Causes of death of transgender and gender diverse veterans

Abstract

Introduction:

Transgender and gender diverse (TGD) veterans face numerous challenges due to stigma and marginalization, which have a significant impact on their health and well-being. However, there is insufficient data on cause-specific mortality in TGD veteran populations in the United States. The purpose of this study was to describe the leading causes of death in a sample of TGD veterans who received care from the Veterans Health Administration (VHA).

Methods:

A secondary data analysis was conducted using VHA electronic health record (EHR) data matched with death certificate records from the National Death Index (NDI) from October 1, 1999 to December 31, 2019. Using record axis codes from NDI data, the 25 most frequent underlying and all causes of death were summarized.

Results:

Deaths occurred in 1,415 TGD veterans. Ranking by any mention on the death certificate, mental and behavioral disorders due to psychoactive substance use (17.2%), conduction disorders and cardiac dysrhythmias (15.3%), chronic obstructive pulmonary disease (15.1%), diabetes mellitus (13.9%), and chronic ischemic heart disease (13.3%) were the top five causes of death. Three distinct methods of suicide appeared as the 7^th (firearms), 17^th (self-poisoning), and 24^th (hanging) underlying causes of death for TGD veterans.

Conclusions:

Targeted prevention efforts or interventions to reduce the frequency and severity of causes of death, particularly mental and behavioral health disorders and metabolic disorders could prevent premature mortality among TGD adults.

Introduction

Due to downstream effects of stigma and discrimination,^1–3 transgender and gender diverse (TGD) people face disproportionately high rates of diseases of physical health such HIV and cardiometabolic diseases and mental health including depression and suicidal ideation and attempt.^4–8 Additionally, TGD people experience high rates of incarceration,^7,9 violence victimization,^7,10 and well-documented barriers to adequate food, housing, and health care compared to cisgender (i.e., non-TGD) people, all of which may increase the risk of premature death.¹¹ Similar social and health disparities have been documented for TGD veterans compared to their cisgender peers, including suicide attempts.¹²

Physical and mental health disparities that TGD people experience may contribute to differing causes of death compared to cisgender individuals. TGD-specific mortality data is needed to adequately detect, understand, and eliminate health disparities. However, few studies have analyzed mortality data for TGD populations primarily because gender identity information is not included on death certificates or in mortality databases.¹³ To circumvent this challenge, researchers have used administrative databases, such as insurance claims or electronic health records (EHR) to analyze mortality among TGD people. Two studies using EHR data from the Veterans Health Administration (VHA) found high risk of suicide mortality among TGD people.^14–15 Two other studies using insurance data found that TGD people had a higher risk of all-cause mortality compared to their cisgender counterparts.^16–17

Since 1949, cause-of-death statistics have been based on the underlying cause of death, defined as “the disease or injury which initiated the train of events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury.”¹⁸ Deaths may be ascribed only one cause, but in some cases, a death represents a consequence of several co-occurring conditions, of which there may be no obvious single underlying cause. Thus, a call-to-action began over 35 years ago to encourage the use of multiple cause of death data to more accurately capture concurrent disease processes that may contribute to death.¹⁹

Despite the value of multiple cause of death data, they have yet to be applied to TGD populations. This study aims to identify the leading underlying and contributing causes of death among TGD veterans. Based on prior literature,^14–17 suicide and cardiovascular events were expected to be leading causes of death for TGD veterans when considering both underlying and multiple causes. Findings can be utilized to inform both research and interventions to address mortality-related disparities and improve survival for TGD veterans. To the authors’ knowledge, this is the first study to explore the nuances of mortality for a population of TGD individuals.

Methods

Study Sample

A secondary analysis was conducted using EHR data from the VHA Corporate Data Warehouse (CDW) from October 1, 1999 to December 31, 2016. The CDW contains administrative and health-related data on all veterans enrolled in VHA care.²⁰ The TGD veterans in the sample were identified based on documentation in health records of any of the following International Classification of Diseases (ICD-9 or ICD-10) codes: 302.3 (Transvestic fetishism), 302.5 (Transsexualism), 302.6 (Gender identity disorder in children), 302.85 (Gender identity disorder in adolescents or adults), F64.X (Gender identity disorders), F65.1 (Transvestic fetishism), and Z87.890 (Personal history of sex reassignment). Identification of TGD veterans based on self-identified gender identity was not possible because the VHA did not begin implementation of a self-identified gender identity field in health records until 2016.²¹ Additionally, entering self-identified gender identity data is still not a universal practice across the VHA. This ICD-based method for identifying TGD patients has been used consistently in health services research using data from VHA,^12,22–24 Centers for Medicare and Medicaid Service,⁵ and private insurance.^16–17 In VHA, its validity has been substantiated by reviewing clinical text notes.^22,25 This study was approved by the Institutional Review Board of the VA Pittsburgh Healthcare System.

Measures

The outcome of interest was cause of death obtained from the National Death Index (NDI), a centralized database of death records housed by the National Center for Health Statistics (NCHS).²⁶ NDI data and VHA administrative and health records are cross-referenced using unique identifiers and probabilistic matching to determine if a patient died, and, if so, to extract relevant mortality information through the Veteran Affairs Mortality Data Repository to provide information from death certificates for all VHA patients.²⁷ Mortality data through December 31, 2019 was used for this study.

Death certificates were used to ascertain cause of death for each veteran. Part I of the death certificate lists the conditions which directly lead to death, forming a causal sequence from the underlying to the immediate cause of death. Part II lists all other conditions which contribute to death but are not in the direct causal pathway.²⁶ The exact conditions written on the death certificate are processed by the NCHS and reported as record axis codes. The first record axis code always refers to the underlying cause of death and the remaining record axis codes refer to any additional causes of death, for up to a total of 20 record axis codes. Further information on the process of creating record axis codes is available from the NCHS.²⁶ The specific ICD-10 codes for causes of death were subsequently categorized into a total of 358 causes of death, consistent with NCHS reporting.²⁸

Select demographic information based on EHR were extracted for all veterans. Age at death was used as a continuous variable. Race, ethnicity, and marital status were also extracted from the veterans’ most current records. Race and ethnicity were combined and categorized as non-Hispanic white, non-Hispanic Black, non-Hispanic other, Hispanic, and unknown/missing. Marital status was coded as married, not married (i.e., separated, divorced, widowed, never married, single) and unknown/missing. Sex was not included as a variable. Veterans may change their sex field in the EHR to be consistent with their gender identity per VHA policy.²⁹ Therefore, it could not be determined whether the EHR-based sex marker reflected a veterans’ gender identity or sex assigned at birth.

Statistical Analysis

Descriptive statistics were used to summarize the demographic characteristics of TGD veterans who died across the observation period. From the underlying causes of death listed in the record axis codes, the 25 most frequent causes of death were identified and the percentage of all deaths attributed to each underlying cause were calculated. Multiple causes of death were summarized by identifying the top 25 causes of death mentioned anywhere in the record axis codes (including the underlying cause) and the percentage of all deaths attributed to each cause. All analyses were conducted using Stata/MP Version 15 (College Station, TX: StataCorp LLC).

Results

Out of the original cohort of 8,981 TGD veterans, death occurred in 1,415 (15.8%) veterans across the observation period. The average age of death was 56.4 years. The majority of TGD veterans who died were non-Hispanic white (71.8%), followed by non-Hispanic Black (7.4%), non-Hispanic other race (4.0%), and Hispanic (1.6%). About 15% of TGD veterans who died had an unknown or missing race. Finally, a majority of the sample was not married (75.8%) (Table 1).

Table 1.

Characteristics of Transgender and Gender Diverse (TGD) Veterans in the Veterans Health Administration, n = 1,415

Characteristics	n (%)
Age,a years (SD)	56.4 (15.6)
Race and ethnicity
White, non-Hispanic	1,016 (71.8)
Black, non-Hispanic	105 (7.4)
Other, non-Hispanic	57 (4.0)
Hispanic	23 (1.6)
Unknown/missing	214 (15.1)
Marital Status
Married	336 (23.8)
Not married	1,072 (75.8)
Unknown/missing	7 (0.5)

Note: This table includes only veterans in the cohort who died throughout the observation period (1999–2019).

^aAt time of death

The 25 most frequent underlying causes of death for TGD veterans are summarized in Table 2. Chronic obstructive pulmonary disease (7.4%) was the leading underlying cause of death, followed by malignant neoplasms of trachea, bronchus, and lung (7.3%), chronic ischemic heart disease (6.3%), diabetes mellitus (4.7%), and acute myocardial infarction (4.5%). Table 2 also ranks the 25 most frequent causes of death mentioned anywhere in the record axis codes. According to this categorization, the most common causes of death for TGD veterans were mental and behavioral disorders due to psychoactive substance use (17.2%), followed by cardiac dysrhythmias (15.3%), chronic obstructive pulmonary disease (15.1%), diabetes mellitus (13.9%) and chronic ischemic heart disease (13.3%). Of note, three distinct methods of suicide appeared as the 7^th (firearms), 17^th (self-poisoning), and 24^th (hanging) causes of death for TGD veterans. On the other hand, neither Parkinson’s disease nor Alzheimer’s disease were within the top 25 causes of death for TGD veterans.

Table 2.

Top 25 Causes of Death for TGD Veterans in the Veterans Health Administration, 1999–2019

Cause	Rank	n (% all deaths)
Underlying Causes
Other chronic obstructive pulmonary disease	1	104 (7.4)
Malignant neoplasms of trachea, bronchus, and lung	2	103 (7.3)
All other forms of chronic ischemic heart disease	3	89 (6.3)
Diabetes mellitus	4	67 (4.7)
Acute myocardial infarction	5	64 (4.5)
Atherosclerotic cardiovascular disease	6	61 (4.3)
Intentional self-harm (suicide) by discharge of firearms	7	48 (3.4)
Hypertensive heart disease	8	28 (2.0)
Malignant neoplasms of liver	9	26 (1.8)
Other malignant neoplasms of other and unspecified sites	10	26 (1.8)
Organic dementia	11	26 (1.8)
Conduction disorders and cardiac dysrhythmias	12	26 (1.8)
Accidental poisoning by and exposure to drugs and other biological substances	13	24 (1.7)
Septicemia	14	22 (1.6)
All other diseases of nervous system	15	21 (1.5)
Stroke, not specified as hemorrhage or infarction	16	21 (1.5)
Intentional self-poisoning (suicide) by and exposure to drugs and other biological substances	17	20 (1.5)
Viral hepatitis	18	18 (1.3)
Malignant neoplasms of pancreas	19	18 (1.3)
Pneumonia due to other or unspecified organisms	20	18 (1.3)
All other diseases of respiratory system	21	17 (1.2)
Mental and behavioral disorders due to use of alcohol	22	16 (1.1)
Renal failure	23	16 (1.1)
Intentional self-poisoning (suicide) by hanging, strangulation, and suffocation	24	16 (1.1)
Other ill-defined and unknown causes of mortality	25	15 (1.1)
All Causesa
Mental and behavioral disorders due to other psychoactive substance use	1	244 (17.2)
Conduction disorders and cardiac dysrhythmias	2	217 (15.3)
Other chronic obstructive pulmonary disease	3	213 (15.1)
Diabetes mellitus	4	196 (13.9)
All other forms of chronic ischemic heart disease	5	188 (13.3)
All other diseases of respiratory system	6	154 (10.9)
Essential (primary) hypertension	7	132 (9.3)
Malignant neoplasms of trachea, bronchus, and lung	8	113 (8.0)
Septicemia	9	107 (7.6)
Renal failure	10	105 (7.4)
Congestive heart failure	11	91 (6.4)
Acute myocardial infarction	12	86 (6.1)
Pneumonia due to other or unspecified organisms	13	78 (5.5)
Atherosclerotic cardiovascular disease	14	72 (5.1)
Hypertensive heart disease	15	66 (4.7)
Other malignant neoplasms of other and unspecified sites	16	63 (4.5)
All other diseases of nervous system	17	63 (4.5)
Intentional self-harm (suicide) by discharge of firearms	18	48 (3.4)
Viral hepatitis	19	44 (3.1)
Fibrosis and cirrhosis of liver	20	44 (3.1)
Organic dementia	21	43 (3.0)
Other metabolic disorders	22	40 (2.8)
Mental and behavioral disorders due to use of alcohol	23	39 (2.8)
Other diseases of liver	24	36 (2.5)
Stroke, not specified as hemorrhage or infarction	25	35 (2.5)

^aIncludes all conditions listed in the record axis codes. The record axis codes include all conditions listed in Part I of the death certificate, including the immediate and underlying cause of death and all conditions related to the causal chain of events that led to death, and all conditions listed in Part II of the death certificate, which refers to all other significant conditions contributing to death but not in the direct causal pathway.

Discussion

Building on earlier work analyzing mortality among TGD veterans,^14–15 this study describes the most frequent causes of death in a large sample of TGD veterans. Two prior studies used a rank-order approach of the top ten underlying causes of death for TGD veterans and found that heart disease and cancer were the first and second leading underlying causes of death, respectively.^14–15 Rather than collapsing all causes of death into ten categories, all ICD codes were combined into a total of 358 causes of death, allowing for a disease-specific exploration of mortality. For example, cancer of the trachea, bronchus, and lung was the most frequent underlying cause of cancer-related death. The present study also used a multiple cause of death approach by ranking the 25 most frequent causes of deaths that were mentioned anywhere on the death certificate.

When ranking diseases by any mention on the death certificate, diabetes emerged as the fourth-leading cause of death for TGD adults. One study using EHR data of 2,869 transfeminine adults matched with 28,3000 cisgender females and 28,258 cisgender males, and 2,133 transmasculine adults matched with 20,997 cisgender females and 20,964 cisgender males from three integrated health care systems in California and Georgia, found that transfeminine adults receiving gender-affirming hormone therapy had higher odds of diabetes compared to cisgender females.³⁰ TGD adults may have high risk of mortality due to diabetes due to experiences of discrimination in healthcare settings which cause TGD people to delay or avoid care.^7,31–35 Additionally, TGD adults experience discrimination in the workforce which leads to high rates of food insecurity.^36–37 Among patients with diabetes, food insecurity is associated with poor glycemic control and lower adherence to diabetes management recommendations.³⁸

The rankings of suicide deaths in this study corroborate evidence from several other studies which found higher rates of suicide death among TGD relative to cisgender veterans.^14–15,33 A prior study conducted by Blosnich et al. used a smaller VHA cohort to explore differences in methods of suicide death among TGD and cisgender veterans.³⁹ An appraisal of the results of the present study in conjunction with those of Blosnich et al. provide a more detailed perspective on suicide deaths among TGD veterans. Blosnich et al. found that for both TGD and cisgender veterans, firearms were the leading cause of suicide death, but when cisgender veterans die by suicide, nearly 7 in 10 (69%) deaths are attributable to firearms compared to about half (53%) of TGD suicide deaths.³⁹ However, the present study demonstrates that for TGD veterans, the frequency of suicide deaths using firearms is overwhelmingly higher relative to suicide deaths from any other cause. Given that firearms were the most frequent method of suicide death for TGD veterans, initiatives that promote firearm safety and lethal means counseling may hold promise,⁴⁰ though critical gaps in identifying effective suicide prevention measures specific for TGD populations remain. The findings around firearm suicides also point to veteran status as an important moderator of firearms suicide deaths among TGD populations because veterans tend to have more familiarity, training, and ownership of firearms compared to non-veterans.⁴¹ Suicide prevention for TGD adults must also address individual experiences of gender minority stress and felt stigma, strive to improve healthcare experiences, and drive protective laws and policies.⁴²

Prior research has reported higher substance use among transgender compared to cisgender populations,⁴³ but the direct implications for mortality remain unknown. This study found that mental and behavioral disorders due to other psychoactive substance use was the most common of all causes of death for TGD individuals, further highlighting the mental health and substance use burden within TGD populations. Despite a well-documented need for TGD-specific substance use treatments, such interventions are scarce.⁴⁴ Building culturally sensitive and affirming treatment paradigms into existing interventions and services may be a promising strategy to reduce substance-use related disparities in mortality.⁴⁵

With the exception of organic dementia, neurodegenerative diseases did not appear within the 25 most frequent causes of death for TGD adults. This finding is likely explained by competing risks for neurodegenerative diseases.⁴⁶ In other words, TGD people at risk of Alzheimer’s or Parkinson’s disease may die of other causes prior to its onset. Indeed, other research among TGD older adults,⁴⁷ including TGD veterans,⁴⁸ have reported an increased risk of Alzheimer’s disease compared to cisgender adults. Another study among TGD and cisgender adults demonstrated that TGD adults died at an age about 12 years younger than cisgender adults and may not have lived long enough to display symptoms of neurodegenerative disease.¹⁷

Limitations

The sample included veterans who were identified as TGD using ICD codes. While ICD codes provide a valid means of identifying TGD people from administrative datasets in the absence of gender identity self-identification, this method likely underestimates the total number of TGD veterans. Further, TGD veterans who did not have transgender-related ICD codes for reasons such as not seeking gender-affirming services in the VA or not wanting to disclose their gender identity were potentially misclassified as cisgender in this study.^2,22,24 Some ICD codes used to identify TGD people in this study may refer to sexuality and not gender identity (e.g., transvestic fetishism). The inclusion of the transvestic fetishism diagnosis is consistent with prior work using the same cohort,^14–15,39 including validity studies.^22,25 People who receive a diagnosis of transvestic fetishism also include those who identify as transgender and who may nor may not pursue social, legal, or medical gender affirmation.⁴⁹ Including transvestism captures more gender variance without exclusively relying on diagnostic criteria; nevertheless, researchers should be aware that this term is offensive and pathologizes gender diversity.^50,51

The sample limits generalizability because only veterans who sought care at the VHA were included in the study. The results may also be most generalizable to transfeminine individuals or people assigned male sex at birth. Although sex was not included as a variable in this analysis due to the unreliability of the EHR-based sex marker, the VHA population predominantly comprises people assigned male sex at birth.⁵² A prior study using this cohort found that HER-based sex and sex listed in the NDI matched for all cisgender veterans, but 4% of transgender veterans had records that did not match between these data sources, underscoring the need for improved sex and gender identity data collection in mortality surveillance.⁵³

One of the limitations of using death certificate data is their potential for error. Causes of death listed on the death certificate represent the opinion of the medical certifier of death (e.g., physician, nurse practitioner, coroner, medical examiner, or justice of the peace) based on an assessment of available laboratory and autopsy results and input from funeral directors and next-of-kin contacts.¹⁸ Death certificates may not include all medical conditions present at death, and disease misclassification or omission are potential sources of error.^54–56 These errors may be higher for TGD people as they may be less likely to have next-of-kin to provide crucial health-related information. Death certificate errors are also more likely if the death occurs suddenly or unobserved. For example, deaths that occur at a hospital or nursing home are less likely to have errors in the death certificate than those that occur at the emergency room or at home.⁵⁷ Because TGD people often avoid medical care due to stigma,^7;31–35 they may be less likely to die in a hospital or nursing home.

Conclusions

Mortality information is a bedrock of public health prevention. One of the best approaches to promoting longevity and wellness is to identify principal threats to both for prioritizing prevention efforts. While this kind of information has been available for the general population and for sub-populations (e.g., racial/ethnic minorities), for TGD populations this has previously not been possible. This study highlighted certain diseases, such as mental and behavioral health and metabolic disorders, that are key targets for prevention programs and interventions for TGD veterans. Notably, discrimination and stigma in healthcare settings is a preventable risk factor that underlies all disparities in mortality among TGD adults.^7;31–35 Addressing discrimination will be a crucial component of creating effective interventions, including efforts at the policy-level to reverse and prevent legislation that threatens equitable access to healthcare for TGD veterans.