INTRODUCTION

Suicide prevention is a critical public health priority in the United States (US). Suicide remains the tenth leading cause of death, with over 47,000 recorded suicides in 2019.1 Suicidal ideation, which describes a range of thoughts related to injury and death with varying degrees of severity,2 is a prevalent antecedent to suicide.3,4 Of those with suicidal ideation, approximately one in three go on to attempt suicide,3 and an estimated one in 30 attempts results in death.5 The identification of suicidal ideation provides an important opportunity for prevention and intervention,6,7 and the number of suicide prevention interventions has increased over time, with new programs developed for specific at-risk populations.8,9,10 Yet, the number of completed suicides has been rising for most of the last two decades.1 Population data suggest that increased access to firearms and lethal means, worsening structural marginalization,11 and the COVID-19 pandemic have exacerbated the risk for suicide.12,13

Identifying which populations are at highest risk is an important step towards improving prevention efforts to mitigate suicide risk. Several socio-demographic and -economic factors have been associated with increased suicide risk, including younger age,4 lower educational attainment,14 homelessness,15,16 long-term unemployment,17 and sexual minority identity18. In recent years, there have been calls to examine suicide risk among specific minority populations, including transgender and gender diverse (TGD) individuals, e.g., transgender men, transgender women, nonbinary (NB) and other non-cisgender persons.19,20 Given identified disparities in psychological and behavioral health issues—including depression, anxiety, and substance use—relative to their cisgender peers,21,22,23,24 as well as disproportionate exposure to interpersonal and structural violence,25,26 TGD individuals may be at increased risk for suicide.

Gender differences in suicide risk between cisgender men and women have been well studied: women are more likely to attempt whereas men are more likely to die by suicide.27,28 Less is known about suicide among TGD individuals, largely due to the lack of systematic gender identity data collection and reporting (e.g., in death certificates)29. Existing data indicate that TGD individuals are at higher risk for suicide. For example, the 2015 US Transgender Survey, the largest epidemiologic report of TGD health in the US to date, found critically high levels of suicidal ideation: among almost 28,000 TGD participants, 48% endorsed serious thoughts about attempting suicide in the past year, and 82% had such thoughts at some point during their lifetime.30 This compares to a prevalence of approximately 2–10% for past-year ideation, and 5–15% for lifetime ideation in the general US adult population.3,4,31,32

The increased prevalence and severity of psychological distress and suicidality among TGD individuals may best be understood within the gender minority stress model. Building on the contributions of Brooks33 and Meyer34, Hendricks and Testa proposed that distal stressors (e.g., discrimination), which are external to the individual, and proximal stressors (e.g., internalized transphobia), which are internal and are often the byproduct of repeated exposure to distal stressors, act on biobehavioral pathways to contribute to the health disparities impacting TGD individuals.35 For example, bullying, which disproportionately affects TGD youth,36,37 is a distal stressor that may contribute to low self-esteem, and feelings of worthlessness and burdensomeness,38,39 which are risk factors for suicide.35,40 Consistent bullying may also lead to the development of proximal stressors; for instance, TGD individuals may conceal their identity and experience anxiety about anticipated discrimination, and thus be less buffered by protective factors, including lower community connectedness and social support.35,41,42 Similar pathways to increased suicide risk have been examined among sexual minority youth.41,42

Although other US-based studies have assessed suicidal ideation among TGD individuals, the literature in this area is largely characterized by small samples, primarily among youth, without sufficient power to examine subgroup differences (e.g., among transgender men, transgender women, NB individuals), and often in the absence of cisgender comparator groups.22 With the ultimate aim of improving prevention and intervention efforts, the current study builds on existing work by evaluating suicidal ideation among TGD individuals with increased nuance in several key ways: (1) examining suicidal ideation in a large sample of almost 30,000 community participants; (2) distinguishing rates of ideation among transgender men, transgender women, NB persons assigned male at birth (AMAB), and NB persons assigned female at birth (AFAB); and (3) including cisgender men and women receiving care at the same center as comparators.

METHODS

Participants and Setting

Participants included 29,988 patients aged 18 and over who presented for care at Fenway Health between October 1, 2015, and October 1, 2018. Fenway Health is a federally qualified community health center in the Northeastern US with a mission to improve the health of lesbian, gay, bisexual, transgender, intersex, asexual and other sexual and gender minority community members.43 The community health center provides primary and specialty care, including obstetrics and gynecology, sexual health, gender affirming care, dentistry, optometry, behavioral health, and public health services.

Procedures and Measures

As part of routine care procedures, participants completed electronic patient-reported outcome measures (ePROs) on a tablet while they waited for a medical appointment in the clinic.44 Patients were informed that the surveys were voluntary and confidential, and survey data were stored in the electronic health record (EHR). The Fenway Health Institutional Review Board approved all study procedures and provided a waiver of consent. The ePROs included the following measures:

Demographics

At time of registration, participants completed a questionnaire pertaining to age, current gender identity, sex assigned at birth, sexual orientation, race, ethnicity, and insurance status, which was used as a proxy for income45. TGD is an inclusive and dynamic term defined variably in different communities.46 For the current study, we describe TGD participants as those who endorsed a current gender identity different from their assigned sex at birth, and refer to this combined sex-gender variable as a gender category throughout. Participants were grouped into six categories using a two-step model.47 Responses to both a current gender identity (“What is your gender?”) and an assigned sex (“What was your sex assigned at birth?”) questions were used to categorize participants. An individual who was assigned female at birth and (a) endorsed female/woman as current gender identity was categorized as a cisgender woman; (b) endorsed male/man as current gender was categorized as a transgender man; and (c) endorsed genderqueer or not exclusively male or female as current gender identity was categorized as a NB AFAB person. Similarly, an individual who was assigned male at birth and (a) endorsed female/woman as current gender identity was categorized as a transgender woman; (b) endorsed male/man as current gender was categorized as a cisgender man; and (c) endorsed genderqueer as current gender identity was categorized as a NB AMAB person.

Suicidal Ideation

Ideation was assessed using the 9-item Patient Health Questionnaire (PHQ-9),48 which is widely used in healthcare settings. The PHQ-9 is a brief measure of depression, which assesses the frequency of symptoms over the past 2 weeks and has established cut-off scores that are indicative of likely depression at mild, moderate, and severe levels. To determine the presence of suicidal ideation, responses to item 9, which assesses the frequency of “thoughts that you would be better off dead, or of hurting yourself” in the past 2 weeks, were dichotomized; participants who endorsed “several days,” “more than half the days,” and “nearly every day” were considered to have suicidal ideation, whereas participants who endorsed “not at all” were not.

Analyses

Data were extracted from the EHR via Structured Query Language.49 For participants who attended the health center more than once between 2015 and 2018, the first complete record of the PHQ-9 was used. Data were analyzed using IBM SPSS for Windows (version 25). Descriptive statistics were calculated for the socio-demographic data, both for the overall sample (presented in the analysis section) and across the gender categories (presented in Table 1). Differences in the proportion of participants endorsing past 2-week suicidal ideation across the gender categories were examined using chi-square tests.

Table 1 Socio-demographic Characteristics by Gender Category
Full size table

A two-step logistic regression was conducted to examine the associations between gender category and the presence/absence of suicidal ideation. The first step of the model included age, which was measured continuously, and several sociodemographic variables that were dichotomized for ease of interpretation, including ethnicity (Latinx/a/o, non-Latinx/a/o), race (of color—including Native American, Pacific Islander, Asian or Asian American, Black or African American, Multiracial—white), sexual orientation (LGBQ+, heterosexual), and insurance status (public, private/self-pay). The six-level gender category was added in the second step of the model, with cisgender men, the largest group with the lowest rate of ideation, as the referent group, to assess the degree to which gender accounted for variability in suicidal ideation beyond the variables included in the first step. In addition to dichotomizing race as described above, an additional regression examined whether suicidal ideation differed between white TGD participants and TGD participants of color, who are likely impacted by compounded stressors (gender and racial minority stressors),50,51 and may thus experience higher rates of suicidal ideation. This subanalysis was driven by recent findings that TGD persons of color have more severe depression, anxiety, and psychological distress than their white TGD peers.52,53

RESULTS

Demographics

The sample included 29,988 participants who had a mean age of 33.9 years (SD=13.1). With respect to race and ethnicity, 69.8% (n=20,946) of the sample was white, 6.5% (n=1945) of the sample was Black, 8.3% (n=2497) of the sample was Asian, 0.4% (n=115) of the sample was Native American or Pacific Islander, and 19.1% (n=1557) were Latinx/a/o. Overall, 49.6% (n=14,873) identified as LGBQ+, and 83.7% (n=25,092) had access to healthcare via private insurance or by paying out of pocket. The majority of the sample were cisgender men (52.3%), followed by cisgender women (32.1%), transgender women (3.3%), transgender men (3.3%), NB AFAB individuals (2.5%), and NB AMAB individuals (1.4%). Demographic data by gender category are presented in Table 1.

Proportions Endorsing Suicidal Ideation

The proportion of patients who endorsed any ideation in the past 2 weeks differed significantly across gender category, χ2(5, n=25,959)=906.454, p<0.001 (see Table 2). Bonferroni-adjusted post hoc z-tests revealed that significantly more transgender women (23.6%) and NB AMAB individuals (26.7%) endorsed any suicidal ideation, compared to transgender men (17.4%), cisgender women (6.6%), and cisgender men (6.1%). In addition, significantly more NB AFAB individuals (22.5%), and transgender men endorsed any suicidal ideation relative to cisgender men and women.

Table 2 Differences in Prevalence Suicidal Ideation by Gender Category
Full size table

Regression Predicting Suicidal Ideation

At the first step of the regression, younger age (OR=0.98, CI=0.97, 0.99), public or grants-based insurance (OR=3.33, CI=2.59, 4.28), and sexual minority identity (OR=1.96, CI=1.53, 2.52) were associated with significantly higher odds of suicidal ideation (all ps<0.001; see Table 3). When gender category was added to the model in the second step, all four TGD subgroups had significantly higher odds of suicidal ideation relative to cisgender men; transgender men had 2.08 times higher odds (p=0.003, CI=1.29, 3.36), transgender women had 3.08 times higher odds (p<0.001, CI=2.05, 4.63), NB AMAB individuals had 3.55 times higher odds (p<0.001, CI=1.86, 6.77), and NB AFAB individuals had 2.49 times higher odds (CI=1.52, 4.07). Age, insurance status, and sexual orientation remained statistically significant (see Table 3). Finally, there were no differences in suicidal ideation between TGD participants of color and White TGD participants (ps>.05).

Table 3 Logistic Regression Examining the Relationship Between Sociodemographic Variables and the Odds of Current Suicidal Ideation
Full size table

DISCUSSION

Study Findings

Among 30,000 community health patients, those who were younger, LGBQ+, TGD, and publicly insured endorsed significantly higher odds of suicidal ideation. In line with minority stress theory, our findings indicate that current suicidal ideation was more prevalent among TGD individuals (17–26% compared to 6–7% among cisgender individuals). Additionally, these disparities were apparent in an adjusted model, in the presence of sociodemographic variables known to contribute to suicide risk (e.g., age, race, ethnicity, sexual orientation, insurance status)4,14,15,16,17,18,54, suggesting that stressors specific to TGD identity may be linked with suicide risk. It is noteworthy that all TGD groups endorsed concerningly high suicidal ideation, which warrants intervention.

A growing body of literature has examined disparities between binary (i.e., transgender men and women) and NB TGD individuals, in order to explain subgroup disparities in psychological distress.55,56,57 Indeed, an analysis utilizing these data indicated differences in mental health symptom severity by gender category: a higher proportion of transgender women and NB patients met the clinical PHQ-9 cut-off score for depression compared to transgender men, cisgender women, and cisgender men.21 Transgender women and NB AMAB individuals in our sample had worse outcomes not only compared to cisgender men and women, but also compared to transgender men. The specific pattern of findings observed here aligns with and extends recent findings reported among TGD youth. For example, in a cross-sectional survey, Newcomb and colleagues found that TGD AMAB youth experienced worse psychosocial outcomes than TGD AFAB youth, including lower social support, higher rates of substance use, more traumatic experiences and higher rates of suicidal ideation and prior attempt.58

The specific mechanisms underlying increased risk among TGD AMAB individuals have not been thoroughly explored, though it is possible that these individuals face higher exposure to stressors even compared to TGD AFAB individuals. Femmephobia, which refers to the regulation of femininity, whereby masculinity is hierarchically valued over femininity,59 may contribute to increased suicidal ideation among TGD AMAB individuals. In a recent examination of the role of femmephobia in experiences of prejudice and oppression among LGBTQ+ individuals, Hoskin connected prejudice to femininity, noting that “for those AFAB, masculinity is a social promotion,” whereas AMAB individuals “experience a demotion for their entrances into the feminine.”60 Moreover, participants in Hoskins’ study reported that perceived femininity was often used as justification for sexual and physical violence, discrimination, and barriers to resources (e.g., housing). Indeed, transgender women and other TGD AMAB individuals are disproportionately impacted by interpersonal victimization and violence.25,26 Similarly, TGD individuals may experience disparate inter- and intra-personal stressors related to their external visibility as non-cisgender (see e.g.,61,62,63,64). These negative experiences associated with how well TGD persons adhere to or transgress from cultural norms of femininity and masculinity may exacerbate suicidal ideation, especially among AMAB persons, for whom the perceived transgressions may be greater.

Implications and Future Directions

These findings have important implications for understanding the pathology of suicide among TGD individuals, as well as for assessment and intervention in clinical and research contexts. Clinical assessment of suicidal ideation and risk among TGD populations should be consistent, comprehensive, and directed, rather than relying on spontaneous patient report.65 Research assessment is also currently limited by assessment protocols, which could be expanded to improve prevention research. For example, although the PHQ-9 is widely used to assess the prevalence of suicidal ideation in research studies,66,67,68 the predictive value of the frequency of thoughts of death or self-harm over the past 2 weeks remains unclear.66,69,70 Moreover, because the item assesses both passive thoughts about death and desire for self-harm, it may yield high “false positives,”68 potentially straining limited mental health resources among providers caring for TGD patients. With the PHQ-9 not sufficiently valid for use as a suicide screening measure,71,72,73 tools like the Columbia Suicide Severity Rating Scale (C-SSRS)74 may be more appropriate for use among groups that have high rates of suicidal risk, both in clinical and research settings. Such tools allow for thorough measurement of passive thoughts about death, desire for self-harm, suicidal ideation, and intent, which may all be differentially predictive of future attempt among TGD individuals.40,75

Given the striking disparities in suicidal ideation between cisgender and TGD individuals, there is also a dire need for consistent assessment of gender identity, which will facilitate the identification of suicide-related outcomes among specific TGD groups. Currently, collection of gender identity demographic data by clinicians and researchers, as well as in national surveillance systems, which serve as the primary source of data related to deaths by suicide, is at best inconsistent.29,44,76,77,78 Systematic and continuous collection of gender identity and modality76 data will help to address methodological limitations in monitoring suicidal ideation, behavior, and associated deaths in this population. Appropriate collection of TGD identity data will also allow for needed moderation analyses by gender identity, modality, or category in epidemiological and etiological studies of suicide, as well as for the investigation of proximal and distal stress mechanisms underlying suicide risk that may be unique to specific subgroups (e.g., among NB individuals, or among persons AMAB).

Finally, evidence-based psychosocial interventions that are developed or modified for TGD persons may be especially efficacious in interrupting the transition from ideation to attempt.6,79 Researchers have already adapted cognitive behavioral therapy protocols to improve mental health outcomes among LGBQ individuals (e.g., ESTEEM)80, following a model for adapting interventions to be LGBQ-affirmative.81 This model, which relies on LGBQ affirmative principles (e.g., highlighting how symptoms of depression and anxiety can be normal responses to minority stress, acknowledging how early and ongoing experiences with minority stress can teach individuals negative lessons about themselves), could be used to adapt and test evidence-based interventions specific to reducing suicidal ideation among TGD individuals. Such interventions may be delivered flexibly via modular approaches, such that modules on TGD-specific minority stress could be delivered to all TGD persons, with additional modules on femmephobia, for example, delivered to TGD AMAB individuals. Qualitative research that centers community knowledge and priorities, and implementation research that situates suicide prevention interventions in relevant settings (e.g., in primary care,65,82 via tele-health83,84) will be invaluable to inform the development or adaptation of interventions that are responsive to the needs of TGD communities.

Limitations

There are a few limitations with respect to measurement tools, and the data that are maintained in the EHR that are worth noting. In line with the limitations reviewed above related to the predictive value of item nine of the PHQ-9,70,71,73 the predictive power of suicidal ideation has also been questioned. Specifically, although ideation is prerequisite for an attempt, which is in turn prerequisite for death by suicide, the progression from ideation to death consists of distinct, multifactorial processes, which are difficult to predict.40 As such, we refer to item 9 as indicative of suicidal ideation rather than of suicidal risk more broadly, and we discuss suicidal ideation as indicative of severe psychological distress, which warrants intervention even in the absence of injury and death.

Similarly, the gender options included in the health record are restrictive, an issue that is not uncommon in research and clinical settings;76,85 we could only examine differences among the six gender categories, and could not examine more complex and layered aspects of gender, which would allow us to examine the impacts of minority stress and femmephobia. These limited identity options are likely not reflective of the diverse and nuanced TGD experiences, and therefore may compromise clinical applicability for this heterogeneous group.85 Finally, patients at the community health center were relatively young, white, and privately insured (indicating they were employed or students), and therefore these findings may not generalize to other settings that serve more diverse communities.86,87,88 Community-embedded research can help to enrich our understanding of TGD experiences, and identify research and treatment priorities.

CONCLUSION

These findings suggest that TGD individuals are two–four times more likely than cisgender men to endorse current suicidal ideation. Suicidal ideation was tiered such that a larger proportion of transgender women and NB AMAB individuals endorsed current suicidal ideation compared to transgender men, and a larger proportion of transgender men and NB AFAB individuals endorsed ideation compared to cisgender men and women. These findings highlight the need for more consistent collection of data related to gender identity by health centers, and the need for better consideration of unique risk profiles by clinicians. Further research to examine the underlying mechanisms contributing to suicide risk among TGD populations and to heightened risk among TGD AMAB individuals is sorely needed, as is work that develops or adapts existing suicide prevention interventions to address gender minority stressors, especially in the context of primary care. Lastly, although the current study highlights differences among TGD subgroups, it is notable that between 1-in-4 and 1-in-6 of all TGD patients endorsed suicidal risk, underlining a critical need for evidence-based and accessible interventions for TGD individuals.