Sexual identity differences in biomarkers of tobacco exposure among women in a national sample

Abstract

Background:

Sexual minority women are consistently at increased risk for tobacco use compared to heterosexual women. Neither biomarkers of nicotine exposure nor biomarkers of tobacco toxicant exposure have been examined by sexual identity.

Methods:

This study used interview and biomarker data from women in the biomarker core sample of Wave 1 of the Population Assessment of Tobacco and Health (PATH) study (2013–2014; n=4,930). We examined associations of sexual identity with nicotine exposure (measured with urinary cotinine and TNE-2) and with tobacco-specific nitrosamines (measured with urinary NNAL). Multivariable regression modeling was used to examine these associations among the full biomarker core sample, among past 30-day tobacco users, and among exclusive established cigarette users before and after controlling for tobacco use quantity and intensity.

Results:

In the full biomarker sample of women, prior to adjusting for tobacco use quantity and intensity, bisexual women had significantly higher cotinine, TNE-2, and NNAL levels compared to heterosexual women. Among exclusive established cigarette users, gay/lesbian women had significantly higher NNAL compared to heterosexual women prior to adjusting for tobacco quantity and intensity. No differences by sexual identity were found after adjusting for tobacco use quantity and intensity.

Conclusions:

This is the first study to demonstrate differences in biological markers of tobacco exposure by sexual identity among women in the U.S. This has important public health implications as greater exposure to both nicotine and to tobacco-specific nitrosamines are strongly linked to cancer risk.

INTRODUCTION

Tobacco use remains the leading cause of preventable death in the U.S., accounting for 480,000 deaths including cancer, cardiovascular diseases, diabetes, and pulmonary illnesses.¹ The health burden of tobacco use is largely driven by the complex mixture of over 7,000 chemical compounds found in tobacco smoke, including multiple toxicants and carcinogens.² Examining population-level exposure to toxicants and carcinogens from tobacco, and understanding whether vulnerable populations differ in their exposure, can provide a clearer picture of the distribution of the public health burden of tobacco use.^3–4 This is particularly important to understand among women, who experience unique challenges related to tobacco use. Women are disproportionately affected by the health consequences of smoking,⁵ have a history of being targeted in tobacco marketing,^6–8 and often face greater smoking-related stigma.^9–10

The most potent carcinogens associated with tobacco use include tobacco-specific nitrosamines (TSNAs).¹ TSNAs are found in tobacco and tobacco smoke^11–12 and are highly specific to tobacco exposure, making these compounds excellent indicators of health risks stemming from tobacco use.^3,13–14 Nicotine derived nitrosamine ketone 4-(metylnitrosamino)-1-(3-pyridyl)-1-butanon, or NNK, is the most understood and commonly studied TSNA in epidemiologic research. The primary metabolite of NNK, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol, or NNAL, provides utility as a biomarker of tobacco exposure as well as a direct marker of cancer risk.^15–16 Nicotine exposure is also critical to examine, as it is the addictive component in tobacco that contributes to tobacco use disorder. While nicotine itself is not considered carcinogenic, nicotine exposure is dose-dependently associated with tobacco use disorder severity,^14,16 the quantity of tobacco used, and lung cancer risk.¹⁴ Further, evidence suggests nicotine may facilitate endogenous formation of TSNAs, induce DNA damage, and serve as a tumor promoter, underscoring the importance of examining nicotine exposure as it relates to cancer prevention and care.¹⁸

Extant research has demonstrated a significantly greater prevalence of tobacco use among sexual minorities compared to heterosexual individuals.^19–23 Sexual identity differences in tobacco use prevalence are most consistently found among women,^22,24,25 and recent studies have found fewer or no differences in tobacco use by sexual identity among men.^25,26 In particular, bisexual women are consistently at greatest risk for tobacco use compared to heterosexual women^{19,22–24,27,28} and compared to gay/lesbian women using self-report survey data.²⁵ Bisexual and gay/lesbian women are also less likely to report receiving preventative care related to cancer and other tobacco-related health consequences including cancer screenings.²⁹ While evidence has clearly linked nicotine and toxicant exposure from tobacco to cancer and other tobacco-related health risks, neither biomarkers of nicotine nor tobacco toxicant exposure have been examined by sexual identity. Thus, it is not known if sexual identity differences in self-reported tobacco use prevalence among women translate into differences in biologically measured exposure to nicotine and tobacco toxicants. Biomarker data may provide an important connection to understanding the unique health risks of sexual minority women and reducing health disparities for sexual minority women.

Examining biological measures of nicotine exposure and tobacco-related toxicant exposure has important implications for understanding cancer and other health risks for sexual minority populations and can complement tobacco use data collected in surveys. Given previous research, we expect the survey data and biologically measured data to be highly correlated.³⁰ However, differences due to the type of tobacco product used, topography of tobacco use, and differences due to stress levels, as well as other factors, could result in differences between tobacco use survey findings and biomarker findings. In this study we had two research questions:

1. Does nicotine exposure and tobacco toxicant exposure differ by sexual identity among adult women in the U.S.?

2. Are any sexual identity differences in nicotine and tobacco toxicant exposure accounted for by self-reported differences in tobacco use?

METHODS

Study population

Data come from adults included in the Wave 1 Biomarker Core Sample from the Population Assessment of Tobacco and Health (PATH) Study (2013–2014), a nationally representative study of the U.S. civilian household population, aged 18 years and older (n=32,320). In order to maximize the sample we had available to analyze sexual minority women subgroups, we used Wave 1 of the PATH study which was not complicated by loss of participants in follow-ups. PATH used a stratified area probability sampling design that oversampled adult tobacco users, young adults (18–24 years), and Black adults. The study used audio computer-assisted self-Interviews (ACASI) available in English and Spanish to collect data. In addition, adults were invited to provide urine biospecimens at each wave. All participants aged 18 years and older provided informed consent. The study was approved by the Westat IRB³¹ and authors of this secondary data analysis study received exempt status by the University of Michigan Health Sciences and Behavioral Sciences IRB.

Of the 32,320 respondents who completed the Wave 1 adult interview, biomarker data was analyzed for 11,522 adult respondents specifically selected to obtain a representative sample of never, former, and current tobacco product users from the non-institutionalized U.S. population.³¹ This included 4,930 women (92.8% heterosexual, 74.8% white, and a mean age of 43.7 years; see Table 1 for additional descriptive characteristics). Respondents reported on their use of all nicotine-containing products for the three days prior to the time of biospecimen collection. The weighted response rate for the household screener was 54.0%; among those who completed the adult interview, the weighted response rate for those providing a urine sample was 63.6%.³¹

Table 1.

Weighted descriptives of women in the PATH Wave 1 Biomarker Core (2013–2014; n=4,930)

	Overall		Heterosexual		Gay/lesbian		Bisexual		Something else
	%/M	nunweighted	%/M	nunweighted	%/M	nunweighted	%/M	nunweighted	%/M	nunweighted
Overall	100%	4,930	92.8	4,505	2.0%	137	3.6%	416	1.6%	102
Race
White	74.83	3769	76.17		52.7		67.72		68.02
Black	15.38	917	15.09		37.01		19.51		12.42
Multiracial/other	9.09	543	8.74		10.29		12.77		19.56
Hispanic	18.66	907	17.73	740	10.94	32	17.39	65	32.40	24
Age	43.74	--	44.46	--	36.51	--	30.61	--	37.46	--
Region
Northeast	17.41	692	17.24	585	30.90	23	15.74	51	14.36	14
Midwest	20.88	1308	21.00	1114	16.29	22	23.61	112	26.24	22
South	39.86	2123	39.91	1801	34.93	57	37.26	154	29.35	36
West	21.84	1123	21.85	944	17.87	32	23.39	86	30.04	26
Urbanicity
Urban	94.90	4879	94.55	4107	99.03	131	97.14	387	94.77	95
Rural	5.10	367	5.45	337	0.97	3	2.86	16	5.23	3
Secondhand smoke exposure	36.19	2560	35.80	2212	37.27	73	49.54	231	27.35	44
Body mass index	28.48	--	28.36	--	28.89	--	29.86		26.76	--
Internalizing symptoms (0–4)	1.46	--	1.43	--	1.76	--	2.35	--	1.51	--
Any past 30-day alcohol use	44.34	2648	43.44	2247	57.57	88	61.19	255	42.38	58
Any past 30-day marijuana use	8.0	881	6.91	668	17.41	44	29.01	144	12.30	25
Any past 30-day tobacco use	38.59	3742	37.29	3196	48.99	112	74.43	360	42.50	74
# of cigarettes per day	4.72	--	4.75	--	4.56	--	6.86	--	3.36	--
Past 30-day cigarette use	34.52	3252	33.37	2797	39.18	93	62.29	307	33.05	55
Past 30-day e-cigarette use	10.86	1230	9.78	1002	19.98	50	29.75	145	19.73	33
Past 30-day other tobacco use	9.10	1145	7.90	889	24.33	60	29.98	166	12.95	30
Past 30 day # of tobacco products (range 0–9)	0.57	---	0.53	---	0.93	---	1.34	---	0.71	---
WISDM score (0–44)	7.38	--	7.21	--	8.33	--	12.58	--	6.80	--

M=mean

^aOther tobacco use includes: cigars, cigarillos, smokeless, hookah, pipes, snus, and dissolvable tobacco

Biospecimen collection and laboratory procedures

Full-void spot urine samples were self-collected by consenting participants in 500 mL polypropylene containers. Samples were immediately placed in a Credo Cube shipper in order to transport samples between 2°C and 8°C, and transported overnight to the biorepository for storage and processing. Biospecimens were analyzed by the Centers for Disease Control and Prevention Division of Laboratory Sciences using highly selective mass spectrometric methods.^32–33

Urinary biomarkers

We examined cotinine and total nicotine equivalents-2 (TNE-2), measured from urine samples, to estimate exposure to nicotine. TNE-2 takes into account two nicotine metabolites (cotinine and trans-3′-hydroxycotinine values) and is calculated by taking their molar sum. TNE-2 is highly correlated (r=0.97) with TNE-7, the gold standard measure of nicotine exposure.³⁴ We examined urinary 4-(methylnitrosamino)-4-(3-pyridyl)-1-butanol (NNAL), the primary metabolite of NNK,³³ to estimate exposure to TSNAs. Analytical limits of detection (LOD) were 0.03 ng/mL for both trans-3’-Hydroxycotinine and cotinine. The LOD for NNAL urine was 0.6 ng/L.³⁵ Creatinine in urine was measured and used for correction in analyses to account for differences in hydration status on sample collection. Individuals with creatinine concentrations outside of the reference range (≤10 mg/dL or >370 mg/dL) were removed from analyses (n=15).³⁵

Self-report measures

Sexual identity.

Individuals were categorized based on their response to the following: “Do you think of yourself as… (1) lesbian or gay, (2) straight, that is not lesbian or gay, (3) bisexual, or (4) something else.”

Survey measures of frequency and intensity of tobacco use. We examined three different measures of tobacco frequency and intensity: (1) average cigarettes per day. Those who smoked cigarettes in the past 30 days were asked to report the number of cigarettes they smoked per day on days that they smoked. (2) WISDM score. The PATH study assesses 11 items from the Wisconsin Inventory on Dependence and Smoking Motives (WISDM).^36–37 Each item (e.g., “I frequently crave tobacco”) is asked on a five-point scale ranging from “not true of me at all” to “extremely true of me”. We combined the eleven items into a continuous scale ranging from 0 to 44. (3) Number of tobacco products. Individuals reported past 30-day use of nine different tobacco products: cigarettes, e-cigarettes, cigars, cigarillos, smokeless, hookah, pipes, snus, and dissolvable tobacco. We summed the number of tobacco products used in the past 30 days (0–9). As a check, we examined the agreement of cotinine with (using a cutoff of 40 ng/ml in accordance with recent guidelines³⁸) past 30-day tobacco use and past 30-day cigarette use. We found that among those who were above the cotinine cutoff, the vast majority of individuals also self-reported tobacco use in the past 30 days; only 6.2% of those above the cutoff self-reported no tobacco use. This was true regardless of sexual identity, but was particularly low for sexual minorities (0% for gay/lesbian, 0.7% for bisexual, and 0% for those who identified as something else).

Other potential confounders.

Regression analyses also included multiple potential confounders shown to be associated with both tobacco use and with sexual identity including: age, race (black, white, other), Hispanic ethnicity, urbanicity, census region (Northeast, Midwest, South, West), past 30-day cannabis use, past 30-day alcohol use, internalizing symptoms (the Global Appraisal of Individual Needs – Short Screener^39–40 score), second hand smoke in home, and body mass index.

Analysis

We used ordinary least squares regression models to examine the association of sexual identity with cotinine, TNE-2, and NNAL in three separate groups: (1) the full biomarker core sample of women (n=4,930), (2) women who were past 30-day tobacco users in the biomarker core sample (n=3,630), and (3) women who were established current exclusive cigarette users (i.e., cigarette smokers who did not currently use any other tobacco products in the past 30 days or nicotine replacement products in the past 3 days and had used at least 100 cigarettes in their lifetime) (n=1,302). Outcomes were natural log transformed and corrected for urinary creatinine; resulting estimates are expressed as geometric means or geometric mean ratios. For each of these subgroups we first examined associations controlling for all potential confounders (except for survey measures of frequency and intensity of tobacco use), and then examined associations after adding additional controls for survey measures of frequency and intensity of tobacco use. We also examined associations of sexual identity with self-reported number of cigarettes per day, number of tobacco products, and WISDM score among the same three subgroups to compare these self-report findings with our findings using biomarker data.

For all analyses, listwise deletion was used given the small amount of missing data on variables of interest for this study (4.1%). Weights to adjust for nonresponse and sample design for the biomarker sample were included in all analyses and were provided by the PATH study team.³¹

RESULTS

Full adult biomarker sample

In the full biomarker sample (see Table 2), prior to adjusting for survey measures of frequency and intensity of tobacco use, bisexual women had significantly higher cotinine levels (Geometric mean ratio [GMR] = 3.72 [95% CI = 1.94, 7.20]) and higher TNE-2 levels (GMR = 3.72 [95% CI = 1.96, 7.07]). Prior to adjusting for frequency and intensity of tobacco use, bisexual women also had significantly higher NNAL levels (GMR = 2.18 [95% CI = 1.41, 2.28]) compared to heterosexual women. Gay/lesbian women did not differ from heterosexual women in cotinine, TNE-2, or NNAL levels. After adjusting for survey measures of frequency and intensity of tobacco use (i.e., number of tobacco products, cigarettes per day, and WISDM score), there were no significant differences in levels of cotinine, TNE-2, or NNAL among women by sexual identity.

Table 2.

Adjusted associations of sexual identity with tobacco use biomarkers among women in the PATH Wave 1 Biomarker Core (2013–2014)

	Full biomarker core sample (n=4,930)		Past 30-day tobacco users (n=3,630)		Exclusive established cigarette users (n=1,302)
	Model 1: WISDM score, # of cigarettes, # of tobacco products excluded (n = 4,930)	Model 2: WISDM score, # of cigarettes, # of tobacco products included (n = 4,930)	Model 1: WISDM, # of cigs, # of tobacco products excluded (n = 3,630)	Model 2: WISDM, # of cigs, # of tobacco products included (n = 3,630)	Model 1: WISDM, # of cigs excluded (n = 1,302)	Model 2: WISDM, # of cigs included (n = 1,302)
	GMR (95% CI)	GMR (95% CI)	GMR (95% CI)	GMR (95% CI)	GMR (95% CI)	GMR (95% CI)
Cotinine
Gay/lesbian (n=137)	1.54 (0.52, 4.52)	0.98 (0.45, 2.12)	0.71 (0.20, 2.43)	0.74 (0.26, 2.06)	2.10 (1.05, 4.23)*	1.25 (0.50, 3.10)
Bisexual (n=416)	3.73 (1.94, 7.20)*	1.05 (0.73, 1.51)	0.97 (0.67, 1.40)	0.98 (0.74, 1.31)	0.63 (0.32, 1.26)	0.68 (0.40, 1.15)
Something else (n=102)	1.37 (0.43, 4.38)	0.89 (0.53, 1.51)	0.59 (0.21, 1.71)	0.66 (0.28, 1.54)	0.37 (0.07, 2.10)	0.57 (0.12, 2.65)
TNE-2
Gay/lesbian (n=137)	1.76 (0.63, 4.91)	1.12 (0.56, 2.25)	1.04 (0.50, 2.16)	0.87 (0.29, 2.66)	1.98 (0.79, 4.96)	1.22 (0.43, 3.45)
Bisexual (n=416)	3.72 (1.96, 7.07)*	1.06 (0.74, 1.53)	0.68 (0.35, 1.32)	0.97 (0.34, 1.34)	0.58 (0.27, 1.21)	0.62 (0.33, 1.13)
Something else (n=102)	1.46 (0.48, 4.47)	0.96 (0.58, 1.58)	1.62 (0.78, 3.37)	0.81 (0.34, 1.93)	0.48 (0.09, 2.72)	0.73 (0.15, 3.54)
NNAL
Gay/lesbian (n=137)	1.40 (0.70, 2.79)	1.08 (0.68, 1.72)	0.92 (0.42, 1.99)	0.93 (0.50, 1.76)	2.02 (1.35, 3.03)*	1.34 (0.81, 2.23)
Bisexual (n=416)	2.18 (1.41, 2.28)*	1.01 (0.76, 1.35)	0.91 (0.68, 1.21)	0.91 (0.74, 1.12)	0.79 (0.49, 1.29)	0.84 (0.59, 1.20)
Something else (n=102)	1.14 (0.54, 2.42)	0.89 (0.62, 1.27)	0.65 (0.31, 1.34)	0.70 (0.42, 1.18)	0.50 (0.19, 1.31)	0.71 (.033, 1.53)

Note: Reference is heterosexual women. Models control for age, race, ethnicity, urbanicity, census region, marijuana use, alcohol use, household second hand smoke exposure, internalizing symptoms, and BMI. 0.3% of values for cotinine, 0.4% of values for trans-3′-hydroxycotinine, and 12.8% of the values for NNAL were imputed due to values below the limit of detection as describe in the methods. TNE-2 = total nicotine equivalents-2 and is calculated by taking their molar sum of cotinine and trans-3′-hydroxycotinine. NNAL is 4-(methylnitrosamino)-4-(3-pyridyl)-1-butanol, the primary metabolite of NNK. GMR = geometric mean ratio.

^*p<0.05

In the full biomarker sample we also found that self-reported number of cigarettes per day (β = 1.30 [95% CI = 0.05, 2.55]), number of tobacco products (β = 0.42 [95% CI = 0.28, 0.56]), and WISDM scores (β = 2.89 [95% CI = 0.96, 4.82]) were greater for bisexual women compared to heterosexual women. The number of tobacco products was also greater for gay/lesbian women (β = 0.31 [95% CI = 0.06, 0.55]) compared to heterosexual women.

Past 30-day tobacco users

Among past 30-day tobacco users, we found no significant differences in cotinine, TNE-2, or NNAL by sexual identity either before or after adjusting for survey measures of tobacco use frequency and intensity.

Among past 30-day tobacco users, we only found significant differences in self-reported tobacco use for the number of tobacco products used with all three sexual minority groups using a greater number of tobacco products compared to heterosexual women (β ranged from 0.21 to 0.36).

Exclusive established cigarette users

Among exclusive established cigarette users, gay/lesbian women had significantly greater cotinine levels (GMR = 2.10 [95% CI = 1.05, 4.23]) and significantly greater NNAL (GMR = 2.02 [95% CI = 1.35, 3.03]) compared to heterosexual women prior to adjusting for tobacco use quantity and intensity. In this subgroup of exclusive established cigarette users, bisexual women did not differ from heterosexual women in levels of cotinine, TNE-2, or NNAL. After adjusting for tobacco quantity and intensity (number of cigarettes per day and WISDM score), there were no significant differences in cotinine, TNE-2, or NNAL by sexual identity among exclusive established cigarette using women.

Among exclusive established cigarette users, we did not find significant differences in self-report tobacco use measures, except for a significantly lower number of cigarettes used per day among women identifying as something else compared to heterosexual women (β = −4.23 [95% CI = −7.02, −1.44]). Gay/lesbian women trended toward using more cigarettes per day and having a higher WISDM score than heterosexual women, but these differences were not significant.

DISCUSSION

Sexual minority women have greater nicotine exposure and tobacco toxicant exposure; all differences were attenuated once accounting for tobacco frequency and intensity. These findings are consistent with previous research using self-report data showing greater prevalence of tobacco use and tobacco use disorder among sexual minority women compared to heterosexual women.^19–27 Still, highlighting and demonstrating these differences at the biological level provides additional insight into the public health burden of tobacco use among sexual minority women. This study also adds to a broader literature from multiple seminal researchers (see Allen et al., 2014⁵ and Pomerleau et al., 2005⁴¹) that sought to understand and highlight the importance of understanding tobacco use among women, the unique context in which they use, and stressors that put women at risk for tobacco use and its consequences. Given that these biomarkers of exposure are strongly linked to cancer and other health risks, these findings warrant further investigation into tobacco-related health consequences.

While we found that overall, bisexual women had greater nicotine and tobacco toxicant exposure compared to heterosexual women, these results were explained by differences in self-reported tobacco quantity and intensity. Interestingly, among both subgroups of tobacco users that we examined, bisexual women did not have significantly different exposure to nicotine or tobacco toxicants. This finding, along with the self-report findings, suggests that while bisexual women are more likely to use tobacco, bisexual women that use tobacco do not use tobacco in greater quantity or intensity than heterosexual women and, thus, do not have higher levels of nicotine and tobacco toxicant exposure. On the other hand, among women that exclusively use cigarettes, gay/lesbian women do have higher exposure to both nicotine and TSNAs. This was explained by greater intensity and frequency of tobacco use among gay/lesbian women in this selective tobacco using subgroup. Thus, at least for a more traditional tobacco user that only smokes cigarettes, gay/lesbian women are at heightened risk of exposure to smoke constituents responsible for health consequences such as cancer and cardiovascular illness.

The subgroup of women with increased exposure changed depending on the tobacco subgroup examined. This highlights the need to consider the generalizability of our results. Often tobacco biomarker studies and clinical studies use a restrictive set of criteria in an effort to have a consistent and uniform sample and control the all potential confounding influences.⁴² This approach has the advantage of reducing residual confounding and increasing internal validity. However, it is also important to consider the potential implications for generalizability when analyzing a restrictive selective sample that may not generalize to the wider population. This may also have implications for comparing across studies that examine different subgroups of tobacco users. Studies that leverage varied approaches in exposure assessments of tobacco use among sexual minority women will be needed in order to reconcile the nature of exposure and their links to health consequences experienced among this population.

This study adds to our understanding of sexual identity disparities among sexual minority women. Showing these consequences of greater nicotine and tobacco toxicant exposure due to the higher prevalence of tobacco use among bisexual women (and for some subgroups of gay/lesbian women) highlights the imperative to reduce tobacco use among sexual minority women. With continued exposure over time, these toxicants can pose many long-term negative health implications including greater cancer risk. Future research is needed to further examine these exposures and the long-term health outcomes by sexual identity. Some sexual minority women may require more intensive cessation strategies and strategies that are sensitive to the unique needs of sexual minority women in order to combat these disparities.⁴³ Tobacco use prevention strategies and tobacco cessation initiatives that reach sexual minority women and are meaningful to them will have significant public health benefit.

Table 3.

Association of sexual identity with tobacco use quantity and intensity among PATH Wave 1 (2013–14) Biomarker Core women.

	Full biomarker core sample (n=4,930)	Past 30-day tobacco users (n=3,630)	Past 30-day excmsive established cigarette users (n=1,302)
	β (95% CI)	β (95% CI)	β (95% CI)
Number of cigarettes per day
Heterosexual	Ref	Ref	Ref
Gay/lesbian	0.33 (−1.33, 2.00)	−0.15 (−2.63, 2.32)	3.89 (−0.07, 7.86)
Bisexual	1.30 (0.05, 2.55)*	−0.17 (−1.51, 1.18)	−1.40 (−3.73, 0.93)
Something else	0.20 (−2.09, 2.48)	−0.85 (−4.68, 2.97)	−4.23 (−7.02, −1.44)*
Number of tobacco products
Heterosexual	Ref	Ref	Ref
Gay/lesbian	0.31 (0.06, 0.55)*	0.36 (0.15, 0.55)*	---
Bisexual	0.42 (0.28, 0.56)*	0.21 (0.09, 0.33)*	---
Something else	0.25 (−0.08, 0.59)	0.21 (0.01, 0.41)*	---
WISDM score (11 items; 0–44)
Heterosexual	Ref	Ref	Ref
Gay/lesbian	0.18 (−2.84, 3.20)	−0.69 (−4.55, 3.18)	6.52 (−1.35, 14.29)
Bisexual	2.89 (0.96, 4.82)*	−0.12 (−1.96, 1.72)	0.47 (−3.31, 4.26)
Something else	0.70 (−2.29, 3.69)	−0.81 (−4.25, 2.63)	−4.02 (−12.02, 3.98)

Note: Reference is heterosexual women. Models control for age, race, ethnicity, urbanicity, census region, marijuana use, alcohol use, household second hand smoke exposure, internalizing symptoms, and BMI.

^*p<0.05

Highlights.

• Sexual minority compared to heterosexual women had higher levels of nicotine exposure

• Sexual minority compared to heterosexual women had higher levels of TSNAs

• Sexual identity differences varied by tobacco subgroup examined