Associations of Sexual Orientation-Related Minority and Military Stressors With Past-Year Cigarette Use Among Lesbian, Gay, and Bisexual Veterans

Abstract

Introduction

Lesbian, gay, and bisexual (LGB) individuals and Veterans are more likely to report current smoking than the general adult population in the United States. The Minority Stress Model may explain the high prevalence of cigarette smoking among LGB individuals, who experience unique interpersonal (e.g., discrimination) and intrapersonal (e.g., identity concealment) stressors related to their minoritized sexual orientation.

Aims and Methods

This study assessed whether three types of stressors (interpersonal, intrapersonal, and LGB-specific military) were associated with past-year smoking among LGB Veterans. Veterans were recruited online for a prospective cohort study. We conducted secondary data analysis of baseline surveys collected from 2019 to 2020. The study sample included cisgender, LGB Veterans (n = 463). Adjusted nested multivariable logistic regression models were used to estimate the association of each stressor with past-year cigarette smoking.

Results

Participants were mostly male (54.0%), non-Hispanic white (82.1%), and at least a college graduate (58.5%). LGB Veterans who were younger, had lower levels of education, income, and healthcare coverage, higher general stressors, and post-traumatic stress disorder and depression symptoms were more likely to smoke in the past year (n = 98, 21.2%). The adjusted odds of past-year smoking were higher among those who reported higher levels of harassment (aOR = 1.13, 95% confidence interval [CI]: 1.01 to 1.26), victimization (aOR = 1.13, 95% CI: 1.02 to 1.43), and family rejection (aOR = 1.13, 95% CI: 1.08 to 1.36).

Conclusions

Multiple interpersonal stressors were associated with past-year smoking, highlighting the need to intervene on these stressors. Future interventions should aim to address policies that reduce prejudice against LGB Veterans, while helping those who smoke identify and develop positive coping skills that support cessation.

Implications

Our findings contribute to the growing body of literature on tobacco use disparities among LGB individuals, particularly LGB Veterans. Results of this analysis provide some evidence for the Minority Stress Model as a conceptual model for understanding and intervening in disparities in smoking prevalence among LGB Veterans.

Introduction

Despite declines in the prevalence of cigarette smoking among adults in the United States (U.S.) from 20.9% in 2005 to 12.5% in 2020,¹ cigarette use disproportionately impacts marginalized groups. For example, those with lower education and income, in rural environments, with minimal to no healthcare coverage, minoritized racial and ethnic identities, minoritized sexual orientations, military service, and mental health conditions (anxiety and depression) have a higher prevalence of cigarette use.^1,2 Moreover, individuals with overlapping and intersecting minoritized identities may be at elevated risk for continued tobacco use and may ultimately bear a disproportionate burden of tobacco-related disease, disability, and death.

Research indicates that lesbian, gay, and bisexual (LGB) individuals are more likely to report current smoking compared to heterosexual individuals.^2–5 Similarly, Veterans are more likely to report ever or current smoking compared to non-Veterans.^6–9 Correlates for smoking among LGB and/or Veteran populations include younger age,^4,7,8,10,11 education,^2,5,7,10,11 income,^7,11 minoritized race and ethnicity,^7,11 lack of healthcare coverage,^7,11 relationship status,^5,7,11 depression,^5,9,10 post-traumatic stress disorder (PTSD),^9,12 and history of deployment.^2,12 Studies of LGB Veterans are less common; however, there is some evidence that LGB Veterans are at increased risk for smoking compared to their heterosexual Veteran peers.^13,14

Minority Stress

The Minority Stress Model, a conceptual framework developed to understand how stress influences health, is the prevailing model used to understand tobacco disparities among LGB individuals.^15–18 According to the model, individuals are exposed to unique stressors based on their minoritized identities (e.g., sexual orientation, gender identity, and/or race and ethnicity) that can cause additive stress above and beyond general life stressors (e.g., work or family stress). The model conceptualizes interpersonal and intrapersonal stressors as two sources of sexual orientation-related minority stress. Interpersonal stressors are external events or experiences of prejudice by others towards LGB individuals, such as discrimination and harassment. Intrapersonal stressors are internal processes, such as LGB identity concealment, which often stem from interpersonal stressors. For example, if harassed, an individual may conceal their LGB identity to prevent or protect from further incidents of prejudice. The model suggests that all minoritized individuals are exposed to minority stressors, yet the adverse impact of stress on health varies by differences in duration, frequency, and intensity of stressors.^15,16 Furthermore, an individual’s perception of events or experiences as stressors, internalization of the stressors, and access to stress-reducing resources such as positive coping can also lead to differential impacts on health.^15,16

The relationship between stress and smoking has been well documented, with stress promoting smoking initiation, maintenance, and relapse.¹⁹ Additionally, stress mediates the relationship between sexual orientation and current smoking when comparing LGB individuals to heterosexual individuals.²⁰ For these reasons, the Minority Stress Model provides a compelling framework for understanding the high prevalence of cigarette smoking among LGB individuals. However, the evidence of the impact of minority stress on cigarette smoking is limited and mixed. Some findings suggest that LGB individuals who experience higher levels of sexual orientation-related discrimination are more likely to report past-year cigarette smoking^21–23 while other studies found no associations.^10,24 Prior research also suggests that family rejection because of sexual orientation is associated with smoking.^10,25 Few studies have examined intrapersonal stressors, and among them, findings have been inconsistent with only one study finding higher levels of internalized homophobia to be associated with smoking.¹⁰

In addition to being exposed to these minority stressors in everyday life, LGB Veterans may have been exposed to minority stressors while in the military. Historically, the U.S. military has excluded LGB individuals from serving openly and only since the repeal of the “Don’t Ask, Don’t Tell” (DADT) Policy in 2011 have LGB individuals been able to serve without having to hide their sexual orientation.^26,27 Under exclusionary policies, LGB service members may have experienced intrapersonal stress (e.g., anxiety about others finding out about their sexual orientation) and interpersonal stress (e.g., being physically isolated from unit because of their perceived or known sexual orientation). These minority stressors may be especially magnified in the military environment, where LGB service members may have to rely on their unit for their physical safety. While policies have changed and the number of service members who are serving openly has increased, military environment cultural norms and anti-LGB attitudes from individual service members may continue to perpetuate minority stressors.²⁸ No research to date has examined the impact of minority stressors and minority stressors while in the military in relation to smoking among LGB Veterans.

The goal of this study was to test whether three types of sexual orientation-related stressors were associated with past-year smoking among a cohort of LGB Veterans. These stressors include interpersonal minority stressors (i.e., discrimination, harassment, victimization, trauma, social exclusion, and family rejection), intrapersonal minority stressors (i.e., identity concealment and internalized homophobia), and LGB-specific military stressors (e.g., interpersonal, and intrapersonal stressors while serving in the military). We hypothesized that LGB Veterans exposed to higher levels of interpersonal stressors, intrapersonal stressors, and/or LGB-specific military stressors would be more likely to smoke in the past year.

Methods

Data Source

The Health for Every Veteran Study uses a prospective cohort design to understand mental health problems and health risk behaviors among heterosexual and lesbian, gay, bisexual, and transgender (LGBT) Veterans. Details on recruitment and eligibility criteria have been previously reported.²⁹ Eligible participants were at least 18 years old; reported having served in the U.S. Armed Forces; were currently living in the U.S.; and had routine access to the Internet. Participants completed four web-based surveys every 9 months for 27 months. We conducted secondary data analysis of baseline surveys collected from September 2019 to December 2020. Those who completed the survey received $30, which took approximately 60 to 90 minutes to complete.

Analytic Sample

The analytic sample for the current study included cisgender (i.e., current gender identity the same as sex assigned at birth) Veterans who identified as lesbian, gay, or bisexual. We excluded cisgender, heterosexual Veterans (n = 343) because our independent variables of interest focused on minoritized sexual orientation-based (i.e., LGB) stressors. Transgender Veterans (n = 189) and gender diverse (e.g., genderqueer, gender fluid, or non-binary) Veterans (n = 66) did not receive the same measures as cisgender LGB participants and were excluded from the current study. We acknowledge that we may have excluded transgender and gender-diverse individuals who identify as LGB in our analyses. One cisgender, LGB Veteran was excluded because they had no responses for our independent variables of interest.

Measures

Dependent Variable

Past-Year Cigarette Smoking Status

Participants who reported having smoked more than 100 cigarettes in their life and having smoked a cigarette (even one or two puffs) in the past year were categorized as LGB Veterans with “past-year smoking,” and all others were categorized as those with “no past-year smoking.”³⁰

Independent Variables: Interpersonal Minority Stressors

Discrimination

The 10-item Everyday Discrimination Scale measures the frequency of discriminatory experiences in the past year, ranging from (1) never to (6) almost every day, and the main reason for the experience (e.g., race, age, and sexual orientation).^31,32 The scale was developed to assess unfair treatment in racially and ethnically diverse samples and demonstrated internal consistency (α = 0.88)³¹ and test–retest reliability.³² Example items of discrimination include, “You were treated with less courtesy than other people.” To assess sexual orientation-related discrimination in the past year, we created a composite variable for each item indicating whether the participant had experienced the discriminatory experience (1) at least once in the past year because of sexual orientation, and (0) never or for some other reason. A total sum score (range 0–10) was calculated for total number of occurrences, with higher scores indicating greater sexual orientation-related discriminatory events or experiences in the past year. Cronbach’s alpha for the current sample was α = 0.83.

Harassment, Victimization, Trauma, Social Exclusion, and Family Rejection

The Daily Heterosexist Experiences Questionnaire is a 50-item measure of sexual orientation-related experiences of distress.³³ The questionnaire was developed and validated among diverse samples of LGBT adults and demonstrated having internal consistency (α = 0.92), concurrent validity, and construct validity.³³ Internal consistency for subscales used in the current study are presented below. We used a 6-item subscale to assess harassment (e.g., “being verbally harassed by strangers for being LGBT”; α = 0.88), a 4-item subscale to assess victimization (e.g., “being punched, hit, kicked, or beaten because you are LGBT”; α = 0.86), a 6-item subscale to assess trauma (e.g., “hearing about LGBT people you know being treated unfairly”; α = 0.85), a 4-item subscale to assess social exclusion (e.g., “having very few people you can talk to about being LGBT”; α = 0.82), and a 6-item subscale to assess family rejection (e.g., “your family avoiding talking about your LGBT identity”; α = 0.87). Participants were asked about the occurrence and how much these events and experiences distressed or bothered them in the past year. Response options ranged from (0) did not happen, or not applicable to (5) it happened, and it bothered me extremely. We followed recommended scoring guidelines,³³ which dichotomized occurrences to (0) did not happen, or not applicable and (1) yes, it happened. We calculated total sum scores by summing all items within a subscale to indicate how many of these events or experiences occurred in the past year, with higher scores indicating a greater number of occurrences.

Independent Variables: Intrapersonal Minority Stressors

Identity Concealment

The 6-item Sexual Orientation Concealment Scale is a measure of behaviors that LGB individuals sometimes use to hide their sexual orientation.³⁴ The scale was developed and validated among LGB university students in the U.S. and has shown internal consistency (α = 0.79) and significantly correlated with other measures of identity concealment.³⁴ Example items include, “In the last year, I have concealed my sexual orientation by telling someone that I was straight or denying that I was LGB.” Participants indicated the frequency of these behaviors on a 5-point Likert scale ranging from (1) not at all to (5) all the time. We calculated a mean score by summing and averaging all items (range 1–5), with higher scores indicating higher identity concealment. Internal consistency reliability in the current sample was α = 0.86.

Internalized Homophobia

The 10-item Internalized Homophobia Scale is a measure of thoughts and feelings regarding being LGB.¹⁷ The scale was originally developed to assess rejection of sexual orientation among gay men but has since been validated among LGB individuals, and has demonstrated internal consistency (range α = 0.79–0.83) and convergent validity in prior studies.³⁵ Example items of internalized homophobia include, “If someone offered you the chance to be completely heterosexual, you would have accepted the offer.” Participants indicated the frequency of these thoughts and feelings in the past year, on a 4-point Likert scale ranging from (1) never to (4) often. Responses were summed and averaged to create a total mean score (range 1–4), with higher mean scores indicating higher internalized homophobia. Cronbach’s alpha for the current sample was α = 0.86.

Independent Variables: LGB-specific Military Stressors

Interpersonal LGB Military Stressors

The 8-item Military Events and Sexual Orientation Scale was developed to assess whether any stressful events or experiences were initiated by the U.S. military to investigate or punish LGB Veterans during military service.³⁶ Example items include, “Were you ever incarcerated in a military jail or prison due to your sexual orientation?” or “Were you ‘outed’ to friends and family in the course of any investigation of your sexual orientation?.” Participants were asked to indicate whether these interpersonal stressors occurred; (1) Yes or (0) No. Based on recommended scoring guidelines, we dichotomized responses to (1) Yes, if participants answered “Yes” to any of the items or (0) No, if they answered “No” to all items.³⁶ Prior studies have also used this approach because items were not correlated with one another as in a continuous scale.³⁷

Intrapersonal LGB Military Stressors

The Military Experiences and Sexual Orientation Scale was developed to assess LGBT U.S. Veterans’ beliefs, attitudes, and experiences related to sexual orientation while in the military and has shown internal consistency (α = 0.73).³⁸ We used an 8-item subscale, which measures anxiety about sexual orientation identity and identity concealment.³⁸ Participants were asked to rate their agreement or disagreement with statements about intrapersonal stressors a LGB individual could have experienced while in the military. Reponses options ranged from (1) strongly disagree to (5) strongly agree. Example items include, “If I had let people in the service know of my sexual orientation, I probably would have been harmed physically” and “In the service, I was constantly trying to conceal my sexual orientation.” A total sum score was calculated by summing all responses (range 8–40), with higher sum scores indicating higher anxiety about concealing sexual orientation during military service. Internal consistency reliability in the current sample was α = 0.78.

Covariates

Sociodemographic

Sociodemographic characteristics included age in years, gender identity (man or woman), sexual orientation (gay/lesbian or bisexual), race and ethnicity (non-Hispanic white, non-Hispanic other racial identities, or Hispanic), relationship status (in a relationship—i.e., married, domestic partnership, or dating; or single), educational attainment (some college/high school or less, or college graduate or more), annual household income ($39 999 or less, $40 000 to $79 999, or $80 000 or more), period of service (before DADT or before DADT repeal, before and after DADT repeal, or after DADT repeal), military branch (Air Force, Army, Coast Guard, Marines, Navy, and National Guard/Reserves), and any kind of healthcare coverage (yes or no).

Mental Health Conditions

PTSD symptoms were assessed using the Diagnostic and Statistical Manual of Mental Disorders – Fifth Edition (DSM-5) PTSD Checklist (PCL).^39,40 The 20-item scale asked participants about problems they may have had after a stressful experience or event that occurred at any point in their life and how much each of the problems bothered them in the past month. Response options ranged from (0) not at all to (4) extremely. A total symptom score was calculated by summing the items (range 0–80), with a cutoff score of ≥33 indicating probable symptoms of PTSD.³⁹ Depression was assessed using the Patient Health Questionnaire (PHQ-8), an 8-item scale that measures the frequency of depression symptoms in the past 2 weeks.^41,42 Participants were asked how often they were bothered by the symptoms, with response options ranging from (0) not at all to (3) nearly every day. A total score was calculated by summing the items, with a cutoff score of ≥10 indicating probable depression symptoms.

General Stress

The 16-item Social Stress Scale was used to measure instances of social and personal life experiences and events in the past year that could have been stressors.⁴³ Participants were asked whether these experiences and events occurred in the past year; (1) Yes or (0) No. A total score was calculated by summing all “Yes” responses (range 0–16), with higher scores indicating a greater degree of general stressors.

Data Analysis

We calculated descriptive statistics for all variables. Bivariate analyses were conducted to assess the associations between participant characteristics, stressors, and cigarette smoking using chi-square tests of independence for categorical variables and unequal variance t-tests for continuous variables. Multivariable logistic regression models with 95% confidence interval (CI) were used to estimate the association of stressors with cigarette smoking. Nested models estimated associations of cigarette smoking with (1) each stressor as separate independent variables modeled individually, (2) stressors grouped together by type (e.g., all interpersonal stressors) as separate independent variables modeled together, and (3) all stressors as separate variables together in a final model. Covariates for adjusted analyses included age, educational attainment, income, healthcare coverage, PTSD, depression, and general stressors. Covariates were identified from existing literature indicating associations with smoking among LGB and/or Veteran populations^{2,4,5,7,8,10-12} and were selected if significant in our bivariate analyses. In sensitivity analyses, models were minimally adjusted for age, educational attainment, and income to assess whether potential over-adjustment biased estimates toward the null.⁴⁴ In additional sensitivity analyses, models were not adjusted for PTSD symptoms and depression symptoms given their potential to mediate the association between stress and past-year smoking status.^45,46 Lastly, in sensitivity analyses, we used a Bonferroni correction to adjust the significant level for multiple comparisons. In post hoc exploratory analyses, using covariate adjustment, we examined potential gender differences (i.e., lesbian and bisexual women versus gay and bisexual men) in the association of stressors and smoking. All analyses were conducted using Stata/MP Version 17.0 (StataCorp, College Station, TX). This research was approved by the Human Subjects Division of the University of Washington.

Results

Participant Characteristics

Table 1 presents participant characteristics for the overall sample and past-year cigarette smoking status (N = 463). The mean age was 49.53 years (standard deviation [SD] = 14.71), the majority were male (54.0%), non-Hispanic white (82.1%), in a relationship (66.9%), at least a college graduate (58.5%). The majority served in the military before “DADT” or before “DADT” was repealed (65.7%) and the largest proportion of participants served in the Army (42.9%) during their military tenure. In terms of cigarette smoking, 80.8% reported ever smoking and, of these, 69.0% reported lifetime smoking (having smoked 100 cigarettes or more). In the sample, 21.2% were categorized as LGB Veterans with past-year smoking (n = 98).

Table 1.

Participant Characteristics for Overall Sample and by Past-Year Cigarette Smoking Status

	Overall sample, N = 463	Past-year smoking, n = 98	No past-year smoking, n = 365	p-value
	Mean (standard deviation, SD)
Age (years)	49.53 (14.71)	43.49 (11.93)	51.13 (14.97)	<.001**
General stressors (range 0–16)	2.73 (2.29)	3.64 (2.59)	2.48 (2.13)	<.001**
	Frequency (%)
Gender identity				.506
Man	250 (54.0)	50 (51.0)	200 (54.8)
Woman	213 (46.0)	48 (49.0)	165 (45.2)
Sexual orientation				.073
Gay/lesbian	346 (74.7)	66 (67.3)	281 (77.0)
Bisexual	117 (25.3)	33 (33.7)	84 (23.0)
Race and ethnicity				.246
Non-Hispanic white	380 (82.1)	78 (79.6)	302 (82.7)
Non-Hispanic other identitya	46 (9.9)	10 (10.2)	36 (9.9)
Hispanic	36 (7.8)	9 (9.2)	27 (7.4)
Relationship status				.928
In a relationshipb	310 (66.9)	65 (67.0)	245 (67.1)
Single	150 (32.4)	32 (33.0)	118 (32.3)
Educational attainment				<.001++
Some college/high school or less	192 (41.5)	56 (57.1)	136 (37.3)
College graduate or more	271 (58.5)	42 (42.9)	229 (62.7)
Income (annual household)				<.001++
$39 999 or less	128 (27.7)	40 (40.8)	88 (24.1)
$40 000 to $79 999	167 (36.1)	39 (39.8)	128 (35.1)
$80 000 or more	167 (36.1)	19 (19.4)	148 (40.6)
Period of servicec
Only before “DADT”/before repeal	304 (65.7)	57 (58.2)	247 (67.7)	.211
Before and after “DADT” repeal	84 (18.1)	22 (22.4)	62 (17.0)
Only after “DADT” repeal	75 (16.2)	19 (19.4)	56 (15.3)
Branch
Air force	101 (21.8)	20 (20.4)	81 (22.2)	.704
Army	199 (42.9)	42 (42.9)	157 (43.0)	.978
Coast guard	12 (2.6)	2 (2.0)	10 (2.7)	.699
Marines	38 (8.2)	8 (8.2)	30 (8.2)	.986
Navy	119 (25.7)	25 (25.5)	94 (25.8)	.961
National guard/reserves	70 (15.1)	15 (15.3)	55 (15.1)	.954
Healthcare coverage – yes	401 (86.6)	77 (78.6)	324 (88.8)	.015+
PTSD symptomsd	182 (39.3)	49 (50.0)	133 (36.4)	.007++
Depression symptomse	182 (39.3)	49 (50.0)	133 (36.4)	.013+
Ever smoking – yes	374 (80.8)
Lifetime smokingf – yes	258 (69.0)

Frequencies and proportions may not equal 100% because of rounding and/or missing data.

^aIncludes non-Hispanic black, multiracial, and “other” racial identities. Race and ethnic groups were combined because of small cell sizes.

^bIncludes being married, in a domestic partnership, or dating.

^c“Don’t Ask, Don’t Tell” (DADT) policy, in effect from February 1994 until September 2011, required service members to conceal their sexual orientation.

^dPost-traumatic stress disorder (PTSD) checklist for DSM-5 (PCL-5) cutoff score of ≥33 indicating probable PTSD symptoms.

^ePatient Health Questionnaire (PHQ-8) cutoff score of ≥10 indicating probable depression symptoms.

^fSmoked 100 or more cigarettes (five packs or more) in entire life.

**Welch’s t-test significant at p-value <.01; ⁺χ² test of independence significant at p-value <.05; ⁺⁺χ² test of independence significant at p-value <.01.

LGB Veterans who reported past-year smoking were younger in age (p-value <.001), had lower educational attainment levels (p-value = .001), lower annual household income (p-value <.001), less healthcare coverage (p-value = .015), higher mean general stressor scores (p-value <.001), and had higher levels of PSTD symptoms (p-value = .007) and depression symptoms (p-value = .013) compared to those with no past-year smoking (see Table 1).

Bivariate Differences by Past-Year Smoking Status

Table 2 presents descriptive statistics (means and SD) for minority and LGB-specific military stressors. In bivariate analyses, LGB Veterans with past-year smoking reported higher past-year harassment (M = 3.20 [SD] = 2.29 vs. M = 2.46 [SD] = 2.31; p-value = .006), victimization (M = 0.99 [SD] = 1.59 vs. M = 0.55 [SD] = 1.28; p-value = .012), social exclusion (M = 2.69 [SD] = 1.36 vs. M = 2.29 [SD] = 1.51; p-value = .012), and family rejection (M = 3.20 [SD] = 2.27 vs. M = 2.25 [SD] = 2.19; p-value < .001) compared to those with no past-year smoking. There were no significant findings for intrapersonal minority stressors nor LGB-specific military stressors (see Table 2).

Table 2.

Minority and LGB-Specific Military Stressors for Overall Sample and by Past-Year Smoking Status

	Overall sample, N = 463	Past-year smoking, n = 98	No past-year smoking, n = 365	p-value
Interpersonal minority stressors	Mean (SD)
Discrimination (range 0–10)	2.14 (2.24)	2.55 (2.61)	2.04 (2.37)	.079
Harassment (range 0–6)	2.62 (2.32)	3.20 (2.29)	2.46 (2.31)	.006**
Victimization (range 0–4)	0.64 (1.36)	0.99 (1.59)	0.55 (1.28)	.012*
Trauma (range 0–6)	5.01 (1.55)	5.21 (1.41)	4.96 (1.58)	.135
Social exclusion (range 0–4)	2.38 (1.49)	2.69 (1.36)	2.29 (1.51)	.012*
Family rejection (range 0–6)	2.45 (2.24)	3.20 (2.27)	2.25 (2.19)	<.001**
Intrapersonal minority stressors
Identity concealment (range 1–5)	1.78 (0.85)	1.93 (0.87)	1.74 (0.84)	.053
Internalized homophobia (range 1–4)	1.56 (0.51)	1.61 (0.56)	1.54 (0.49)	.264
LGB-specific military stressors
LGB military experiences (range 8–40)	26.32 (6.83)	25.95 (7.46)	26.42 (6.67)	.572
LGB military events – yes (N, %)	159 (34.3%)	30 (30.6%)	129 (35.3%)	.391

*Welch’s t-test comparing mean scores between past-year smoking group and no past-year smoking group was found to be significant at p-value <.05.

**Welch’s t-test comparing mean scores between past-year smoking group and no past-year smoking group was found to be significant at p-value <.01.

Multivariable Analyses of Past-Year Smoking Status

Results of the multivariable logistic regression models, where each stressor was modeled individually, are presented in Table 3. The adjusted odds of past-year smoking were higher among individuals who reported higher past-year harassment (aOR = 1.13 [95% CI: 1.01 to 1.26], p-value = .032), victimization (aOR = 1.21 [95% CI: 1.02 to 1.43], p-value = .024), and family rejection (aOR = 1.21 [95% CI: 1.08 to 1.36], p-value < .001).

Table 3.

Adjusted Logistic Regression Models Estimating Associations Between Minority Stressors and LGB-Specific Military Stressors With Past-Year Smoking Status

	β	aOR	SE	95% CI	p-value
Interpersonal minority stressors
Discrimination	0.057	1.06	0.056	(0.95 to 1.17)	.285
Harassment	0.119	1.13	0.063	(1.01 to 1.26)	.032*
Victimization	0.189	1.21	0.102	(1.02 to 1.43)	.024*
Trauma	0.071	1.07	0.093	(0.91 to 1.27)	.411
Social exclusion	0.097	1.10	0.100	(0.92 to 1.32)	.285
Family rejection	0.191	1.21	0.070	(1.08 to 1.36)	<.001**
Intrapersonal minority stressors
Identity concealment	0.058	1.06	0.153	(0.79 to 1.41)	.687
Internalized homophobia	−0.058	0.94	0.246	(0.57 to 1.57)	.824
LGB-specific military stressors
LGB military experiences	−0.001	0.99	0.020	(0.96 to 1.04)	.965
LGB military events	−0.210	0.81	0.220	(0.47 to 1.38)	.440

aOR = adjusted odds ratio, SE = standard error, CI = confidence interval.

Models adjusted for age, educational attainment, income, healthcare coverage, general stress, post-traumatic stress disorder symptoms, and depression symptoms.

*p-value <.05.

**p-value <.01.

Only the association between family rejection and past-year smoking remained significant (aOR = 1.21 [95% CI: 1.03 to 1.42], p-value = 0.017) when interpersonal stressors were modeled together as separate independent variables (see Supplemental Table 1). Similarly, in the fully adjusted model, where all stressors were modeled together as separate independent variables, the estimated odds of past-year smoking were higher among those with higher family rejection scores (aOR = 1.23 [95% CI: 1.03 to 1.45], p-value = .019) (see Supplemental Table 1).

Sensitivity Analysis

Results of minimally adjusted logistic regression models, only adjusted for age, educational attainment, and income, show consistent results with our main analysis (see Supplemental Table 2). In logistic regression models not adjusted for mental health conditions, including probable PTSD and major depression symptoms, the estimated odds of being a LGB Veteran who reported past-year smoking was higher among those with higher mean victimization scores and family rejection scores (see Supplemental Table 3). However, the association between mean harassment scores and past-year smoking status did not remain significant. When the significance level was adjusted using a Bonferroni correction for multiple comparisons, our results remained the same (results shown). Lastly, in post hoc exploratory analyses we did not find any differences in higher versus lower levels of stressors being associated with increased odds of past-year smoking by gender identity.

Discussion

This study was guided by the Minority Stress Model,^15–18 which suggests that LGB individuals are exposed to unique stressors based on their minoritized sexual orientation which, in turn adversely impact health. To the best of our knowledge, this is the first study to comprehensively examine whether sexual orientation-related interpersonal minority stressors, intrapersonal minority stressors, and military stressors are associated with past-year cigarette smoking among LGB Veterans.

We hypothesized that LGB Veterans exposed to higher levels of sexual orientation-related stressors would be more likely to smoke in the past year. We found that higher levels of several interpersonal stressors, including harassment, victimization, and family rejection were associated with past-year cigarette smoking. Family rejection was the most robust predictor of past-year cigarette use. Prior studies among LGB youth and adolescents have found that family rejection because of sexual orientation is associated with smoking.^10,25 Our results extend these findings to adult LGB Veterans, highlighting the potential influence of family rejection on tobacco use well into adulthood. Future studies should further examine this relationship among adults. Further research is also needed to better understand how distress and its impact on smoking varies by stressor type. The Minority Stress Model suggests that all minoritized individuals are exposed to minority stressors; however, differences in duration, frequency, and intensity of stressors can have varied impacts on health and adverse health behaviors.^15,16 Compared to other interpersonal stressors, family rejection is a proximal source of prejudice since it is committed by known people, and may persist for a longer duration of time (e.g., over months or years), may reoccur or be frequent, and may be perceived as more intense than acts of prejudice committed by strangers.

In our study, we found no significant associations between past-year sexual orientation-related discrimination with past-year smoking status. This is in contrast to studies among LGB individuals found that those who experienced higher levels of past-year sexual orientation-related discrimination had significantly greater odds of past-year cigarette use, tobacco use disorder, and dual or poly-tobacco use.^21–23,47 One potential explanation for this conflicting finding may be because of measurement differences. Prior studies that found significant associations between discrimination and tobacco outcomes used adapted scales that specifically assess sexual orientation-related discrimination.^22,23 Our measure of past-year sexual orientation-related discrimination was derived from the Everyday Discrimination Scale,^31,32 which was originally developed to assess racial discrimination and measures both frequency and attribution of discriminatory events or experiences based on several aspects of identity. We created an attribution variable to examine past-year discrimination because of sexual orientation compared to no discrimination or discrimination for some other attribution. A study that used this same approach to measure sexual orientation-based discrimination among lesbian and bisexual women, also did not find an association between discrimination and current smoking.²⁴

Our study did not find significant associations between intrapersonal minority stressors (i.e., identity concealment and internalized homophobia) and past-year smoking. These results may be partially explained by selection bias.⁴⁸ Participants for the Health for Every Veteran Study were recruited through various approaches including online ads to LGBT Veteran groups and LGBT groups, and eligibility criteria included willingness to answer key demographic questions including sexual orientation questions. Participants who self-selected into the study may have higher LGB identity centrality, which refers to how important an individual’s sexual orientation is to their sense of self. Research has found that those with higher levels of LGB identity centrality have higher levels of self-acceptance and lower levels of identity concealment and stigma.⁴⁹ Lastly, we found no associations between LGB-specific military stressors and past-year smoking status, which may be limited by temporality issues. Measures of LGB-specific military stressors assessed interpersonal and intrapersonal minority stressors while participants were in the military, the lapse of time since separation from service may have diminished effects on past-year smoking behavior. Additionally, the lapse of time from service may have resulted in underreporting of LGB-specific military stressors and/or may be subject to recall bias. Additional research is needed to better understand how minority stressors while in the military impact cigarette smoking among LGB Veterans.

While our study focused on sexual orientation-related stressors, we observed several differences in smoking status. Our findings suggest that age, educational attainment, income, lack of healthcare coverage, higher levels of PTSD and depression symptoms, and general stressors are important correlates of tobacco use among LGB Veterans. These correlates are similar to those of non-Veteran LGB individuals.^4,5,10 However, future research among Veterans should consider how these characteristics interact with LGB identity to impact cigarette smoking. For example, prior research indicates that health outcomes among gay and bisexual men are moderated by socioeconomic status.⁵⁰ Our analyses adjusted for educational attainment and annual household income, but we did not explore the potential moderating effects of socioeconomic status on past-year cigarette smoking status. Previous research has also found that mental health conditions, including PTSD and depression, mediate the associations between stress and smoking.^45,46 In sensitivity analysis, we found high levels of PTSD and depression symptoms to be associated with past-year smoking status, particularly for past-year experiences of harassment. Identifying moderators and mediators may help identify important targets for smoking cessation treatment.

There are limitations to this study. First, we used cross-sectional data for this analysis, and therefore we cannot make any causal interpretations. Second, due to the sample size we were unable to conduct stratified analysis by gender identity and/or sexual orientation identity, a consistent gap in the literature. Future research with larger samples of LGB Veterans who smoke are needed to better understand these potential differences. Third, study inclusion criteria included having routine access to the internet, an e-mail address, a postal address, and a phone number which may have excluded Veterans who are most vulnerable and likely to smoke cigarettes. Finally, we operationalized our outcome as past-year smoking versus no past-year smoking to be time-concordant with most of our independent variables, which assessed exposure to sexual orientation-related stressors in the past year. This approach is common in studies using large national epidemiological survey data.^22,23 However, we acknowledge that the minimal frequency and/or heaviness of smoking (i.e., having smoked a cigarette, even one or two puffs in the past year) could lead to misclassification. A relatively small proportion of our sample (2.8%, n = 13) indicated smoking more than 100 cigarettes in their lifetime but smoking ≤1 day or ≤1 cigarette on the days they smoked within the past 30 days.

Despite these limitations, this study has many strengths. It is the first study to examine constructs within the Minority Stress Model, specifically the associations of sexual orientation-related interpersonal stressors, intrapersonal stressors, and military stressors with past-year cigarette smoking among LGB Veterans. The Minority Stress Model is the guiding framework used to understand the high prevalence of cigarette smoking among LGB individuals, yet evidence of its utility has lagged. With the exception of the NESARC-III, evidence has largely lagged because nationally representative data sources often lack measures of sexual orientation and/or measures specific to LGB experiences, including minority stress.⁵¹ Even so, the NESARC-III has its limitations. First, data were collected between 2012 and 2013, and the sociopolitical landscape for LGB individuals has changed including the increase in anti-LGBT legislation in the U.S.⁵² Second, NESARC-III only measures sexual orientation-related discrimination, leaving continuous gaps in our understanding of the relationship between other sexual orientation-related minority stressors and smoking. In contrast, our study measured six interpersonal minority stressors, two intrapersonal minority stressors, and two LGB-specific military stressors. Our findings provide some evidence for the Minority Stress Model, particularly with regard to interpersonal stressors. However, future studies that comprehensively examine constructs within the Minority Stress Model using nationally representative samples are needed.

Conclusion

Our findings show that LGB Veterans who reported experiencing more occurrences of harassment, victimization, and family rejection in the past year were more likely to report smoking cigarettes in the past year. These results suggest the need for interventions that address interpersonal minority stressors. At the state or national level, enactment of anti-discrimination laws to protect LGBT individuals against prejudice may be essential for reducing health disparities. While smoking cessation interventions cannot prevent events or experiences of prejudice towards LGB individuals because of their sexual orientation, at the individual level interventions can help LGB people who smoke identify these stressors and help develop positive coping skills that support smoking cessation. This may be particularly important for LGB individuals who experience family rejection because of their sexual orientation.

Supplementary Material

Supplementary material is available at Nicotine and Tobacco Research online.

Funding

This work was supported by the Health Services Research & Development (HSR&D) grant (I01HX002423 to K.L.) from the U.S. Department of Veterans Affairs. R.A.R acknowledges support from the Biobehavioral Cancer Control and Prevention Training grant (T32CA092408) from the National Cancer Institute at the National Institutes of Health.

Author Contributions

Raymond Ruiz (Conceptualization [Lead], Formal analysis [Lead], Project administration [Lead], Writing – original draft [Lead], Writing – review & editing [Lead]), Keren Lehavot (Conceptualization [Equal], Data curation [Lead], Formal analysis [Equal], Funding acquisition [Lead], Methodology [Equal], Writing – original draft [Equal], Writing – review & editing [Equal]), Jaimee Heffner (Conceptualization [Equal], Formal analysis [Equal], Methodology [Equal], Writing – original draft [Equal], Writing – review & editing [Equal]), Christine Kava (Conceptualization [Equal], Formal analysis [Equal], Methodology [Equal], Writing – original draft [Equal], Writing – review & editing [Equal]), and India Ornelas (Conceptualization [Equal], Formal analysis [Equal], Methodology [Equal], Supervision [Lead], Writing – original draft [Equal], Writing – review & editing [Equal]).

Declaration of Interests

This content is solely the responsibility of the authors and does not necessarily represent the official views of the U.S. Department of Veterans Affairs or the National Cancer Institute. The authors declare no potential conflicts of interest.

Data Availability

Data not publicly available.