Sleep Health by Sexual Orientation among Women in the United States and Interrelations with Race/Ethnicity, Age, and Generational Cohort

Abstract

Purpose

To examine associations between sexual orientation and sleep; interrelations with race/ethnicity, age, and generation; and mediation by perceived sexual orientation discrimination among United States (US) women.

Methods

Eligible Sister Study participants (N=50,790) identified as heterosexual or non-heterosexual at enrollment (2003–2009), and self-reported sleep duration, sleep quality, and sleep mask and sleep medication use. We used latent class analyses to determine patterns of sleep health (good, moderate, poor). We investigated race/ethnicity, age, and generation as modifiers and perceived sexual orientation discrimination as a modifier and/or mediator. We used Poisson regression with robust variance to estimate prevalence ratios (PRs; 95% confidence intervals [CIs]) and inverse odds weighting to assess mediation.

Results

Median age was 55 (interquartile range: 49–62) years; 2% identified as non-heterosexual; 86% identified as non-Hispanic White, 9% non-Hispanic Black (NHB), and 5% Hispanic/Latina. Overall, sleep masks were more common among non-heterosexuals than heterosexuals (PR: 1.75 [95% CI: 1.17, 2.61]), and we did not identify other sexual orientation-sleep associations. Among NHB women, sleep medications were more prevalent among non-heterosexual orientation (1.54 [1.01, 2.35]). Among women who ever perceived discrimination, shorter than recommended duration was less common among non-heterosexuals than heterosexuals (0.78 [0.63, 0.96]) as was moderate sleep health compared to good sleep health (0.77 [0.63, 0.94]). Among women who never perceived discrimination, poor sleep health was more common among non-heterosexuals (1.29 [1.02, 1.63]).

Conclusion

Prevalent sleep aid use among sexual minoritized women may help reduce sleep disparities by sexual orientation. Differential sleep experiences among intersecting identities can shape health inequities, and identification of mediating pathways can inform interventions.

Plain Language Summary

Marginalization of non-heterosexual women may lead to poor sleep. Sexual orientation-sleep disparities have been documented; yet, few studies have evaluated multiple sleep measures or interrelations across social identities. We examined sexual orientation and self-reported sleep among women; intersections with race/ethnicity, age, and generation; and both mediation and moderation by perceived sexual orientation discrimination. Non-heterosexual, compared to heterosexual women were more likely to use sleep masks. Among non-Hispanic Black women, non-heterosexual women were more likely to use sleep medications than heterosexual women. Overall, perceived sexual orientation discrimination may modify associations between sexual orientation and both sleep health and sleep duration; perceived sexual orientation discrimination may also mediate the association between sexual orientation and sleep health. Women with intersecting disadvantaged identities experience unique stressors. Future studies should investigate objective sleep aid use over time, discrimination, and other mediating pathways.

Introduction

In the United States, approximately 4.5% of adults identify as lesbian, gay, bisexual, transgender, or queer (LGBTQ).^1,2 These groups may experience a range of oppressive or discriminatory behaviors on the basis of sexual and gender identities, which can lead to poor health behaviors and health-related disparities. For instance, compared to heterosexual persons, sexual minoritized persons have a higher prevalence of heavy alcohol consumption, smoking, illicit drug use, delayed healthcare use, poor mental health (eg, worry, sadness, distress), obesity, and cardiovascular disease.^3–9 These poor health outcomes have also been associated with sleep deficiency and disturbances. Sleep health is essential for health maintenance, but poor sleep may disproportionately burden sexual minoritized individuals throughout the life course.^9–13 Specifically, sexual minoritized persons are more likely to have sleep durations considered either too short or too long, report trouble falling or staying asleep, and use sleep medications.^9,12–17 Further, women have more insomnia and less restful sleep than men,^18,19 and women in same-sex relationships may have less restful sleep than women in different-sex relationships.²⁰ Sleep has been hypothesized as a fundamental contributor to disparities in various health outcomes, including cardiovascular and other chronic conditions, and tailored efforts to improve sleep could help close health gaps.^9–11,17

Socioenvironmental stressors (eg, discrimination, stigma, diminished social support) among sexual minoritized groups may play key roles in producing disparities in sleep and among a range of other health outcomes like cardiovascular health.^9,21,22 For example, internalization of negativity and depletion of emotional, psychological, and coping resources due to these stressors may be harmful to health.^{9,12,14,22–24} Few studies have examined intersections or identification with multiple minoritized social groups, even though poor sleep among sexual minoritized groups may be exacerbated by additional social stressors linked to race/ethnicity (eg, racial/ethnic discrimination) or older age (eg, age-related health conditions).^9,22,25 Some,^14,26 but not all,²⁷ studies suggest that people doubly-minoritized due to sexual orientation and race/ethnicity had worse sleep than both heterosexuals and sexual minoritized White individuals. For aging adults, sleep can be disrupted by changes in circadian rhythms due to hormonal and internal clock shifts, and increases in multimorbidities and their medications.²⁵ However, a nationally representative study by Jackson et al did not identify sexual orientation-sleep differences by age among adults.⁸ More research is needed to clarify the sexual orientation-sleep association among vulnerable subpopulations.

Additionally, while age refers to the number of years lived, generation or birth cohort is defined by a range of birth years to represent a group of individuals who share similar life experiences and historical context. Sociocultural norms and values can be shaped by shared historical and social events, and social acceptance of lesbian, gay, bisexual, transgender, queer or questioning, or another diverse gender identity (LGBTQ+) has increased over time.²⁸ For example, research has shown that for generations ranging from the Silent Generation to Millennials, each younger generation shows increased support for same-sex marriage when compared to the preceding generation, and LGBTQ+ legal protections have accumulated over time.^29,30 However, studies of generational differences in the health of sexual minoritized groups have been mixed.^31–33 For example, among younger generations, studies have found earlier milestones in coming out for sexual minoritized individuals but have also found little change in well-being and more distress and disparities in health.^31–33

Although sleep is multidimensional, existing studies have largely assessed single dimensions (eg, duration) that do not fully capture the breadth of sleep health.^8,9,14,26,27 Investigations into how intersectional identities between sexual orientation, race/ethnicity, age, and generation associate with more comprehensive sleep assessments are needed, including the social stressors that shape them. Findings would illuminate the complex relationship between sexual orientation and multidimensional sleep health, which can guide efforts to tailor sleep promotion within the sexual minoritized community.

To address the research gap, we sought to examine the relationship between sexual orientation and multiple sleep outcomes (ie, duration, quality measures, aids, and a comprehensive sleep health measure) among adult US women in the Sister Study. We then examined these associations within racial/ethnic, age, and generational (or birth) groups to elucidate potential interrelations. Finally, we conducted exploratory analyses of the potential mediating or moderating role of perceived sexual orientation discrimination given its role as a socioenvironmental stressor that may negatively impact sleep.³⁴ We hypothesized that sexual minoritized women have worse sleep – across all measures – than heterosexuals, with the magnitude of association higher among women who are racially/ethnically minoritized and of older age and generational cohort than their counterparts who are non-Hispanic White (NHW) and of younger age and generation. We also hypothesized that perceived sexual orientation discrimination may modify the association of minoritized sexual orientation with sleep among sexual minoritized individuals or partially mediate the association.

Materials and Methods

Study Population: The Sister Study

We used data (Release 10.1) from The Sister Study, a prospective cohort study of 50,884 women which began in 2003. Eligible participants were women aged 35–74 years in the United States, including Puerto Rico, who had a biological sister diagnosed with breast cancer but were free from breast cancer themselves at enrollment. Baseline data collection included a computer-assisted telephone interview and self-administered questionnaires that assessed socioeconomic measures, medical history, and lifestyle factors; anthropometric measurements and biologic specimens were collected during a home visit. Additional methods and recruitment strategies have been described in greater detail elsewhere.³⁵ Variables were largely collected at baseline (2003–2009); however, ever perceived sexual orientation discrimination was collected during follow-up (2–3 years after enrollment; 2008–2012). The Sister Study is overseen and approved by the National Institutes of Health Institutional Review Board (protocol number 02EN271). All participants provided written informed consent.

Study Participants

Of the original sample of 50,884 women, we excluded 94 participants (0.18%). Five withdrew from the study. We excluded those who identified as asexual due to small sample size (n=54) and those who refused the question or were missing data (n=32). We then restricted to participants who had at least one sleep outcome of interest, resulting in an exclusion of three participants and a final analytic sample of 50,790 women (99.8% of the original study population). Analyses that included race/ethnicity were further restricted to the 49,454 participants who self-identified as NHW, non-Hispanic Black (NHB), or Hispanic/Latina (hereafter, Latina) race/ethnicity due to small sample sizes in other racial/ethnic categories (NH American Indian/Alaska Native, NH Asian, NH Native Hawaiian/Pacific Islander; n=1336 excluded). Analyses including perceived sexual orientation discrimination were limited to the subset with these data available (n=46,216).

Measures

Exposure Assessment: Sexual Orientation

We defined participants as heterosexual or non-heterosexual based on responses to, “Would you say you are sexually attracted only to men; sexually attracted only to women; or sexually attracted to both men and women?” Participants who responded as attracted to women or attracted to women and men were categorized as homosexual or bisexual, respectively, and grouped as sexual minoritized non-heterosexuals; participants who said they were attracted only to men were categorized as heterosexual sexual orientation.

Outcome Assessment: Sleep Measures

We examined 11 self-reported sleep measures. Average sleep duration was based on multiple survey questions. Participants were asked, “which of the following best describes your pattern for waking up/going to sleep during the past six weeks?”. If they selected both “I wake up [go to sleep] at about the same time, that is, within 1 hour, every day of the week”, sleep duration was calculated as the difference of sleep-wake times in response to “About what time to you usually go to sleep [wake up for the day]?” If they selected both “I wake up [go to sleep] at about the same time on workdays, but I have a different wake-up time [bedtime] on my days off”, then average difference in sleep-wake times was calculated and weighted based on reported days working and not working. If they responded both, “the time when I wake up [go to bed] varies by 2 or more hours depending on what day of the week it is, but the pattern is consistent from week to week”, then duration was averaged based on sleep-wake times for each day of the week. If participants lacked reported wake and bedtimes due to survey structure then response to a separate, single question was used, “about how many hours and/or minutes of sleep per [night/day] do you get on average?” Duration was categorized as non-mutually exclusive categories: ≤5 hours (very short), <7 hours (short), 7–9 hours (recommended), and >9 hours (long);³⁶ very short and short sleep categories overlap.

Consistent with prior studies,^37–39 we assessed six dichotomous measures of sleep quality: inconsistent sleep, sleep debt, frequent napping, difficulty falling asleep, difficulty staying asleep, and insomnia symptoms. Inconsistent sleep patterns (yes/no) were defined as “yes” if participants reported inconsistent wake up and bedtimes in the previous six weeks and “no” for all other non-missing responses to both wake/sleep questions. Sleep debt was based on the difference between longest and shortest sleep duration, and dichotomized as ≥2 hours vs <2 hours.

For frequent napping, participants were asked “How often do you nap?” which we defined as ≥3 times/week vs <3 times/week (infrequent).^40,41 For difficulty falling asleep, participants were asked “About how long does it take you to fall asleep on average? Would you say you fall asleep in…”, and it was defined as ≥30 minutes vs <30 minutes (no difficulty).⁴² For difficulty staying asleep, participants were asked “When you are asleep, how often do you wake up for any reason? Would you say…” and “on those [nights/days] how many times do you usually wake up…”. Difficulty staying asleep was defined as waking ≥3 nights/week ≥3 times/night vs waking <3 nights/week or waking ≥3 nights/week <3 times/night (no difficulty).⁴³ Insomnia symptoms were defined as either difficulty falling or staying asleep vs neither of those symptoms.^40,44

We used latent class analysis (LCA)⁴⁵ to create comprehensive classes of sleep health. LCA methods classify unmeasured or unobservable class membership using measured variables. We based our latent sleep health assessment on the SATED scale, an acronym for measures of satisfaction, alertness, timing, efficiency, and duration.⁴⁶ We used the following measures to reflect the scale as fully as possible: dichotomized recommended (7–9 hours) vs non-recommended (<7 and >9 hours) sleep duration, napping frequency, difficulty falling asleep, difficulty staying asleep, and inconsistent sleep. We sequentially fit baseline models for 2–5 latent sleep classes and assessed statistical parameters for best fit. The goodness of fit parameters, of which lower estimates are better, included the following: G² relative to degrees of freedom, Akaike information criterion, and Bayesian information criterion (Table S1).^45,47 Next, we used posterior probabilities to evaluate interpretability of each set of 2–5 class models. The best model of three classes was selected based on goodness of fit and interpretability; we then adjusted for the following predictors for best model convergence: sexual orientation, age (<55, ≥55 years), race/ethnicity, educational attainment, annual household income, occupational class, nativity, and self-rated health to strengthen classification (Table S2).^45,48 Because of sample size changes between adjusted and unadjusted models, we re-ran the unadjusted model limited to those with covariate data, obtained the optimized seed, and re-ran the adjusted models for final estimates. We assigned participants to the class in which they had the highest posterior probability of membership and used that class assignment variable as an outcome in subsequent analyses.⁴⁸ Based on available sleep and covariate data, 31,022 (61%) participants were categorized into latent sleep classes.

We also assessed three self-reported sleep aid measures: two sleep medication assessments and sleep mask use. Sleep medication use was extracted from a combination of survey questions and prescription assessments coded using the Slone Drug Dictionary⁴⁹ into two measures: with antihistamines and without antihistamines. A subset of Tricyclic Antidepressants & other Norepinephrine Reuptake Inhibitor antidepressants also used to treat sleep problems (not all antidepressants) were included under medications without antihistamine; medications with antihistamine did not include antidepressants. For sleep mask use, participants were asked “Do you usually sleep with a mask on to keep out light?” and responded yes or no.

Confounders

Sociodemographic characteristics and self-rated health were included in model adjustment sets. For race, participants were asked “What race do you consider yourself to be? You may choose one or more of the following”, and selected between American Indian or Alaska Native, Asian, Black or African American, Native Hawaiian or other Pacific Islander, and White. For ethnicity, participants responded yes/no to “Do you consider yourself to be Hispanic or Latina?” After aforementioned restriction to White and Black race categories due to sample size, race/ethnicity was categorized as NHW, NHB, and Latina. Age was calculated from self-reported date of birth and treated as continuous. Marital status was categorized as legally married or cohabitating vs not married and not cohabitating. Educational attainment was dichotomized as ≤high school and >high school based on the reported highest year or level of school completed. Annual household income before taxes was dichotomized as <$50,000 and ≥$50,000. Current occupational class was categorized into 3 levels based on the US Department Health and Human Services 2003 Industry and Occupation Coding for Death Certificates⁵⁰ and included the following: professional/management, support services, and laborer, with those in the military (n=42) being classified as support services. Furthermore, participants were asked whether they were born in the US (US-born) or in a US territory or outside the US (non-US-born). Region of residence was determined based on reported home address and census regions: Northeast, Midwest, South, and West. For self-rated health, participants were asked “In the past 12 months, would you say your health has generally been…”, and responses were categorized as excellent/very good, good, and fair/poor.

For supplemental information, we examined the prevalence of the following health conditions: cancer (non-invasive and invasive); depression; type 2 diabetes; heart disease; hypertension; obesity; and stroke. Diseases were defined based on either or a combination of self-report of physician diagnosis, medical record or medication documentation, or measurements collected at home visits. Specific definitions and descriptive statistics for these health conditions are provided in Table S3.

Potential Modifiers: Race/Ethnicity, Age, and Generational Cohort

Race/ethnicity and age were previously described. As a modifier, age was dichotomized at the median: <55 years and ≥55 years. We examined the following generational cohorts or generations based on range of birth years: Silent Generation (1928–1945), Baby Boomer (1946–1964), and Generation X (1965–1980).⁵¹

Potential Modifier or Mediator: Sexual Orientation Discrimination

Ever perceived discrimination for sexual orientation was analyzed as both a potential modifier and mediator between minoritized sexual orientation and sleep health. Participants responded yes or no to “Have you ever felt discriminated against because of your sexual orientation?”. As follow-up, participants who responsed yes were asked “If yes, has this [discrimination] happened in the past five years?”.

Statistical Analysis

We presented descriptive sociodemographic and health characteristics for our study population by sexual orientation status in the overall sample and within racial/ethnic groups. First, using Poisson regression with robust variance, we estimated prevalence ratios and 95% confidence intervals (PR [95% CI]) to investigate the association between sexual orientation and sleep outcomes, comparing minoritized non-heterosexuals to heterosexuals in the overall population. There were three model adjustment sets: 1) Model 1 was age-adjusted to account only for age-specific sleep patterns; 2) Model 2 included Model 1 plus sociodemographic factors that influence or are influenced by sexual orientation and sleep patterns: race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, and region of residence; and 3) Model 3 included Model 2 plus self-rated health. For sleep debt, Models 2 and 3 additionally adjusted for inconsistent sleep.³⁷

Second, we examined three potential modifiers: race/ethnicity (NHW, NHB, Latina), age group (<55 years, ≥55 years), and generational cohort (Silent Generation, Baby Boomer, Generation X). To determine effect measure modification, we estimated PRs within each stratum and Wald p-values for interaction terms. As a secondary analysis, we assessed cumulative disadvantage of racial/ethnic and sexual minoritization in relation to sleep by comparing sleep between sexual minoritized women of each racial/ethnic group to NHW heterosexuals.

Third, we assessed the contribution of ever perceived sexual orientation discrimination to the relationship between sexual orientation and sleep in three ways: as an exposure among sexual minoritized women, and post-hoc based on results, as a modifier and then mediator between sexual orientation and sleep. (1) Among only sexual minoritized women, we examined the association between perceived sexual orientation discrimination and all sleep outcomes. (2) For modification, we examined whether perceived sexual orientation discrimination status modified sexual orientation and sleep associations among the overall population. (3) For mediation, we examined whether perceived discrimination mediated the select associations identified in our original analysis between sexual orientation and sleep among the overall population. We used inverse odds weighting (IOW) to estimate natural direct and indirect effects,⁵² which takes the inverse of the odds from a regression of exposure on the mediator and covariates and uses it to weight the primary analytical regression of the sleep outcome on sexual orientation (natural direct effect). Indirect effects were identified by subtracting direct effects from total effects. We adjusted for only sociodemographic factors (Model 2) as self-rated health may also mediate the association. Lastly, we conducted a sensitivity assessment excluding those who had perceived sexual orientation discrimination in the past five years (N=1551) to ensure discrimination did not occur solely after the sleep assessment. The average length of time between baseline and follow-up is 2.8 years. We thus excluded anyone who may have only perceived discrimination in the time between baseline (sexual orientation and sleep measures) and follow-up (ever perceived sexual orientation discrimination) to provide confidence in temporality for interpretation of results.

All analyses were conducted using SAS 9.4 (Cary, NC), and a two-sided p-value of 0.05 was used to determine statistical significance (α ≤ 0.05).

Results

Sociodemographic and Health Characteristics by Sexual Orientation

Sociodemographic and health characteristics of the study population by sexual orientation status, overall (n=50,790), and within races/ethnicities are shown in Tables 1 and S3. Two percent of participants self-identified as lesbian or bi-sexual (n=1128) sexual orientation, herein referenced as “non-heterosexual” since our subgroup does not encompass all sexual minoritized groups. A slightly higher percentage of non-heterosexual women identified as NHW than heterosexual women (89% vs 86%), and a slightly smaller percentage of non-hetersexual women identified as NHB (6% vs 9%) race/ethnicity. Median age was 55 years and 52 years among heterosexual and non-heterosexual women, respectively. The majority of participants attained >high school education, had an annual household income ≥$50,000, and worked in support services; over 70% reported excellent/very good self-rated health. Fifty-three percent of non-heterosexual women reported ever perceiving discrimination vs 5% of heterosexual women.

Table 1.

Sociodemographic and Health Characteristics Among Heterosexual and Sexual Minoritized Women at Enrollment ^a, Overall and by Race/Ethnicity, 2003–2009

		Overall				NH White				NH Black					Latina
		N=50,790 (100%)				N=42,500 (85.9%)				N=4445 (9.0%)					N=2509 (5.1%)
		Heterosexual		Sexual Minoritized		Heterosexual		Sexual Minoritized		Heterosexual		Sexual Minoritized			Heterosexual		Sexual Minoritized
		N	%	N	%	N	%	N	%	N	%	N		%	N	%	N	%
		49,662	(97.8)	1128	(2.2)	41,517	(97.7)	983	(2.3)	4378	(98.5)	67		(1.5)	2459	(98.0)	50	(2.0)
Age (years)	Median (IQR)	55	(49, 62)	52	(47, 59)	56	(49, 62)	52	(47, 59)	53	(47, 59)	50		(43, 55)	53	(46, 60)	50	(46, 57)
	< 55	23,383	(47.1)	678	(60.1)	18,842	(45.4)	581	(59.1)	2440	(55.7)	50		(74.6)	1424	(57.9)	32	(64.0)
	≥ 55	26,279	(52.9)	450	(39.9)	22,675	(54.6)	402	(40.9)	1938	(44.3)	17		(25.4)	1035	(42.1)	18	(36.0)
Race/ethnicity	NH White	41,517	(85.9)	983	(89.4)	41,517	(100.0)	983	(100.0)	0	(0.0)	0		(0.0)	0	(0.0)	0	(0.0)
	NH Black	4378	(9.1)	67	(6.1)	0	(0.0)	0	(0.0)	4378	(100.0)	67		(100.0)	0	(0.0)	0	(0.0)
	Latina	2459	(5.1)	50	(4.5)	0	(0.0)	0	(0.0)	0	(0.0)	0		(0.0)	2459	(100.0)	50	(100.0)
Generational cohort	Silent generation (1928–1945)	14,600	(29.4)	210	(18.6)	13,098	(31.5)	192	(19.5)	739	(16.9)	3		(4.5)	432	(17.6)	10	(20.0)
	Baby boomer (1946–1964)	31,742	(63.9)	814	(72.2)	26,016	(62.7)	708	(72.0)	3172	(72.5)	53		(79.1)	1679	(68.3)	34	(68.0)
	Generation X (1965–1980)	3320	(6.7)	104	(9.2)	2403	(5.8)	83	(8.4)	467	(10.7)	11		(16.4)	348	(14.2)	6	(12.0)
Nativity	US born	47,104	(94.8)	1071	(94.9)	40,663	(97.9)	956	(97.3)	4265	(97.4)	64		(95.5)	1064	(43.3)	25	(50.0)
	Non-US-born	2558	(5.2)	57	(5.1)	854	(2.1)	27	(2.7)	113	(2.6)	3		(4.5)	1395	(56.7)	25	(50.0)
Region of residence	Northeast	8278	(17.0)	242	(21.7)	7548	(18.2)	217	(22.1)	405	(9.3)	17		(25.4)	185	(11.6)	5	(13.9)
	Midwest	13,446	(27.6)	218	(19.6)	12,041	(29.0)	200	(20.4)	988	(22.6)	10		(14.9)	130	(8.1)	4	(11.1)
	South	16,383	(33.6)	324	(29.1)	12,625	(30.4)	273	(27.8)	2648	(60.5)	30		(44.8)	679	(42.5)	12	(33.3)
	West	10,669	(21.9)	329	(29.6)	9283	(22.4)	292	(29.7)	335	(7.7)	10		(14.9)	604	(37.8)	15	(41.7)
Marital status	Married/cohabitating	37,265	(75.0)	702	(62.2)	32,286	(77.8)	630	(64.1)	2278	(52.1)	28		(41.8)	1698	(69.1)	27	(54.0)
	Not married/cohabitating	12,391	(25.0)	426	(37.8)	9228	(22.2)	353	(35.9)	2097	(47.9)	39		(58.2)	761	(30.9)	23	(46.0)
Educational attainment	≤ High school	7711	(15.5)	74	(6.6)	6432	(15.5)	62	(6.3)	444	(10.1)	5		(7.5)	608	(24.7)	5	(10.0)
	> High school	41,947	(84.5)	1054	(93.4)	35,082	(84.5)	921	(93.7)	3933	(89.9)	62		(92.5)	1851	(75.3)	45	(90.0)
Annual household income	< $50,000	12,227	(25.6)	311	(28.1)	9283	(23.3)	264	(27.3)	1357	(31.9)	19		(29.7)	1213	(51.2)	18	(36.7)
	≥ $50,000	35,477	(74.4)	797	(71.9)	30,534	(76.7)	703	(72.7)	2892	(68.1)	45		(70.3)	1156	(48.8)	31	(63.3)
Occupational class	Professional/ management	9495	(29.8)	306	(35.8)	7772	(29.5)	270	(36.0)	1098	(34.2)	14		(26.9)	361	(24.4)	15	(41.7)
	Support services	20,573	(64.5)	483	(56.5)	17,184	(65.2)	428	(57.1)	1925	(59.9)	28		(53.8)	936	(63.2)	17	(47.2)
	Laborers	1844	(5.8)	66	(7.7)	1415	(5.4)	51	(6.8)	190	(5.9)	10		(19.2)	183	(12.4)	4	(11.1)
General self-rated health	Excellent/very good	36,574	(73.7)	813	(72.1)	32,108	(77.4)	726	(73.9)	2419	(55.3)	40		(59.7)	1196	(48.6)	31	(62.0)
	Good	9812	(19.8)	233	(20.7)	7248	(17.5)	192	(19.6)	1465	(33.5)	21		(31.3)	766	(31.2)	13	(26.0)
	Fair/poor	3267	(6.6)	81	(7.2)	2153	(5.2)	64	(6.5)	494	(11.3)	6		(9.0)	497	(20.2)	6	(12.0)
Sexual orientation discrimination	No	42,844	(94.8)	478	(46.7)	36,643	(95.4)	409	(45.4)	3126	(88.0)	33		(58.9)	1996	(96.9)	25	(58.1)
	Yes	2348	(5.2)	546	(53.3)	1775	(4.6)	491	(54.6)	428	(12.0)	23		(41.1)	63	(3.1)	18	(41.9)

Notes: ^a Sexual minoritized includes homosexual or bisexual.

Abbreviations: IQR, interquartile range; NH, non-Hispanic; US, United States.

By sexual orientation, health conditions were comparable between heterosexual and non-heterosexual women (Table S3), though depression was more common among non-heterosexual participants (45% vs 32%) and hypertension among heterosexual participants (49% vs 42%). By race/ethnicity, cancer was most prevalent among NHW heterosexual women and depression among NHW non-heterosexual women. Type 2 diabetes was most common among NHB and Latina participants, hypertension among NHB participants, and obesity among NHB and non-heterosexual Latina participants.

Sleep Characteristics by Sexual Orientation

Sleep Duration and Quality

For sleep outcomes (Table 2), overall, 74% reported recommended durations of 7–9 hours of sleep/night. Among NHB women, heterosexuals reported more recommended sleep duration (51%) than non-heterosexuals (44%). Minoritized racial/ethnic groups had a higher prevalence of insomnia symptoms. NHB women, particularly non-heterosexuals, and heterosexual Latina women reported inconsistent sleep and sleep debt more often than other groups.

Table 2.

Sleep Characteristics by Sexual Orientation ^a, Overall and by Race/Ethnicity, 2003–2009

		Overall					NH White				NH Black				Latina
		N=50,790 (100%)					N=42,500 (85.9%)				N=4445 (9.0%)				N=2509 (5.1%)
		Heterosexual		Sexual Minoritized			Heterosexual		Sexual Minoritized		Heterosexual		Sexual Minoritized		Heterosexual		Sexual Minoritized
		N	%	N	%		N	%	N	%	N	%	N	%	N	%	N	%
		49,662	(97.8)	1128	(2.2)		41,517	(97.7)	983	(2.3)	4378	(98.5)	67	(1.5)	2459	(98.0)	50	(2.0)
Sleep duration b	Very Short (≤5 hours)	2007	(4.0)	51	(4.5)		1227	(3.0)	39	(4.0)	526	(12.0)	9	(13.4)	182	(7.4)	3	(6.0)
	Short (<7 hours)	10,879	(21.9)	243	(21.6)		7853	(18.9)	184	(18.7)	1975	(45.4)	36	(54.5)	712	(29.1)	15	(30.0)
	Recommended (7–9 hours)	36,652	(73.9)	828	(73.5)		31,904	(76.9)	748	(76.2)	2228	(51.2)	29	(43.9)	1613	(66.0)	34	(68.0)
	Long (> 9)	2040	(4.1)	55	(4.9)		1712	(4.1)	50	(5.1)	151	(3.5)	1	(1.5)	120	(4.9)	1	(2.0)
Sleep quality	Inconsistent sleep	7799	(15.7)	182	(16.1)		5988	(14.4)	150	(15.3)	999	(22.8)	18	(26.9)	571	(23.2)	7	(14.0)
	Sleep debt ≥ 2 hours	11,883	(24.6)	301	(27.8)		9332	(23.1)	249	(26.3)	1417	(33.8)	28	(44.4)	778	(33.0)	11	(23.9)
	Frequent napping ≥ 3 times/week	5421	(10.9)	115	(10.2)		4251	(10.2)	95	(9.7)	638	(14.6)	11	(16.4)	346	(14.1)	4	(8.0)
	Difficulty falling asleep ≥ 30 minutes	9026	(18.2)	212	(18.8)		6763	(16.3)	157	(16.0)	1219	(27.9)	27	(40.3)	747	(30.4)	19	(38.0)
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night	6894	(13.9)	170	(15.1)		5779	(13.9)	141	(14.4)	547	(12.5)	12	(17.9)	399	(16.3)	6	(12.0)
	Insomnia symptoms c	13,646	(27.5)	324	(28.7)		10,799	(26.0)	259	(26.3)	1515	(34.6)	30	(44.8)	937	(38.1)	21	(42.0)
Latent sleep health	Good	19,237	(63.7)	556	(67.2)		18,014	(70.3)	524	(70.8)	455	(14.5)	10	(19.6)	768	(53.1)	22	(61.1)
	Moderate	7194	(23.8)	160	(19.3)		5102	(19.9)	135	(18.2)	1769	(56.5)	21	(41.2)	323	(22.4)	4	(11.1)
	Poor	3764	(12.5)	111	(13.4)		2502	(9.8)	81	(10.9)	908	(29.0)	20	(39.2)	354	(24.5)	10	(27.8)
Sleep aids	Sleep medications ever use, with antihistamines	17,369	(35.5)	424	(38.3)		14,854	(36.1)	361	(37.4)	1199	(29.1)	28	(43.8)	844	(35.4)	20	(41.7)
	Sleep medication ever use, no antihistamines	15,008	(30.7)	385	(34.9)		12,954	(31.5)	332	(34.4)	911	(22.2)	24	(38.1)	734	(30.8)	17	(36.2)
	Sleep mask use	875	(1.8)	36	(3.2)		676	(1.6)	33	(3.4)	92	(2.1)	2	(3.0)	76	(3.1)	1	(2.0)

Notes: ^a Heterosexual or sexual minoritized (homosexual or bisexual). ^b Very short and short duration categories are not mutually exclusive. ^c Insomnia symptoms refer to difficulty falling asleep ≥ 30 minutes or difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night.

Abbreviation: NH, non-Hispanic.

Latent Classes of Sleep Health

Based on results for best fit and interpretability, we identified three classes of sleep health (Figure 1). Fit statistics generally improved for crude models as number of classes increased (Table S1), but adjusted models were limited by small cell sizes in the 4- and 5-class models. We therefore created a covariate-adjusted 3-class latent variable of sleep health (Table S2) and used maximum probability to assign a class. Among NHW participants, comparable by sexual orientation, the majority had good sleep health (70%), followed by moderate (18-20%) and then poor (10-11%) (Table 2 and Figure 2). Among NHB women, the majority had moderate sleep health, followed by poor, and then good; non-heterosexual participants had more poor sleep (40%) than heterosexual participants (29%) and all other racial/ethnic-sexual orientation subgroups. Among Latina women, the majority had good sleep health, followed by poor and then moderate; good and poor sleep were both more common among non-heterosexual than heterosexual Latina participants.

Figure 1.

Adjusted probabilities of endorsement for poor sleep measures given latent 3-class memberships: The Sister Study (2003–2009).

Notes: Adjusted for sexual orientation, age (<55, ≥55), race/ethnicity, educational attainment, annual household income, occupational class, nativity, self-rated health.

Figure 2.

Distribution of latent sleep health classes by sexual orientation, overall and within race/ethnicity: The Sister Study (2003–2009).

Sleep Aids

Ever use of sleep medication, with and without antihistamines, was more common among non-heterosexual than heterosexual participants overall (Table 2). NHB non-heterosexual women had the highest use compared to other racial/ethnic-sexual orientation subgroups (with antihistamines: 44%) and the greatest intra-racial/ethnic disparity by sexual orientation (without antihistamines: 38% vs 22%). Sleep mask use ranged from 2–3% among all race/ethnicity and sexual orientation sub-groups.

Associations Between Sexual Minoritized Status and Sleep in the Overall Population

When adjusting only for age, non-heterosexual women were less likely than heterosexuals to have moderate sleep health than good (PR [95% CI]: 0.82 [0.71, 0.94]), and more likely to have ever used sleep medication (with antihistamines: 1.10 [1.02, 1.19]; without antihistamines: 1.17 [1.07, 1.26]; Table 3). However, when adjusting for social and health factors, the associations were attenuated. After full adjustment, the prevalence of sleep mask use was more common among non-heterosexual women than heterosexuals (1.75 [1.17, 2.61]). We did not identify associations between sexual orientation and other sleep outcomes in sociodemographic and health adjusted models.

Table 3.

Adjusted ^a,b,c Prevalence Ratios (95% Confidence Intervals) ^d for Sleep Characteristics by Sexual Orientation (Reference=heterosexuals), Overall and by Race/Ethnicity: The Sister Study (2003–2009; N=50,782)

		PR (95% CI) for Sexual Minoritized (vs Heterosexuals) Women
		Overall	NH White	NH Black	Latina	p-value e
Model 1 a
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.13 (0.86, 1.48)	1.39 (1.02, 1.89)*	1.30 (0.73, 2.33)	0.81 (0.27, 2.43)	0.535
	Short, <7 hours	0.98 (0.88, 1.10)	1.00 (0.88, 1.14)	1.18 (0.95, 1.47)	1.00 (0.65, 1.53)	0.480
	Long, >9 hours	1.19 (0.92, 1.54)	1.24 (0.95, 1.63)	0.53 (0.08, 3.67)	0.41 (0.06, 2.87)	0.120
Sleep quality	Frequent napping ≥3 times/week (vs infrequent)	1.04 (0.88, 1.24)	1.07 (0.88, 1.30)	1.31 (0.75, 2.29)	0.59 (0.23, 1.51)	0.232
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.05 (0.93, 1.19)	1.01 (0.87, 1.16)	1.49 (1.11, 2.00)*	1.26 (0.88, 1.80)	0.089
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.13 (0.98, 1.30)	1.08 (0.92, 1.26)	1.50 (0.89, 2.53)	0.75 (0.35, 1.60)	0.305
	Insomnia symptoms (vs none)	1.07 (0.98, 1.17)	1.04 (0.94, 1.16)	1.34 (1.03, 1.75)*	1.11 (0.80, 1.54)	0.298
	Inconsistent sleep (vs consistent)	1.08 (0.95, 1.24)	1.13 (0.97, 1.31)	1.27 (0.86, 1.89)	0.61 (0.31, 1.23)	0.083
	Sleep debt ≥ 2 hours f (vs none)	1.04 (0.96, 1.12)	1.03 (0.94, 1.12)	1.10 (0.87, 1.39)	0.90 (0.59, 1.36)	0.677
Latent sleep health (vs good)	Moderate	0.82 (0.71, 0.94)*	0.93 (0.80, 1.09)	0.86 (0.67, 1.10)	0.52 (0.21, 1.29)	0.277
	Poor	1.03 (0.87, 1.22)	1.13 (0.92, 1.38)	1.03 (0.80, 1.34)	1.00 (0.59, 1.68)	0.834
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.10 (1.02, 1.19)*	1.05 (0.97, 1.15)	1.54 (1.16, 2.04)*	1.18 (0.84, 1.65)	0.095
	Sleep medication ever use, no antihistamines (vs never)	1.17 (1.07, 1.26)*	1.12 (1.02, 1.22)*	1.77 (1.29, 2.44)*	1.18 (0.81, 1.73)	0.093
	Sleep mask use (vs none)	1.83 (1.31, 2.53)*	2.09 (1.48, 2.94)*	1.44 (0.36, 5.72)	0.65 (0.09, 4.58)	0.179
Model 2 b
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.16 (0.83, 1.62)	1.10 (0.72, 1.69)	1.28 (0.72, 2.27)	1.41 (0.37, 5.38)	0.887
	Short, <7 hours	0.97 (0.85, 1.11)	0.94 (0.81, 1.10)	1.15 (0.89, 1.48)	0.97 (0.53, 1.76)	0.471
	Long, >9 hours	1.05 (0.73, 1.52)
Sleep quality	Frequent napping ≥3 times/week (vs infrequent)	1.03 (0.81, 1.32)	0.99 (0.76, 1.30)	1.49 (0.78, 2.85)	0.74 (0.20, 2.75)	0.522
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.07 (0.92, 1.25)	1.03 (0.86, 1.23)	1.29 (0.88, 1.90)	1.38 (0.77, 2.49)	0.463
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.04 (0.86, 1.25)	1.00 (0.82, 1.22)	1.47 (0.77, 2.80)	1.54 (0.62, 3.84)	0.488
	Insomnia symptoms (vs none)	1.05 (0.94, 1.19)	1.03 (0.91, 1.18)	1.15 (0.80, 1.64)	1.35 (0.82, 2.23)	0.599
	Inconsistent sleep (vs consistent)	1.11 (0.92, 1.33)	1.10 (0.90, 1.34)	1.14 (0.69, 1.89)	1.15 (0.46, 2.88)	0.989
	Sleep debt ≥ 2 hours f (vs none)	1.00 (0.91, 1.10)	0.98 (0.89, 1.10)	1.16 (0.88, 1.51)	0.91 (0.57, 1.45)	0.532
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.91 (0.78, 1.06)	0.84 (0.66, 1.06)	0.49 (0.17, 1.42)	0.336
	Poor	1.10 (0.93, 1.30)	1.11 (0.90, 1.36)	1.03 (0.78, 1.35)	1.33 (0.71, 2.50)	0.765
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.03 (0.94, 1.14)	1.01 (0.91, 1.12)	1.44 (1.01, 2.06)*	1.08 (0.62, 1.89)	0.278
	Sleep medication ever use, no antihistamines (vs never)	1.10 (1.00, 1.22)	1.08 (0.97, 1.21)	1.54 (1.00, 2.37)	1.13 (0.62, 2.07)	0.420
	Sleep mask use (vs none)	1.75 (1.17, 2.62)*
Model 3 c
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.15 (0.82, 1.61)	1.10 (0.72, 1.68)	1.23 (0.71, 2.15)	1.47 (0.37, 5.82)	0.900
	Short, <7 hours	0.97 (0.85, 1.10)	0.94 (0.81, 1.09)	1.15 (0.89, 1.49)	0.98 (0.54, 1.79)	0.449
	Long, >9 hours	1.05 (0.73, 1.52)
Sleep quality	Frequent napping ≥3 times/week (vs infrequent)	1.03 (0.81, 1.31)	0.99 (0.75, 1.29)	1.51 (0.79, 2.86)	0.78 (0.21, 2.84)	0.524
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.07 (0.92, 1.25)	1.02 (0.86, 1.21)	1.31 (0.90, 1.91)	1.45 (0.80, 2.64)	0.372
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.03 (0.86, 1.24)	0.99 (0.82, 1.21)	1.48 (0.78, 2.80)	1.61 (0.65, 4.02)	0.446
	Insomnia symptoms (vs none)	1.05 (0.93, 1.18)	1.02 (0.90, 1.17)	1.16 (0.81, 1.65)	1.41 (0.86, 2.32)	0.495
	Inconsistent sleep (vs consistent)	1.10 (0.92, 1.32)	1.09 (0.90, 1.33)	1.15 (0.70, 1.89)	1.22 (0.49, 3.05)	0.962
	Sleep debt ≥ 2 hours f (vs none)	1.00 (0.90, 1.10)	0.98 (0.88, 1.09)	1.15 (0.87, 1.51)	0.92 (0.58, 1.47)	0.569
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.91 (0.78, 1.06)	0.83 (0.66, 1.05)	0.50 (0.17, 1.44)	0.353
	Poor	1.11 (0.94, 1.30)	1.10 (0.90, 1.34)	1.04 (0.78, 1.38)	1.49 (0.76, 2.93)	0.680
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.03 (0.94, 1.13)	1.00 (0.91, 1.11)	1.44 (1.01, 2.06)*	1.11 (0.63, 1.95)	0.253
	Sleep medication ever use, no antihistamines (vs never)	1.10 (0.99, 1.22)	1.08 (0.97, 1.20)	1.54 (1.01, 2.35)*	1.16 (0.63, 2.15)	0.394
	Sleep mask use (vs none)	1.75 (1.17, 2.61)*

Notes: *p ≤ 0.05 and estimates bolded for significance. ^a Adjusted for age (continuous). ^b Adjusted for age (continuous), race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence. ^c Adjusted for age (continuous), race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence, self-rated health. ^d Missing estimates due to model non-convergence. ^e Wald p-value for interaction between sexual orientation and race. ^f Sleep debt models additionally adjusted for inconsistent sleep. These results show the associations between sexual minoritized status and sleep in the overall population and by race/ethnicity, the primary and secondary aims of this research.

Abbreviations: CI, confidence interval; PR, prevalence ratio.

Associations Between Sexual Minoritized Status and Sleep by Race/Ethnicity

We did not find statistical evidence that associations between sexual orientation and sleep were modified by race/ethnicity (Table 3). However, results were suggestive for sleep medication: among NHB participants, use was more prevalent among non-heterosexual than heterosexual women in fully-adjusted models (with antihistamines: 1.44 [1.01, 2.06]; no antihistamines: 1.54 [1.01, 2.35]); associations were attenuated particularly among NHW participants (with antihistamines: 1.00 [0.91, 1.11]; no antihistamines: 1.08 [0.97, 1.20]). Additionally, when solely adjusting for age, among NHB participants, we identified an increased prevalence of difficulty falling asleep (1.49 [1.11, 2.00]) and insomnia symptoms (1.34 [1.03, 1.75]) among non-heterosexuals, while the same associations were not identified among NHW or Latina women; all associations attenuated when adjusting for social and health factors.

In supplemental analyses, we examined cumulative disadvantage via a common analytic referent of heterosexual NHW women within each racial/ethnic strata (Table S4). NHB non-heterosexual women had a higher prevalence of multiple harmful sleep outcomes compared to NHW heterosexual women, including short sleep duration, insomnia symptoms, and moderate and poor sleep health in fully adjusted models. Latina non-heterosexual participants had higher prevalences of difficulty falling asleep and poor sleep health (vs good) than NHW heterosexuals.

Associations Between Sexual Minoritized Status and Sleep by Age Group and Generation

We did not find statistical evidence that sexual orientation and sleep were modified by age group (Table 4) or generation (Table 5).

Table 4.

Adjusted ^a,b,c Prevalence Ratios (95% Confidence Intervals) for Sleep Characteristics by Sexual Orientation (Reference=heterosexuals), Overall and by Age Group: The Sister Study (2003–2009; N=50,782)

		PR (95% CI) for Sexual Minoritized (vs Heterosexuals) Women
		Overall	< 55 Years	≥ 55 Years	p-value d
Model 1 a
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.12 (0.86, 1.47)	1.08 (0.75, 1.54)	1.19 (0.79, 1.79)	0.720
	Short, <7 hours	0.98 (0.88, 1.10)	0.96 (0.84, 1.11)	1.01 (0.85, 1.21)	0.671
	Long, >9 hours	1.19 (0.92, 1.55)	1.04 (0.72, 1.51)	1.40 (0.97, 2.02)	0.278
Sleep quality	Inconsistent sleep (vs consistent)	1.07 (0.94, 1.22)	1.18 (0.99, 1.41)	0.95 (0.77, 1.17)	0.115
	Sleep debt ≥ 2 hours e (vs none)	1.05 (0.97, 1.13)	1.03 (0.94, 1.14)	1.09 (0.96, 1.23)	0.537
	Frequent napping ≥3 times/week (vs infrequent)	0.99 (0.83, 1.18)	1.04 (0.81, 1.34)	0.95 (0.74, 1.21)	0.611
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.05 (0.93, 1.18)	1.05 (0.90, 1.24)	1.04 (0.86, 1.26)	0.925
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.12 (0.97, 1.29)	1.14 (0.94, 1.38)	1.09 (0.89, 1.34)	0.762
	Insomnia symptoms (vs none)	1.06 (0.97, 1.17)	1.07 (0.94, 1.21)	1.06 (0.92, 1.22)	0.962
Latent sleep health (vs good)	Moderate	0.82 (0.71, 0.94)*	0.79 (0.67, 0.94)*	0.88 (0.70, 1.11)	0.490
	Poor	1.03 (0.86, 1.22)	1.06 (0.85, 1.30)	0.97 (0.72, 1.31)	0.654
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.09 (1.01, 1.18)*	1.12 (1.01, 1.23)*	1.07 (0.95, 1.20)	0.555
	Sleep medication ever use, no antihistamines (vs never)	1.16 (1.07, 1.25)*	1.19 (1.07, 1.33)*	1.11 (0.98, 1.26)	0.415
	Sleep mask use (vs none)	1.83 (1.32, 2.54)*	1.91 (1.25, 2.92)*	1.71 (1.01, 2.89)*	0.741
Model 2 b
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.16 (0.83, 1.63)	1.22 (0.83, 1.80)	0.98 (0.50, 1.91)	0.552
	Short, <7 hours	0.97 (0.86, 1.11)	0.95 (0.81, 1.11)	0.99 (0.80, 1.24)	0.760
	Long, >9 hours	1.06 (0.73, 1.53)	0.83 (0.50, 1.37)	1.58 (0.92, 2.70)	0.117
Sleep quality	Inconsistent sleep (vs consistent)	1.10 (0.92, 1.33)	1.20 (0.97, 1.50)	0.93 (0.66, 1.30)	0.179
	Sleep debt ≥ 2 hours e (vs none)	1.00 (0.91, 1.10)	1.02 (0.91, 1.13)	0.99 (0.81, 1.21)	0.826
	Frequent napping ≥3 times/week (vs infrequent)	1.04 (0.81, 1.32)	1.15 (0.86, 1.54)	0.77 (0.49, 1.20)	0.122
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.07 (0.92, 1.25)	1.10 (0.91, 1.32)	1.04 (0.79, 1.36)	0.746
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.04 (0.86, 1.25)	1.06 (0.84, 1.33)	0.97 (0.71, 1.32)	0.628
	Insomnia symptoms (vs none)	1.05 (0.94, 1.19)	1.07 (0.93, 1.24)	1.01 (0.82, 1.24)	0.643
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.85 (0.72, 1.00)	0.95 (0.75, 1.19)	0.483
	Poor	1.10 (0.93, 1.30)	1.11 (0.91, 1.36)	1.07 (0.80, 1.43)	0.810
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.03 (0.94, 1.14)	1.08 (0.96, 1.21)	0.93 (0.79, 1.11)	0.162
	Sleep medication ever use, no antihistamines (vs never)	1.10 (1.00, 1.22)	1.15 (1.02, 1.30)*	1.00 (0.83, 1.19)	0.186
	Sleep mask use (vs none)	1.75 (1.17, 2.62)*	2.01 (1.26, 3.21)*	1.25 (0.56, 2.78)	0.275
Model 3 c
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.15 (0.82, 1.61)	1.21 (0.82, 1.77)	0.98 (0.50, 1.91)	0.580
	Short, <7 hours	0.97 (0.85, 1.11)	0.95 (0.81, 1.12)	0.99 (0.79, 1.24)	0.780
	Long, >9 hours	1.05 (0.73, 1.52)	0.83 (0.50, 1.37)	1.57 (0.92, 2.68)	0.117
Sleep quality	Inconsistent sleep (vs consistent)	1.10 (0.92, 1.32)	1.21 (0.97, 1.50)	0.91 (0.65, 1.27)	0.147
	Sleep debt ≥ 2 hours e (vs none)	1.00 (0.90, 1.10)	1.02 (0.91, 1.13)	0.98 (0.80, 1.20)	0.780
	Frequent napping ≥3 times/week (vs infrequent)	1.03 (0.81, 1.32)	1.15 (0.86, 1.54)	0.76 (0.49, 1.18)	0.102
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.07 (0.92, 1.25)	1.10 (0.91, 1.32)	1.02 (0.78, 1.34)	0.658
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.04 (0.86, 1.24)	1.06 (0.84, 1.33)	0.95 (0.70, 1.30)	0.586
	Insomnia symptoms (vs none)	1.05 (0.93, 1.18)	1.07 (0.93, 1.24)	1.00 (0.81, 1.22)	0.561
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.85 (0.72, 1.00)	0.94 (0.74, 1.19)	0.497
	Poor	1.11 (0.94, 1.30)	1.12 (0.92, 1.37)	1.06 (0.80, 1.41)	0.760
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.03 (0.94, 1.13)	1.08 (0.96, 1.20)	0.92 (0.78, 1.09)	0.134
	Sleep medication ever use, no antihistamines (vs never)	1.10 (0.99, 1.22)	1.15 (1.01, 1.30)*	0.99 (0.82, 1.18)	0.158
	Sleep mask use (vs none)	1.75 (1.17, 2.61)*	2.01 (1.26, 3.21)*	1.24 (0.55, 2.77)	0.269

Notes: *p ≤ 0.05 and estimates bolded for significance. ^a Overall models adjusted for age. Age stratified models unadjusted for other covariates. ^b Adjusted for race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence. ^c Adjusted for race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence, self-rated health. ^d Wald p-value for interaction between sexual orientation and age. ^e Sleep debt models additionally adjusted for inconsistent sleep. These results show the associations between sexual minoritized status and sleep in the overall population and by age group, the primary and secondary aims of this research.

Abbreviations: CI, confidence interval; PR, prevalence ratio.

Table 5.

Adjusted ^a,b,c Prevalence Ratios (95% Confidence Intervals) for Sleep Characteristics by Sexual Orientation (Reference=heterosexuals), Overall and by Generation: The Sister Study (2003–2009; N=50,782)

		PR (95% CI) for Sexual Minoritized (vs Heterosexuals) Women
		Overall	Generation X (1965–1980)	Baby Boomer (1946–1964)	Silent Generation (1928–1945)	p-value d
Model 1 a
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.13 (0.86, 1.48)	1.99 (0.96, 4.12)	1.12 (0.81, 1.54)	0.84 (0.41, 1.75)	0.341
	Short, <7 hours	0.98 (0.88, 1.10)	1.33 (0.96, 1.84)	0.98 (0.86, 1.12)	0.82 (0.60, 1.12)	0.143
	Long, >9 hours	1.19 (0.92, 1.54)	1.21 (0.55, 2.63)	1.12 (0.81, 1.56)	1.42 (0.85, 2.36)	0.773
Sleep quality	Inconsistent sleep (vs consistent)	1.08 (0.95, 1.24)	0.85 (0.51, 1.42)	1.20 (1.02, 1.40)*	0.88 (0.65, 1.19)	0.098
	Sleep debt ≥ 2 hours e (vs none)	1.04 (0.96, 1.12)	1.07 (0.82, 1.40)	1.04 (0.95, 1.13)	0.97 (0.80, 1.19)	0.812
	Frequent napping ≥3 times/week (vs infrequent)	1.04 (0.88, 1.24)	1.03 (0.52, 2.04)	1.06 (0.85, 1.32)	1.01 (0.74, 1.38)	0.965
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.05 (0.93, 1.19)	1.01 (0.66, 1.55)	1.06 (0.92, 1.23)	1.03 (0.77, 1.36)	0.957
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.13 (0.98, 1.30)	1.73 (1.13, 2.67)*	1.08 (0.91, 1.28)	1.10 (0.81, 1.50)	0.248
	Insomnia symptoms (vs none)	1.07 (0.98, 1.17)	1.25 (0.93, 1.69)	1.06 (0.94, 1.18)	1.05 (0.86, 1.30)	0.611
Latent sleep health (vs good)	Moderate	0.82 (0.71, 0.94)*	0.86 (0.56, 1.30)	0.80 (0.69, 0.94)*	0.93 (0.63, 1.38)	0.801
	Poor	1.03 (0.87, 1.22)	1.12 (0.67, 1.88)	1.09 (0.90, 1.31)	0.54 (0.25, 1.17)	0.076
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.10 (1.02, 1.19)*	1.30 (1.00, 1.69)*	1.08 (0.98, 1.18)	1.11 (0.94, 1.31)	0.446
	Sleep medication ever use, no antihistamines (vs never)	1.17 (1.07, 1.26)*	1.40 (1.04, 1.89)*	1.14 (1.04, 1.26)*	1.16 (0.97, 1.39)	0.496
	Sleep mask use (vs none)	1.83 (1.31, 2.53)*	3.09 (1.37, 6.99)*	1.48 (0.96, 2.27)	2.58 (1.34, 4.94)*	0.247
Model 2 b
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.16 (0.83, 1.62)	1.78 (0.71, 4.47)	1.16 (0.80, 1.68)	0.43 (0.06, 2.86)	0.194
	Short, <7 hours	0.97 (0.85, 1.11)	1.17 (0.81, 1.70)	0.97 (0.84, 1.12)	0.81 (0.51, 1.31)	0.480
	Long, >9 hours	1.05 (0.73, 1.52)	0.89 (0.29, 2.74)	1.05 (0.69, 1.60)	1.33 (0.51, 3.50)	0.866
Sleep quality	Inconsistent sleep (vs consistent)	1.11 (0.92, 1.33)	0.90 (0.48, 1.69)	1.19 (0.98, 1.45)	0.74 (0.38, 1.45)	0.229
	Sleep debt ≥ 2 hours e (vs none)	1.00 (0.91, 1.10)	1.01 (0.76, 1.36)	1.00 (0.90, 1.11)	0.85 (0.54, 1.35)	0.765
	Frequent napping ≥3 times/week (vs infrequent)	1.03 (0.81, 1.32)	1.22 (0.55, 2.69)	1.09 (0.83, 1.43)	0.66 (0.31, 1.42)	0.328
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.07 (0.92, 1.25)	1.08 (0.68, 1.73)	1.12 (0.95, 1.33)	0.69 (0.37, 1.29)	0.208
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.04 (0.86, 1.25)	1.38 (0.76, 2.49)	1.01 (0.82, 1.25)	1.01 (0.58, 1.76)	0.692
	Insomnia symptoms (vs none)	1.05 (0.94, 1.19)	1.12 (0.77, 1.64)	1.08 (0.95, 1.23)	0.81 (0.53, 1.25)	0.373
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.94 (0.65, 1.38)	0.86 (0.74, 1.01)	1.00 (0.68, 1.49)	0.748
	Poor	1.10 (0.93, 1.30)	1.29 (0.82, 2.03)	1.13 (0.94, 1.36)	0.66 (0.31, 1.44)	0.214
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.03 (0.94, 1.14)	1.16 (0.85, 1.59)	1.04 (0.93, 1.15)	0.92 (0.67, 1.25)	0.578
	Sleep medication ever use, no antihistamines (vs never)	1.10 (1.00, 1.22)	1.30 (0.92, 1.84)	1.10 (0.98, 1.23)	1.01 (0.73, 1.38)	0.573
	Sleep mask use (vs none)	1.75 (1.17, 2.62)*	4.61 (2.00, 10.62)*	1.41 (0.85, 2.35)	1.75 (0.57, 5.42)	0.232
Model 3 c
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.15 (0.82, 1.61)	1.64 (0.68, 3.96)	1.16 (0.80, 1.68)	0.45 (0.07, 3.03)	0.252
	Short, <7 hours	0.97 (0.85, 1.10)	1.17 (0.81, 1.70)	0.96 (0.84, 1.11)	0.82 (0.51, 1.31)	0.487
	Long, >9 hours	1.05 (0.73, 1.52)	0.89 (0.29, 2.74)	1.04 (0.68, 1.60)	1.35 (0.51, 3.53)	0.860
Sleep quality	Inconsistent sleep (vs consistent)	1.10 (0.92, 1.32)	0.90 (0.48, 1.68)	1.19 (0.97, 1.45)	0.75 (0.38, 1.47)	0.253
	Sleep debt ≥ 2 hours e (vs none)	1.00 (0.90, 1.10)	1.01 (0.75, 1.36)	1.00 (0.90, 1.11)	0.85 (0.53, 1.35)	0.761
	Frequent napping ≥3 times/week (vs infrequent)	1.03 (0.81, 1.31)	1.21 (0.55, 2.68)	1.08 (0.83, 1.42)	0.67 (0.31, 1.45)	0.364
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.07 (0.92, 1.25)	1.08 (0.69, 1.70)	1.12 (0.94, 1.32)	0.70 (0.38, 1.30)	0.239
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.03 (0.86, 1.24)	1.38 (0.77, 2.44)	1.00 (0.82, 1.23)	1.03 (0.59, 1.79)	0.672
	Insomnia symptoms (vs none)	1.05 (0.93, 1.18)	1.12 (0.78, 1.62)	1.07 (0.94, 1.22)	0.82 (0.54, 1.26)	0.418
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.95 (0.65, 1.38)	0.86 (0.74, 1.00)	1.00 (0.68, 1.48)	0.744
	Poor	1.11 (0.94, 1.30)	1.24 (0.80, 1.90)	1.13 (0.95, 1.36)	0.71 (0.33, 1.52)	0.325
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.03 (0.94, 1.13)	1.16 (0.85, 1.59)	1.03 (0.93, 1.14)	0.92 (0.68, 1.25)	0.601
	Sleep medication ever use, no antihistamines (vs never)	1.10 (0.99, 1.22)	1.30 (0.92, 1.84)	1.09 (0.97, 1.22)	1.01 (0.74, 1.39)	0.590
	Sleep mask use (vs none)	1.75 (1.17, 2.61)*	4.62 (2.01, 10.64)*	1.41 (0.85, 2.34)	1.76 (0.57, 5.44)	0.229

Notes: *p ≤ 0.05 and estimates bolded for significance. ^a Adjusted for age (continuous). ^b Adjusted for age (continuous), race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence. ^c Adjusted for age (continuous), race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence, self-rated health. ^d Wald p-value for interaction between sexual orientation and generation. ^e Sleep debt models additionally adjusted for inconsistent sleep. These results show the associations between sexual minoritized status and sleep in the overall population and by generation, the primary and secondary aims of this research.

Abbreviations: CI, confidence interval; PR, prevalence ratio.

Perceived Sexual Orientation Discrimination

Ever Perceived Sexual Orientation Discrimination-Sleep Associations Among Only Sexual Minoritized Women

In adjusted models, we did not find evidence of an association between perceived sexual orientation discrimination and sleep among non-heterosexual women (Table 6). Results were consistent in supplemental analyses (Table S5).

Table 6.

Adjusted ^a,b,c Prevalence Ratios (95% Confidence Intervals) for Sexual Orientation-Sleep Associations by Perceived Sexual Orientation Discrimination Strata and Perceived Sexual Orientation Discrimination-Sleep Association Among Sexual Minoritized: The Sister Study (2003–2009)

		PR (95% CI) for Sexual Minoritized (vs Heterosexuals) Women N=50,782				Ever Discrimination (vs None)
		Overall	No Discrimination	Discrimination	p-value d	Among only Sexually Minoritized N=1024
Model 1 a
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.13 (0.86, 1.48)	1.47 (1.01, 2.16)*	0.50 (0.32, 0.80)*	0.002*	0.65 (0.36, 1.15)
	Short, <7 hours	0.98 (0.88, 1.10)	1.09 (0.92, 1.28)	0.64 (0.53, 0.78)*	<0.001*	0.79 (0.62, 1.00)
	Long, >9 hours	1.19 (0.92, 1.54)	1.08 (0.70, 1.66)	0.86 (0.58, 1.30)	0.467	1.16 (0.66, 2.06)
Sleep quality	Inconsistent sleep (vs consistent)	1.08 (0.95, 1.24)	1.27 (1.05, 1.53)*	0.82 (0.66, 1.03)	0.005*	0.79 (0.60, 1.05)
	Sleep debt ≥ 2 hours e (vs none)	1.04 (0.96, 1.12)	1.04 (0.93, 1.17)	0.98 (0.87, 1.11)	0.484	1.00 (0.85, 1.17)
	Frequent napping ≥3 times/week (vs infrequent)	1.04 (0.88, 1.24)	1.11 (0.86, 1.44)	0.79 (0.59, 1.06)	0.087	0.86 (0.59, 1.24)
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.05 (0.93, 1.19)	1.07 (0.89, 1.29)	0.81 (0.66, 0.99)*	0.052*	0.94 (0.72, 1.23)
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.13 (0.98, 1.30)	1.16 (0.93, 1.43)	1.02 (0.81, 1.27)	0.422	0.98 (0.73, 1.31)
	Insomnia symptoms (vs none)	1.07 (0.98, 1.17)	1.04 (0.90, 1.21)	0.93 (0.81, 1.08)	0.307	1.03 (0.85, 1.26)
Latent sleep health (vs good)	Moderate	0.82 (0.71, 0.94)*	0.87 (0.70, 1.08)	0.57 (0.46, 0.70)*	0.009*	0.91 (0.67, 1.22)
	Poor	1.03 (0.87, 1.22)	1.37 (1.07, 1.74)*	0.48 (0.36, 0.65)*	<0.001*	0.62 (0.43, 0.89)*
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.10 (1.02, 1.19)*	1.12 (1.00, 1.26)*	0.99 (0.88, 1.12)	0.144	0.96 (0.82, 1.13)
	Sleep medication ever use, no antihistamines (vs never)	1.17 (1.07, 1.26)*	1.17 (1.04, 1.33)*	1.05 (0.92, 1.20)	0.235	0.98 (0.83, 1.17)
	Sleep mask use (vs none)	1.83 (1.31, 2.53)*	2.01 (1.24, 3.28)*	1.07 (0.61, 1.87)	0.112	0.81 (0.41, 1.62)
Model 2 b
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.16 (0.83, 1.62)	1.34 (0.82, 2.20)	0.71 (0.40, 1.29)	0.116	0.74 (0.35, 1.56)
	Short, <7 hours	0.97 (0.85, 1.11)	1.01 (0.83, 1.23)	0.77 (0.62, 0.95)*	0.061	0.87 (0.66, 1.16)
	Long, >9 hours	1.05 (0.73, 1.52)	0.70 (0.34, 1.47)	0.85 (0.50, 1.43)	0.679
Sleep quality	Inconsistent sleep (vs consistent)	1.11 (0.92, 1.33)	1.31 (1.01, 1.71)*	0.90 (0.67, 1.21)	0.069	0.81 (0.55, 1.19)
	Sleep debt ≥ 2 hours e (vs none)	1.00 (0.91, 1.10)	0.99 (0.85, 1.15)	0.91 (0.78, 1.06)	0.427	1.02 (0.83, 1.26)
	Frequent napping ≥3 times/week (vs infrequent) f	1.03 (0.81, 1.32)	1.03 (0.70, 1.51)	0.79 (0.53, 1.18)	0.365
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.07 (0.92, 1.25)	1.14 (0.90, 1.44)	0.84 (0.65, 1.08)	0.088	0.91 (0.65, 1.28)
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.04 (0.86, 1.25)	1.09 (0.82, 1.45)	0.92 (0.70, 1.22)	0.411	1.10 (0.75, 1.61)
	Insomnia symptoms (vs none)	1.05 (0.94, 1.19)	1.03 (0.85, 1.25)	0.93 (0.77, 1.11)	0.427	1.06 (0.82, 1.38)
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.87 (0.70, 1.07)	0.77 (0.62, 0.94)*	0.428	0.98 (0.73, 1.31)
	Poor	1.10 (0.93, 1.30)	1.29 (1.01, 1.64)*	0.76 (0.57, 1.01)	0.006*	0.80 (0.55, 1.17)
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.03 (0.94, 1.14)	1.06 (0.91, 1.23)	0.92 (0.80, 1.07)	0.212	0.98 (0.80, 1.20)
	Sleep medication ever use, no antihistamines (vs never)	1.10 (1.00, 1.22)	1.14 (0.98, 1.34)	0.95 (0.81, 1.12)	0.123	0.96 (0.77, 1.18)
	Sleep mask use (vs none) f	1.75 (1.17, 2.62)*	1.71 (0.89, 3.28)	0.96 (0.50, 1.86)	0.258
Model 3 c
Mean sleep duration (vs 7–9 hours)	Very short, ≤ 5 hours	1.15 (0.82, 1.61)	1.34 (0.83, 2.18)	0.76 (0.42, 1.37)	0.151	0.75 (0.35, 1.59)
	Short, <7 hours	0.97 (0.85, 1.10)	1.01 (0.83, 1.23)	0.78 (0.63, 0.96)*	0.073	0.88 (0.67, 1.17)
	Long, >9 hours f	1.05 (0.73, 1.52)	0.71 (0.34, 1.47)	0.86 (0.51, 1.45)	0.654
Sleep quality	Inconsistent sleep (vs consistent)	1.10 (0.92, 1.32)	1.30 (1.00, 1.69)	0.93 (0.69, 1.25)	0.108	0.81 (0.55, 1.20)
	Sleep debt ≥ 2 hours e (vs none)	1.00 (0.90, 1.10)	0.99 (0.85, 1.14)	0.91 (0.78, 1.06)	0.477	1.03 (0.83, 1.27)
	Frequent napping ≥3 times/week (vs infrequent) f	1.03 (0.81, 1.31)	1.02 (0.69, 1.49)	0.82 (0.55, 1.21)	0.432
	Difficulty falling asleep ≥ 30 minutes (vs no difficulty)	1.07 (0.92, 1.25)	1.13 (0.89, 1.42)	0.87 (0.68, 1.11)	0.131	0.92 (0.66, 1.29)
	Difficulty staying asleep ≥ 3 nights/week ≥ 3 times/night (vs no difficulty)	1.03 (0.86, 1.24)	1.08 (0.81, 1.42)	0.96 (0.72, 1.27)	0.568	1.13 (0.77, 1.66)
	Insomnia symptoms (vs none)	1.05 (0.93, 1.18)	1.02 (0.85, 1.23)	0.95 (0.80, 1.15)	0.605	1.08 (0.83, 1.40)
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.86 (0.70, 1.07)	0.77 (0.63, 0.94)*	0.442	0.97 (0.73, 1.30)
	Poor	1.11 (0.94, 1.30)	1.29 (1.02, 1.63)*	0.84 (0.64, 1.10)	0.018*	0.80 (0.56, 1.15)
Sleep aids	Sleep medications ever use, with antihistamines (vs never)	1.03 (0.94, 1.13)	1.05 (0.90, 1.21)	0.95 (0.82, 1.10)	0.345	0.99 (0.81, 1.20)
	Sleep medication ever use, no antihistamines (vs never)	1.10 (0.99, 1.22)	1.13 (0.97, 1.33)	0.98 (0.84, 1.15)	0.223	0.97 (0.78, 1.20)
	Sleep mask use (vs none) f	1.75 (1.17, 2.61)*	1.70 (0.89, 3.27)	0.97 (0.50, 1.88)	0.268

Notes: *p ≤ 0.05 and estimates bolded for significance. ^a Adjusted for age (continuous). ^b Adjusted for age (continuous), race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence. ^c Adjusted for age (continuous), race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence, self-rated health. ^d Wald p-value for interaction between sexual orientation and discrimination. ^e Sleep debt models additionally adjusted for inconsistent sleep. ^f Missing estimate due to model non-convergence. We sought to assess the contribution of ever perceived sexual orientation discrimination to the relationship between sexual orientation and sleep. These results show (1) the association between sexual orientation and sleep by strata of ever perceived sexual orientation discrimination (no discrimination/discrimination) among the overall population, and (2) the association between perceived sexual orientation discrimination and sleep among only sexual minoritized women.

Abbreviations: CI, confidence interval; PR, prevalence ratio.

Sexual Orientation-Sleep Association Stratified by Perceived Sexual Orientation Discrimination

In fully adjusted models, ever perceived discrimination modified the association between sexual orientation and poor sleep health (vs good; Table 6; p-value for interaction = 0.018). Among those who never reported discrimination, non-heterosexual participants had a higher prevalence of poor than good sleep health when compared to heterosexuals (1.29 [1.02, 1.63]); conversely, the prevalence of poor sleep health (vs good) among those who had ever reported discrimination was lower among non-heterosexual than heterosexual women (0.84 [0.64, 1.10]) despite wide confidence intervals. Notably, among participants who had ever perceived discrimination, non-heterosexuals had a lower prevalence of moderate sleep health than good when compared to heterosexuals (0.77 [0.63, 0.94]), yet the association was attenuated among those who had never perceived discrimination (0.86 [0.70, 1.07]). For sleep duration, among those who had ever perceived discrimination, non-heterosexual participants had a lower prevalence of short sleep (vs recommended durations) than heterosexual participants (0.78 [0.63, 0.96]); the association was not evident among those who had never perceived sexual orientation discrimination (1.01 [0.83, 1.23]). Associations were attenuated for supplemental results with participants who had perceived sexual orientation discrimination in the last five years removed (Table S5).

Sexual Orientation-Sleep Associations Mediated by Ever Perceived Sexual Orientation Discrimination

Perceived discrimination mediated the potential association (Table 7; total effect: 1.13 [0.92, 1.39]) between sexual orientation and poor sleep health among women aged <55 years (indirect effect: 0.79 [0.65, 0.99]). The indirect effect for sleep medications without antihistamines among women aged <55 years was suggestive of the same suppressive trend (0.85 [0.75, 1.02]). In supplemental analyses (Table S6), results were comparable and did not change our interpretation of these findings.

Table 7.

Adjusted ^a Prevalence Ratios (95% Confidence Intervals) for Mediation by Perceived Sexual Orientation Discrimination Between Sexual Orientation and Sleep Outcomes (Reference=heterosexuals), Overall and by Select Strata: The Sister Study (2003–2009; N=27,525)

		PR (95% CI) for Sexual Minoritized (vs Heterosexuals) Women
		Total Effect	Direct Effect	Indirect Effect
Overall
Latent sleep health (vs good)	Moderate	0.88 (0.77, 1.01)	0.88 (0.70, 1.11)	1.00 (0.83, 1.25)
	Poor	1.10 (0.93, 1.30)	1.26 (0.97, 1.63)	0.88 (0.72, 1.12)
Sleep medication ever use, with antihistamines (vs never)		1.03 (0.94, 1.14)	1.13 (0.97, 1.32)	0.91 (0.81, 1.06)
Sleep medication ever use, no antihistamines (vs never)		1.10 (1.00, 1.22)	1.25 (1.07, 1.47)*	0.88 (0.78, 1.04)
Sleep mask use (vs none) b
NH Black
Sleep medication ever use, with antihistamines (vs never)		1.38 (0.97, 1.97)	1.42 (1.02, 1.97)*	0.97 (0.71, 1.36)
Sleep medication ever use, no antihistamines (vs never)		1.51 (0.98, 2.33)	1.40 (0.97, 2.01)	1.08 (0.75, 1.58)
< 55 years
Latent sleep health (vs good)	Moderate	0.86 (0.73, 1.01)	0.82 (0.61, 1.09)	1.05 (0.83, 1.40)
	Poor	1.13 (0.92, 1.39)	1.43 (1.09, 1.86)*	0.79 (0.65, 0.99)*
Sleep medication ever use, no antihistamines (vs never)		1.16 (1.02, 1.31)*	1.36 (1.13, 1.63)*	0.85 (0.75, 1.02)
Sleep mask use (vs none) b
Generation X
Sleep mask use (vs none) b
Baby Boomers
Latent sleep health (vs good)	Moderate	0.86 (0.74, 1.01)	0.79 (0.60, 1.05)	1.09 (0.88, 1.40)
	Poor	1.13 (0.94, 1.35)	1.21 (0.91, 1.60)	0.93 (0.74, 1.25)

Notes: *p ≤ 0.05 and estimates bolded for significance. ^a Adjusted for age (continuous), race/ethnicity, marital status, educational attainment, annual household income, occupational class, nativity, region of residence. ^b Missing estimates due to model non-convergence. We sought to assess the contribution of ever perceived sexual orientation discrimination to the relationship between sexual orientation and sleep. These results show mediation by perceived sexual orientation discrimination between sexual orientation and select sleep measures among the overall population.

Abbreviations: CI, confidence interval; PR, prevalence ratio.

Discussion

We identified differences in use of sleep aids and sleep between Sister Study participants who identified as heterosexual and who identified as homosexual or bisexual, with little evidence of variation by race/ethnicity, age, and generation. Perceived sexual orientation discrimination may modify the sexual orientation-sleep health association and may also mediate the relationship with sleep health. Overall, sleep mask use was more common among non-heterosexual than heterosexual women. Among NHB women, non-heterosexuals had a higher use of sleep medications compared to heterosexuals. Among non-heterosexual compared to heterosexual women, good sleep health was more common than moderate sleep health among those who ever perceived sexual orientation discrimination, while good sleep health was less common than poor sleep among those who never perceived discrimination. Our findings illuminate that non-heterosexual women, particularly NHB women, use more sleep aids than heterosexuals, and that discriminatory experiences based on sexual orientation may differentially shape sleep.

The use of sleep aids requires initiative to seek out aids and access to them. Sleep mask use is an affordable, accessible aid behavior to block out light during sleep,⁵³ and has been shown to improve sleep quality among patients in hospital settings⁵⁴ as well as daytime alertness and memory in a healthy population.⁵³ Sleep medications can be more costly than sleep masks and are sought to treat sleep disorders and conditions that make it difficult to stay or fall asleep. Our findings align with two previous studies that found certain sexual minoritized groups use sleep medications more often than heterosexuals,^16,17 and add to a limited body of literature examining sleep masks and sexual orientation. Additional examination of the interplay between sexual orientation and sleep aid use would further elucidate our findings.

Perceived sexual orientation discrimination was a potential modifier, and also possibly a mediator of the relationship between sexual orientation and sleep health: non-heterosexual women who had ever perceived discrimination were less likely to have moderate than good sleep health compared to heterosexual women, while non-heterosexual women who had never perceived discrimination were more likely to have poor sleep health (vs good) than heterosexuals. The prevalence of heterosexuals with perceived sexual orientation discrimination in our population is small (5%). While less common, heterosexuals may perceive discrimination, for example, in social or work settings where power dynamics or cultural norms favor minoritized groups,⁵⁵ though experiences would differ given that heterosexuals still align with cultural and systemic norms. For sexual minoritized groups, to perceive discrimination can imply a level of “outness” and a community of support, which can help navigate discriminatory experiences and has been linked to improved health outcomes.^56–58 Modification by perceived sexual orientation discrimination was attenuated when controlling for socioeconomic factors like race/ethnicity and education that can shape how and when discrimination is experienced. Exploration of the intricacies of perceived sexual orientation discrimination, including detailed assessments of discrimination and coping (eg, source, timing, setting), are warranted given differing sociocultural experiences.

Contrary to our hypothesis, we did not find worse sleep outcomes among non-heterosexual women compared to heterosexual women. We also did not find that sexual minoritized individuals who intersect with minoritized racial/ethnic identities or older age groups and generations had worse sleep outcomes than their heterosexual counterparts. When exploring cumulative disadvantage, non-heterosexual NHB women had multiple worse sleep outcomes compared to NHW heterosexuals, even when adjusting for social and health factors. Age hypotheses were based on known age-related health declines and biological changes in sleep architecture that can negatively impact sleep.²⁵ While some health conditions were more common among non-heterosexual women (ie, depression, obesity) in our study, we did not see strong disparities by sexual orientation overall among health conditions like heart disease and hypertension. Moreover, our sample of adults ≥65 years was limited, and non-heterosexuals skewed younger than heterosexuals. Interrogation of the role of sleep masks and medications between sexual orientation and sleep is warranted as well as a focused assessment of cumulative disadvantage and further exploration of larger older populations.

This study had limitations. Our study design was cross-sectional which limits interpretation of causality. The prevalence of non-heterosexuals in our study population, collected from 2003–2009, was 2.2%, and more recent data estimates LGBT and LGBTQ+ prevalences between 4.5–7.6%.^2,28 Social acceptance of LGBTQ+ populations was less common then than today,^28,30 and participants may have been less forthcoming about their sexuality, which limits the generalizability of our findings. Sleep assessments were self-reported rather than objective, gold standard measures (eg, polysomnography, actigraphy), and validity can vary by race/ethnicity.^59–63 Subjective measures can introduce bias and thus limits validity and generalizability. Moreover, greater differentiation of sleep medications would have provided more context to our interpretations. We also lacked a measure of “outness” in our study which limited interpretation as outness has been linked to the health of sexual minoritized individuals.^57,58 The generalizability of results were further limited by: (1) exclusion of sexual minoritized participants who were not homosexuals or bisexuals due to sample size; we used the term non-heterosexual to not generalize to all sexual minoritized groups; (2) restriction of minoritized racial/ethnic groups to NHB and Latina women due to small sample sizes of other subgroups; and (3) the design of the Sister Study which is all female and each has a sister with breast cancer. Compared to the US population, Sister Study participants are older, and more commonly of White race, a higher socioeconomic status, and better health.³⁵

Our study also has important strengths. To our knowledge, we are the first study to examine intersections of race/ethnicity, age, and generation with sexual orientation and sleep using a multidimensional sleep measure.⁹ While existing sexual orientation literature largely explores singular sleep outcomes (eg, duration),^{12,13,15,17,26} we used latent class methods⁴⁵ to develop a multidimensional sleep health measure that elevates the current body of literature. We also examined sleep mask use, which has received little attention within general populations, and sleep medication use, which provides novel insight into treatment seeking behaviors for sleep problems among sexual minoritized groups. Our large overall sample size powered main analyses and the rich dataset allowed us to explore multiple sleep measures while adjusting for a variety of sociodemographic factors. Additionally, we examined perceived discrimination for sexual orientation, a novel and highly plausible mediating pathway between sexual orientation and sleep with a robust inverse odds weighting approach.⁵²

Conclusion

Our study provides a comprehensive, novel assessment of sexual orientation disparities in sleep; intersection with race/ethnicity, age, and generation; and the role of perceived sexual orientation discrimination among adult US women. Sleep mask use was more common among non-heterosexual than heterosexual women; among NHB women, non-heterosexual women also had higher use of sleep medications. Among those who ever perceived sexual orientation discrimination, non-heterosexual women had more good than moderate sleep health compared to heterosexual women; among those who never perceived discrimination, non-heterosexual women had more poor than good sleep health. Sleep aid use among non-heterosexuals may contribute to fewer sleep disparities by sexual orientation, and findings highlight how intersecting identities may shape sleep inequities. Future studies should investigate findings among larger, more diverse populations using longitudinal and mediation methods, objective sleep measures, and detailed discrimination assessments, with the ultimate goal of informing tailored public health efforts to improve sleep.

Funding Statement

This research was supported by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences (Z1AES103325 and Z01ES044005).

Abbreviations

CI, Confidence Interval; IOW, Inverse Odds Weighting; LCA, Latent Class Analysis; LGBTQ, Lesbian, Gay, Bisexual, Transexual, or Queer; LGBTQ+, lesbian, gay, bisexual, transgender, queer or questioning, or another diverse gender identity; NHB, non-Hispanic Black; NHW, non-Hispanic White; PR, Prevalence Ratio; US, United States.

Data Sharing Statement

All data necessary to reproduce the current analysis are publicly available following procedures described on the Sister Study website (https://sisterstudy.niehs.nih.gov/English/datarequests.htm).

Consent for Publication

The authors provide consent to publication of this manuscript and its contents.

Disclosure

The authors report no conflicts of interest in this work.