Characteristics of Transgender Individuals With Emergency Department Visits and Hospitalizations for Mental Health
Abstract
Objective:
Transgender individuals experience significant oppression resulting in mental health disparities. Factors associated with their need for acute mental health care are unknown. This study compared characteristics of transgender individuals who presented for acute mental health care with population-based comparison samples.
Methods:
This cross-sectional study examined transgender individuals who had a mental health–related emergency department (ED) visit (N=728) or hospitalization (N=454). Transgender individuals were identified, and their data were linked with health administrative data. The transgender ED and hospitalization samples were each compared with two samples: all individuals in Ontario who had an ED visit or hospitalization (unmatched) and individuals matched on age, region of residence, and mental health care utilization history. Individuals’ sociodemographic and clinical factors were compared.
Results:
After matching, transgender individuals in the ED sample were more likely than those in the comparison group to be in the lowest neighborhood income quintile (37% versus 27%) and the highest residential instability quintile (47% versus 38%) and to be diagnosed as having a mood (26% versus 19%) or personality disorder (4% versus 1%). Transgender individuals in the hospitalization sample were more likely to be in the lowest neighborhood income quintile (36% versus 27%) and the highest residential instability quintile (45% versus 35%) and to be diagnosed as having a mood (40% versus 35%) or personality disorder (5% versus 2%).
Conclusions:
Transgender individuals who accessed acute mental health care had unique sociodemographic and clinical factors associated with their presentation that persisted after matching. More research into the factors associated with their acute care presentation is warranted, including how experiences of marginalization play a role.
HIGHLIGHTS
Indicators of socioeconomic marginalization (living in areas of low income and high residential instability and material deprivation) were associated with samples of transgender individuals who had a mental health–related emergency department (ED) visit or hospitalization.
Transgender individuals who presented for a psychiatric ED visit were more likely to be diagnosed as having a mood or personality disorder and less likely to have a substance-related disorder than were individuals in population-based comparison groups.
Transgender individuals who had a psychiatric hospitalization were more likely to be admitted for a mood or personality disorder and less likely to be admitted for a psychotic or substance-related disorder than were individuals in population-based comparison groups.
Transgender people are individuals whose gender identity differs from their sex assigned at birth (1). They are estimated to represent at least 0.5% of the population globally (2, 3). International studies have found a two- to fivefold increase in diagnoses of depression and anxiety among transgender people compared with cisgender individuals (3–5), with 10-fold increases in the rate of suicide attempts (3, 6). High rates of mental illness, substance use, and suicidality are related to experiences of marginalization and oppression, including experiences of transphobia, violence, lack of social support, barriers to education, homelessness, and unemployment (3, 7–9). The minority stress model posits that those with marginalized identities, including transgender people, face chronically high levels of stress due to discrimination, which leads to increased rates and severity of general medical and mental illness (7, 8, 10, 11).
Higher rates of mental illness among transgender individuals likely produce greater need for acute mental health care, including emergency department (ED) visits and hospitalizations. Health administrative data can be useful to examine patterns of mental health care access for populations. Recent studies have explored transgender patients’ use of mental health care by using health administrative data in the United States and Canada (12–16). Studies have found that transgender individuals were more likely to have had psychiatric outpatient visits, hospitalizations, and ED visits (12–16) than were cisgender people. However, these studies have had several important limitations. One limitation was the use of ICD-9 diagnoses (gender identity disorders) and involvement with medical transition to identify transgender individuals in several U.S. studies (12, 13). These methods are not representative of many transgender people without these diagnoses and many who do not seek medical transition (i.e., gender-affirming hormone therapy or surgeries). One U.S. study used a convenience sample from one site, which often confers selection bias (14).
None of the prior studies clarified the factors and individual characteristics associated with the increased need for acute mental health care. Examining factors associated with this need can highlight contributors to mental illness that are amenable to intervention to improve mental health care for transgender people. Using linked health administrative data, this study aimed to compare characteristics of transgender individuals who had had a psychiatric hospitalization or ED visit in Ontario, Canada, with characteristics of population-based comparison samples.
Methods
Design and Setting
This cross-sectional study used linked health administrative data to compare characteristics of transgender individuals who had a psychiatric ED visit or a psychiatric hospitalization in Ontario with characteristics of individuals in the general population of acute psychiatric care users. This project was conducted by using linked health administrative data held at ICES. ICES is an independent, nonprofit research institute. Its legal status under Ontario’s health information privacy law allows it to collect and analyze health care and demographic data, without consent, for the purposes of health system evaluation, planning, and monitoring. The use of data in this project was authorized under section 45 of Ontario’s Personal Health Information Protection Act, which does not require review by a research ethics board. The linkage of external data to ICES for this project was approved by the research ethics board at the University of Toronto (Research Information System protocol 38210).
Data Sources
The ICES data repository includes individual-level longitudinal data on most publicly funded health care services for individuals covered by Ontario health insurance. These data sets were linked by using unique encoded identifiers and were analyzed at ICES. Data resources used in this study included the Registered Persons Database, which includes demographic information for all individuals in Ontario with a health card number. The Ontario Health Insurance Plan database includes physician billing information. The National Ambulatory Care Reporting System contains information about ED visits. The Ontario Mental Health Reporting System database includes information from psychiatric hospitalizations in hospital beds designated for mental health care. The Hospital Discharge Abstract Database captures information from medical hospitalizations in nonmental health hospital beds. The Ontario Marginalization Index (ON-Marg) is a provincial adaptation of the Canadian Marginalization Index and uses the smallest census area–level data to create validated variables measuring multidimensional aspects of marginalization through four dimensions: dependency, residential instability, material deprivation, and ethnic concentration (17).
Study Population
All individuals age ≥16 who were discharged from a psychiatric hospitalization (hospitalized sample) or mental health–related ED visit (ED sample) in Ontario between January 1, 2012, and December 31, 2018, were eligible to be in the study population. The index discharge date was the individual’s first discharge within the period. Patients were excluded if they were not Ontario residents, did not have a valid health card number for data linkage, or were missing valid data for discharge diagnosis, age, or their Ontario region of residence (i.e., Local Health Integration Network [LHIN]) during the study period.
Identifying Transgender Individuals
Transgender individuals were systematically identified through data obtained from electronic medical records (EMRs) of four outpatient health clinics in three cities across Ontario (Thunder Bay, Ottawa, Toronto). Each patient was asked to describe their gender identity by answering questions on the clinic’s intake form. The manager of health information at each clinic identified all transgender patients from 2012 to 2016 through their EMR. Individuals’ Ontario health insurance number and date of birth were used to link individuals to administrative databases at ICES. This process has been previously described (16). All transgender individuals were identified from the study population and included in the two transgender samples in this study.
Comparison Groups
Two comparison groups were created for each sample of transgender individuals to understand the ways in which the transgender population differs from the general Ontario population of acute care users (unmatched) and from a cisgender population from the same regions with similar demographic characteristics and history of mental health care (matched). In each matched comparison group, four individuals from the comparison group were found to match with each transgender individual on the basis of shared age, region of residence, and mental health care utilization history. Region of residence was a variable identifying in which of 14 LHINs the individual resided. LHINs are the health authorities responsible for regional administration of public health care services in Ontario. Individuals were placed into four categories of mental health care utilization history on the basis of their mental health service use in the 24 months prior: having any psychiatric hospitalization history, having no psychiatric hospitalization but any psychiatric ED visit history, having no history of psychiatric hospitalization or ED visit but any outpatient mental health visit history, or having none of the above. This categorization strategy uses patient’s use of acute psychiatric care to approximate the severity of prior mental illness (18).
Variables
The following sociodemographic variables were measured for each group: age, LHIN, rurality (derived from postal code), neighborhood-level income (measured in quintiles at the census tract level), and ON-Marg quintiles. Gender identity was not included because of the high proportion of records with unknown gender in the transgender sample (16). The clinical diagnosis for the ED visit or hospitalization was captured in one of seven psychiatric diagnostic categories used in previous studies (19). Prior health service use variables captured the number of prior mental health–related ED visits, hospitalizations, outpatient visits, and self-harm–related ED visits or hospitalizations in the 24 months before the index admission date.
Statistical Analysis
Descriptive and baseline characteristics (frequencies and means) were calculated across both samples. Differences between samples were explored by using chi-square tests and Student’s t tests, as appropriate. Statistical significance comparing samples was set to p<0.05 in two-tailed tests. All analyses were conducted with SAS, version 9.4.
Results
Transgender ED Sample
Comparing the transgender ED sample (N=728) with the unmatched general population of ED users (N=581,708), we found that the transgender ED sample was younger (mean±SD age=28.8±11.7 versus 38.3±18.1, t=21.90, df=731, p<0.001) and less rural (3% versus 14.4%, χ2=79.64, N=580,768, df=1, p<0.001). They were more likely to be in the lowest quintile for neighborhood income (37% versus 27.6%, χ2=30.91, N=582,436, df=5, p<0.001) and dependency (30% versus 23.2%, χ2=37.95, N=571,637, df=4, p<0.001) and in the highest quintile for residential instability (47% versus 29.2%, χ2=118.47, N=571,637, df=4, p<0.001), material deprivation (33% versus 27.6%, χ2=9.58, N=571,637, df=4, p=0.048), and ethnic concentration (29% versus 22.1%, χ2=78.78, N=571,637, df=4, p<0.001) (Table 1). The transgender ED sample was more likely than the unmatched general population of ED users to be diagnosed as having a mood disorder (26% versus 15.6%), schizophrenia or psychotic disorder (5% versus 3.4%), personality disorder (4% versus 0.7%), and deliberate self-harm (7% versus 5.2%) (χ2=208.83, N=582,436, df=6, p<0.001) (Table 2). They were less likely to be diagnosed as having an anxiety, trauma, or obsessive-compulsive–related disorder (38% versus 46.0%) or a substance-related disorder (14% versus 25.0%). The transgender ED sample had much greater prior psychiatric service use than did the sample of unmatched general population of ED users, including having more prior hospitalizations (21% versus 8.1%, χ2=172.73, N=582,436, df=1, p<0.001), ED visits (16% versus 9.9%, χ2=30.27, N=582,436, df=1, p<0.001), outpatient visits (83% versus 57.3%, χ2=198.21, N=582,436, df=1, p<0.001), and visits for deliberate self-harm (7% versus 2.3%, χ2=81.92, N=582,436, df=1, p<0.001) (Table 3).
Transgender ED sample (N=728) | Unmatched comparison sample (N=581,708) | Matched comparison sample (N=2,912) | ||||||
---|---|---|---|---|---|---|---|---|
Characteristic | N | % | N | % | pb | N | % | pb |
Age at admission (M±SD) | 28.8±11.7 | 38.3±18.1 | <.001 | 29.0±11.7 | .76 | |||
Rural residence | <.001 | <.001 | ||||||
Rural (population <10,000) | 20 | 3 | 83,604 | 14.4 | 247 | 9 | ||
Urban (population ≥10,000) | 704 | 97 | 496,440 | 85.3 | 2,653 | 91 | ||
Neighborhood income quintile | <.001 | <.001 | ||||||
Q1 (poorest) | 266 | 37 | 160,320 | 27.6 | 774 | 27 | ||
Q2 | 139 | 19 | 123,313 | 21.2 | 563 | 19 | ||
Q3 | 110 | 15 | 108,753 | 18.7 | 518 | 18 | ||
Q4 | 105 | 14 | 97,229 | 16.7 | 517 | 18 | ||
Q5 (wealthiest) | 103 | 14 | 88,884 | 15.3 | 518 | 18 | ||
ON-Marg dependency quintile | <.001 | .67 | ||||||
Q1 (least dependent) | 216 | 30 | 135,070 | 23.2 | 901 | 31 | ||
Q2 | 167 | 23 | 113,131 | 19.4 | 706 | 24 | ||
Q3 | 137 | 19 | 103,316 | 17.8 | 508 | 17 | ||
Q4 | 94 | 13 | 102,665 | 17.6 | 355 | 12 | ||
Q5 (most dependent) | 106 | 15 | 116,735 | 20.1 | 384 | 13 | ||
ON-Marg residential instability quintile | <.001 | <.001 | ||||||
Q1 (least unstable) | 70 | 10 | 86,013 | 14.8 | 405 | 14 | ||
Q2 | 61 | 8 | 91,172 | 15.7 | 385 | 13 | ||
Q3 | 110 | 15 | 101,314 | 17.4 | 414 | 14 | ||
Q4 | 138 | 19 | 122,734 | 21.1 | 556 | 19 | ||
Q5 (most unstable) | 341 | 47 | 169,684 | 29.2 | 1,094 | 38 | ||
ON-Marg material deprivation quintile | .048 | <.001 | ||||||
Q1 (least deprived) | 108 | 15 | 89,744 | 15.4 | 580 | 20 | ||
Q2 | 113 | 16 | 98,194 | 16.9 | 466 | 16 | ||
Q3 | 119 | 16 | 105,155 | 18.1 | 517 | 18 | ||
Q4 | 141 | 19 | 117,440 | 20.2 | 538 | 19 | ||
Q5 (most deprived) | 239 | 33 | 160,384 | 27.6 | 753 | 26 | ||
ON-Marg ethnic concentration quintile | <.001 | .78 | ||||||
Q1 (lowest ethnic concentration) | 77 | 11 | 112,499 | 19.3 | 336 | 12 | ||
Q2 | 92 | 13 | 109,298 | 18.8 | 395 | 14 | ||
Q3 | 145 | 20 | 106,611 | 18.3 | 567 | 20 | ||
Q4 | 196 | 27 | 114,077 | 19.6 | 775 | 27 | ||
Q5 (highest ethnic concentration) | 210 | 29 | 128,432 | 22.1 | 781 | 27 | ||
Transgender hospitalized sample (N=454) | Unmatched comparison sample (N=217,507) | Matched comparison sample (N=1,808) | ||||||
Age at admission (M±SD) | 28.3±11.9 | 40.9±18.5 | <.001 | 28.3±11.9 | .97 | |||
Rural residence | <.001 | .02 | ||||||
Rural (population <10,000) | 17 | 4 | 26,927 | 12.4 | 122 | 7 | ||
Urban (population ≥10,000) | 435 | 96 | 190,075 | 87.4 | 1,682 | 93 | ||
Neighborhood income quintile | .002 | <.001 | ||||||
Q1 (poorest) | 163 | 36 | 62,642 | 28.8 | 481 | 27 | ||
Q2 | 108 | 24 | 46,543 | 21.4 | 360 | 20 | ||
Q3 | 71 | 16 | 39,429 | 18.1 | 337 | 19 | ||
Q4 | 52 | 12 | 35,325 | 16.2 | 284 | 16 | ||
Q5 (wealthiest) | 57 | 13 | 32,509 | 14.9 | 331 | 18 | ||
ON-Marg dependency quintile | <.001 | .18 | ||||||
Q1 (least dependent) | 137 | 30 | 48,244 | 22.2 | 567 | 31 | ||
Q2 | 98 | 22 | 42,602 | 19.6 | 406 | 23 | ||
Q3 | 95 | 21 | 39,060 | 18.0 | 303 | 17 | ||
Q4 | 52 | 12 | 38,522 | 17.7 | 258 | 14 | ||
Q5 (most dependent) | 66 | 15 | 45,620 | 21.0 | 243 | 13 | ||
ON-Marg residential instability quintile | <.001 | <.001 | ||||||
Q1 (least unstable) | 47 | 10 | 30,277 | 13.9 | 258 | 14 | ||
Q2 | 38 | 8 | 33,062 | 15.2 | 249 | 14 | ||
Q3 | 60 | 13 | 36,933 | 17.0 | 283 | 16 | ||
Q4 | 97 | 21 | 45,864 | 21.1 | 360 | 20 | ||
Q5 (most unstable) | 206 | 45 | 67,912 | 31.2 | 627 | 35 | ||
ON-Marg material deprivation quintile | .34 | .023 | ||||||
Q1 (least deprived) | 70 | 15 | 32,255 | 14.8 | 321 | 18 | ||
Q2 | 64 | 14 | 35,500 | 16.3 | 318 | 18 | ||
Q3 | 71 | 16 | 38,705 | 17.8 | 321 | 18 | ||
Q4 | 96 | 21 | 44,368 | 20.4 | 355 | 20 | ||
Q5 (most deprived) | 147 | 32 | 63,220 | 29.1 | 462 | 26 | ||
ON-Marg ethnic concentration quintile | <.001 | .94 | ||||||
Q1 (lowest ethnic concentration) | 42 | 9 | 41,010 | 18.9 | 182 | 10 | ||
Q2 | 60 | 13 | 41,801 | 19.2 | 245 | 14 | ||
Q3 | 79 | 17 | 40,709 | 18.7 | 330 | 18 | ||
Q4 | 125 | 28 | 42,364 | 19.5 | 485 | 27 | ||
Q5 (highest ethnic concentration) | 142 | 31 | 48,164 | 22.1 | 535 | 30 |
Transgender ED sample (N=728) | Unmatched comparison sample (N=581,708) | Matched comparison sample (N=2,912) | ||||
---|---|---|---|---|---|---|
Diagnosis | N | % | N | % | N | % |
Mood disordersb | 191 | 26 | 90,528 | 15.6 | 566 | 19 |
Anxiety, trauma/stressor, and obsessive-compulsive–related disorders | 278 | 38 | 267,868 | 46.0 | 1,112 | 38 |
Schizophrenia and psychotic disorders | 36 | 5 | 19,874 | 3.4 | 148 | 5 |
Substance-related disorders | 98 | 14 | 145,610 | 25.0 | 729 | 25 |
Personality disorders | 26 | 4 | 4,013 | .7 | 42 | 1 |
Deliberate self-harm | 54 | 7 | 30,053 | 5.2 | 188 | 7 |
Other mental disorders | 45 | 6 | 23,762 | 4.1 | 127 | 4 |
Transgender hospitalized sample (N=454) | Unmatched comparisonsample (N=217,507) | Matched comparisonsample (N=1,808) | ||||
Mood disordersb | 182 | 40 | 77,346 | 35.6 | 626 | 35 |
Anxiety, trauma/stressor, and obsessive-compulsive–related disorders | 63 | 14 | 29,779 | 13.7 | 273 | 15 |
Schizophrenia and psychotic disorders | 60 | 13 | 35,968 | 16.5 | 335 | 19 |
Substance-related disorders | 54 | 12 | 45,363 | 20.9 | 324 | 18 |
Personality disorders | 24 | 5 | 3,889 | 1.8 | 41 | 2 |
Deliberate self-harm | 23 | 5 | 12,962 | 6.0 | 107 | 6 |
Other mental disorders | 48 | 11 | 12,200 | 5.6 | 102 | 6 |
Transgender ED sample (N=728) | Unmatched comparison sample (N=581,708) | Matched comparison sample (N=2,912) | ||||||
---|---|---|---|---|---|---|---|---|
Service use | N | % | N | % | pb | N | % | pb |
Mental health hospitalizations in past 2 years | ||||||||
Any | 156 | 21 | 47,178 | 8.1 | <.001 | 624 | 21 | 1.00 |
M±SD | .4±1.1 | .1±.6 | <.001 | .4±1.0 | .27 | |||
Mental health ED visits in past 2 years | ||||||||
Any | 116 | 16 | 57,306 | 9.9 | <.001 | 425 | 15 | .36 |
M±SD | .3±1.4 | .2±.9 | .001 | .3±1.0 | .19 | |||
Mental health outpatient visits in past 2 years | ||||||||
Any | 605 | 83 | 333,198 | 57.3 | <.001 | 2,420 | 83 | 1.00 |
M±SD | 9.1±12.7 | 4.9±11.8 | <.001 | 8.3±15.5 | .16 | |||
Nonfatal self-harm in past 2 years | ||||||||
Any | 53 | 7 | 13,216 | 2.3 | <.001 | 156 | 5 | .046 |
M±SD | .1±.6 | .0±.3 | <.001 | .1±.4 | .05 | |||
Transgenderhospitalizedsample (N=454) | Unmatchedcomparisonsample(N=217,507) | Matchedcomparisonsample (N=1,808) | ||||||
Mental health hospitalizations in past 2 years | ||||||||
Any | 74 | 16 | 26,009 | 12.0 | .004 | 288 | 16 | .85 |
M±SD | .3±1.0 | .2 ±.7 | .02 | .2 ±.8 | .32 | |||
Mental health ED visits in past 2 years | ||||||||
Any | 214 | 47 | 71,871 | 33.0 | <.001 | 814 | 45 | .42 |
M±SD | 1.1±2.2 | .7±2.3 | <.001 | .9±1.8 | .06 | |||
Mental health outpatient visits in past 2 years | ||||||||
Any | 406 | 89 | 165,400 | 76.0 | <.001 | 1,624 | 90 | .80 |
M±SD | 10.7±13.0 | 8.5±15.0 | <.001 | 10.1±15.8 | .38 | |||
Nonfatal self-harm in past 2 years | ||||||||
Any | 37 | 8 | 12,379 | 5.7 | .024 | 132 | 7 | .54 |
M±SD | .1 ±.5 | .1 ±.4 | .05 | .1 ±.4 | .29 |
Comparing the transgender ED sample with the matched comparison sample of ED users (N=2,912), we found that no significant differences persisted after matching in the dependency or ethnic concentration quintiles. However, the transgender ED sample remained more likely than the matched comparison sample of ED users to be in the lowest neighborhood income quintile (37% versus 27%, χ2=31.00, N=3,640, df=5, p<0.001), highest residential instability quintile (47% versus 38%, χ2=32.17, N=3,574, df=4, p<0.001), and highest material deprivation quintile (33% versus 26%, χ2=19.19, N=3,574, df=4, p<0.001) (Table 1). The transgender ED sample remained more likely than the matched comparison sample of ED users to be diagnosed as having a mood disorder (26% versus 19%) or personality disorder (4% versus 1%) and less likely to be diagnosed as having a substance-related disorder (14% versus 25%) (χ2=66.33, N=3,640, df=6, p<0.001) (Table 2). The differences in diagnoses of anxiety, trauma, or obsessive-compulsive–related disorders and psychotic disorders did not persist after matching. None of the differences in prior mental health service utilization between the transgender ED sample and the unmatched general population of ED users persisted after matching, except that the transgender ED sample still had more prior self-harm (7% versus 5%, χ2=3.98, N=3,640, df=1, p=0.046) (Table 3).
Transgender Hospitalized Sample
Comparing the transgender hospitalized sample (N=454) with the unmatched general population of hospitalized individuals (N=217,507), we found that the transgender hospitalized sample was younger (age=28.3±11.9 versus 40.9±18.5, t=22.30, df=458, p<0.001) and less rural (4% versus 12.4%, χ2=31.07, N=217,454, df=1, p<0.001). They were more likely to be in the lowest quintile for neighborhood income (36% versus 28.8%, χ2=19.10, N=217,961, df=5, p=0.002) and dependency (30% versus 22.2%, χ2=35.07, N=214,496, df=4, p<0.001) and in the highest quintile for residential instability (45% versus 31.2%, χ2=50.75, N=214,496, df=4, p<0.001) and ethnic concentration (31% versus 22.1%, χ2=63.07, N=214,496, df=4, p<0.001) (Table 1). The transgender hospitalized sample was more likely than the unmatched general population of hospitalized individuals to be diagnosed as having a mood disorder (40% versus 35.6%) or personality disorder (5% versus 1.8%) (χ2=74.46, N=217,961, df=6, p<0.001) (Table 2). They were less likely to be diagnosed as having schizophrenia or a psychotic disorder (13% versus 16.5%) or substance-related disorder (12% versus 20.9%). The transgender hospitalized sample had much greater prior psychiatric service use than did the unmatched general population of hospitalized individuals, including having more prior hospitalizations (16% versus 12.0%, χ2=8.11, N=217,961, df=1, p=0.004), ED visits (47% versus 33.0%, χ2=40.66, N=217,961, df=1, p<0.001), outpatient visits (89% versus 76.0%, χ2=44.58, N=217,961, df=1, p<0.001), and visits for deliberate self-harm (8% versus 5.7%, χ2=5.10, N=217,961, df=1, p=0.024) (Table 3).
Comparison of the transgender hospitalized sample with the matched comparison sample of hospitalized individuals (N=1,808) revealed no persistent differences in the dependency or ethnic concentration quintiles. However, the transgender hospitalized sample remained more likely to be in the lowest neighborhood income quintile (36% versus 27%, χ2=27.02, N=2,262, df=5, p<0.001) and the highest residential instability quintile (45% versus 35%, χ2=25.48, N=2,225, df=4, p<0.001). The transgender hospitalized sample was more likely to be in the highest material deprivation quintile than the matched comparison sample of hospitalized individuals (32% versus 26%, χ2=11.36, N=2,225, df=4, p=0.023) (Table 1). The transgender hospitalized sample remained more likely than the matched comparison sample of hospitalized individuals to be diagnosed as having a mood (40% versus 35%) or personality disorder (5% versus 2%) and less likely to be diagnosed as having schizophrenia or a psychotic disorder (13% versus 19%) or substance-related disorder (12% versus 18%) (χ2=42.42, N=2,262, df=6, p<0.001) (Table 2). None of the differences in prior mental health service utilization between the transgender hospitalized sample and the unmatched general population of hospitalized individuals persisted after matching (Table 3).
Discussion
Compared with the unmatched comparison group presenting for acute mental health care, transgender individuals in our samples were younger, were more likely to experience marginalization, had different diagnostic patterns for their acute mental health care presentation, and had more prior mental health care utilization (as an outpatient, in the ED, and for hospitalization). Matching was required to isolate the unique contribution of transgender status on sociodemographic and clinical factors that were not adequately explained by differences in age, region of residence, and prior mental health care utilization. Even after matching, data indicated that transgender individuals were more likely to live in lower-income neighborhoods and areas of greater residential instability and material deprivation. Transgender individuals were more likely to be diagnosed as having a mood disorder and twice as likely to be diagnosed as having a personality disorder during their acute mental health care presentation. They were less likely to be diagnosed as having a substance-related disorder and, among those who were hospitalized, less likely to be diagnosed as having a psychotic disorder. Transgender individuals in the ED sample were more likely to have presented to acute care for self-harm, even after we accounted for other differences in mental health care use through matching.
The diagnostic patterns in the transgender samples are consistent with the two- to fivefold higher prevalence of depression among transgender individuals than in the general population (3–5, 14, 15). Although transgender individuals also have increased rates of anxiety- and trauma-related disorder diagnoses (3, 14, 15, 20), this pattern was not reflected in our results. The multifold increase in personality disorder diagnoses we found in the transgender samples in this study may be partially explained by the minority stress model (11, 21). Symptoms and behaviors consistent with diagnostic criteria of borderline personality disorder may be better understood as reactions to stress and oppression experienced by members of minority groups (21, 22). A society that subscribes to a rigid sex-gender binary viewpoint (23), where transgender individuals are regularly misgendered and face daily discrimination and marginalization, creates environments that can produce such symptoms and behaviors in response. Diagnosis of borderline personality disorder is also prone to health care provider bias (24, 25), which may have influenced our results. Furthermore, transgender people are frequently pathologized, and those seeking medical transition are required to have a mental health assessment, both of which can lead to psychiatric diagnoses (26).
Our study provides evidence for an association between acute mental health care utilization and increased measures of marginalization in the transgender samples. Our study echoes previous U.S. research, which found that transgender inpatients were more likely to be in the lowest neighborhood income quartile than were cisgender inpatients (15), and extends that finding to transgender individuals presenting to the ED for mental health care. Our study clarifies that residential instability and material deprivation disproportionately affect transgender individuals presenting for acute psychiatric care. Residential instability refers to high rates of family or housing instability, whereas material deprivation assesses indicators of poverty and difficulty accessing basic material needs (27). That these findings were persistent after matching adds strength to the possibility that experiences of marginalization are an important contributor to acute mental health care utilization for transgender individuals. This possibility is consistent with evidence that high rates of mental illness among transgender people are associated with elevated rates of homelessness, unemployment, and poverty (3, 7, 8, 28, 29).
A strength of the study was the use of two comparison samples: one unmatched and one matched on key characteristics. This method allowed for an understanding of the ways transgender individuals differ from samples from the general population who access acute mental health care and of the unique contributions of transgender status to sociodemographic and clinical factors associated with acute mental health care presentation. The large sample of transgender individuals also allowed for adequate power to examine differences in multiple characteristics associated with psychiatric hospitalizations and ED visits for this population. In total, the unmatched samples are inclusive of nearly every psychiatric hospitalization and ED visit in Ontario within the study period. Unlike previous studies that have relied on diagnostic definitions and involvement with medical transition to identify transgender individuals (12, 13), the current study used self-reported gender identities to identify individuals, which is more representative of the total transgender population.
This study had several limitations. Transgender individuals were identified from clinics in larger cities, which may not be representative of individuals in smaller cities and rural areas. Because the unmatched comparison groups represented all regions in Ontario, geographic variation in psychiatric service availability and utilization may have affected results. Matching was thus important to control for regional differences. Some of the sociodemographic differences may reflect our specific sample of transgender individuals (16) and may not be characteristic of all transgender individuals with acute psychiatric care utilization. Because they represented a clinical population, individuals in the samples were also more likely to have had prior outpatient visits than other transgender individuals, which may limit generalizability of findings. Prior outpatient visits for transgender individuals can include assessments to facilitate access to medical transition. These assessments may differentially contribute to the number of prior outpatient visits for transgender samples. However, these visits should have limited impact on other measures of prior mental health care. Matching likely did not remove all variation in mental illness severity between transgender and comparison groups, which may have also affected results. Because transgender status was based on self-report, some individuals may not have disclosed their gender identity and thus would have been excluded from the transgender sample. The comparison samples likely included transgender individuals who were not identified as part of the transgender samples. However, the vast majority of individuals who comprised the comparison samples were likely to be cisgender individuals (13, 15).
Conclusions
This study found that transgender individuals presenting for acute mental health care were more likely to experience marginalization than cisgender individuals and to present to acute care with different diagnostic patterns. More research is warranted into the experiences of transgender individuals presenting for acute mental health care and the factors associated with their presentation, particularly in regard to how experiences of marginalization and discrimination may play a role.
1 : Transgender-inclusive care. CMAJ 2019; 191:E79Crossref, Medline, Google Scholar
2 : Transgender demographics: a household probability sample of US adults, 2014. Am J Public Health 2017; 107:213–215Crossref, Medline, Google Scholar
3 : Transgender people: health at the margins of society. Lancet 2016; 388:390–400Crossref, Medline, Google Scholar
4 : Depression in male-to-female transgender Ontarians: results from the Trans PULSE Project. Can J Commun Ment Health 2011; 30:113–133Crossref, Google Scholar
5 : Prevalence of and risk and protective factors for depression in female-to-male transgender Ontarians: Trans PULSE Project. Can J Commun Ment Health 2011; 30:135–155Crossref, Google Scholar
6 : Ontario’s Trans Communities and Suicide: Transphobia is Bad for our Health. Trans PULSE e-Bulletin, November 12, 2010. http://transpulseproject.ca/research/ontarios-trans-communities-and-suicide. Accessed December 11, 2020Google Scholar
7 : Gender minority stress and depressive symptoms in transitioned Swiss transpersons. Biomed Res Int (Epub April 19, 2018)Google Scholar
8 : Educational attainment of transgender adults: does the timing of transgender identity milestones matter? Soc Sci Res 2018; 74:146–160Crossref, Medline, Google Scholar
9 : Intervenable factors associated with suicide risk in transgender persons: a respondent driven sampling study in Ontario, Canada. BMC Public Health 2015; 15:525Crossref, Medline, Google Scholar
10 : Global health burden and needs of transgender populations: a review. Lancet 2016; 388:412–436Crossref, Medline, Google Scholar
11 : Stigma and minority stress as social determinants of health among lesbian, gay, bisexual, and transgender youth: research evidence and clinical implications. Pediatr Clin North Am 2016; 63:985–997Crossref, Medline, Google Scholar
12 : Identifying gender minority patients’ health and health care needs in administrative claims data. Health Aff 2018; 37:413–420Crossref, Google Scholar
13 : Trends in mental health care use in Medicare from 2009 to 2014 by gender minority and disability status. LGBT Health 2019; 6:297–305Crossref, Medline, Google Scholar
14 : Mental health of transgender youth in care at an adolescent urban community health center: a matched retrospective cohort study. J Adolesc Health 2015; 56:274–279Crossref, Medline, Google Scholar
15 : Psychiatric disorders in the US transgender population. Ann Epidemiol 2019; 39:1–7.e1Crossref, Medline, Google Scholar
16 : Assessment of health conditions and health service use among transgender patients in Canada. JAMA Netw Open 2020; 3:e2015036Crossref, Medline, Google Scholar
17 : Characterising violent deaths of undetermined intent: a population-based study, 1999–2012. Inj Prev 2018; 24:424–430Crossref, Medline, Google Scholar
18 : Cancer diagnosis and risk of suicide after accounting for prediagnosis psychiatric care: a matched-cohort study of patients with incident solid-organ malignancies. Cancer 2019; 125:2886–2895Medline, Google Scholar
19 : Discharge and post-discharge outcomes of psychiatric inpatients with a lifetime history of exposure to interpersonal trauma: a population-based study. Gen Hosp Psychiatry 2020; 65:82–90Crossref, Medline, Google Scholar
20 : Mental health diagnoses among transgender patients in the clinical setting: an all-payer electronic health record study. Transgend Health 2019; 4:313–315Crossref, Medline, Google Scholar
21 : Personality disorders and personality profiles in a sample of transgender individuals requesting gender‐affirming treatments. Int J Environ Res Public Health 2020; 17:5–7Crossref, Google Scholar
22 Distinguishing and addressing gender minority stress and borderline personality symptoms. Harv Rev Psychiatry 2019; 27:317–325Crossref, Medline, Google Scholar
23 : Defending the sex/gender binary: the role of gender identification and need for closure. Soc Psychol Personal Sci (Epub July 16, 2020)Google Scholar
24 : The impact of client sexual orientation and gender on clinical judgments and diagnosis of borderline personality disorder. J Clin Psychol 2006; 62:751–770Crossref, Medline, Google Scholar
25 : Is there a bias in the diagnosis of borderline personality disorder among lesbian, gay, and bisexual patients? Assessment 2021; 28:724–738Crossref, Medline, Google Scholar
26 : Trans health care from a depathologization and human rights perspective. Public Health Rev 2020; 41:3Crossref, Medline, Google Scholar
27 : 2016 Ontario Marginalization Index: User Guide. Toronto,
28 : Suicidality among trans people in Ontario: implications for social work and social justice. Serv Soc 2013; 59:35–62Crossref, Google Scholar
29 : A transgender refugee woman experiencing posttraumatic stress disorder symptoms and homelessness. CMAJ 2020; 192:E9–E11Crossref, Medline, Google Scholar