, 2008. Available at: www.policyalternatives.ca/sites/default/files/uploads/publications/Manitoba_Pubs/2008/Poverty_and_Social_Exclusion.pdf Accessed August 17, 2015 .

55.

Ingram

, Wasik

, Cormier

, et al. Social inequities and mental health: A scoping review. Vancouver: Centre for the Study of Gender, Social Inequities and Mental Health, 2013.

56.

Rossiter

, Morrow

. Intersectional frameworks in mental health: Moving from theory to practice. In: Hankivksy

, ed. Health and intersectionality inquiry in Canada. Vancouver: UBC Press, 2011:312–330.

57.

Deacon

. The biomedical model of mental disorder: A critical analysis of its validity, utility, and effects on psychotherapy research. Clin Psychol Rev, 2013; 33:846–861.