Sexual orientation-related disparities in healthcare access in three cohorts of U.S. adults

Abstract

The objective of this study was to quantify sexual orientation differences in insurance access, healthcare utilization, and unmet needs for care. We analyzed cross-sectional data from three longitudinal U.S.-based cohorts (N = 31,172) of adults ages 20–54 years in the Growing Up Today Studies 1 and 2 and the Nurses’ Health Study 3 from 2015 to 2019. Adjusted log-binomial models examined sexual orientation differences (reference: completely heterosexual) in insurance access, healthcare utilization, and unmet needs for care. Compared to completely heterosexuals, mostly heterosexual and bisexual adults were more likely to report emergency departments as a usual source of care and less likely to be privately insured. Sexual minorities (mostly heterosexual, bisexual, gay/lesbian) were also more likely than completely heterosexuals to delay needed care for reasons of not wanting to bother a healthcare provider, concerns over cost/insurance, bad prior healthcare experiences, and being unable to get an appointment. Differences by sex and sexual orientation also emerged for healthcare utilization and unmet needs. For example, mostly heterosexual women were more likely than completely heterosexual women to delay care due to perceiving symptoms as not serious enough, while gay men were less likely than lesbian women to delay for this reason. Findings indicate that sexual minorities experience disparities in unmet needs for and continuity of care. Provider education should be attentive to how perceptions, like perceived severity, can shape healthcare access in tandem with socioeconomic barriers.

1. Introduction

In the United States, sexual minority (e.g., gay, lesbian, bisexual) individuals are more likely to have worse health compared to heterosexuals, including increased prevalence of stroke, heart disease, and physical or psychological disability (Jackson et al., 2016). Employment discrimination (Sears and Mallory, 2011) and poverty (Badgett et al., 2013) place sexual minorities at increased risk for experiencing disparities in insurance access, healthcare utilization, and having unmet needs for care (Jackson et al., 2016; Buchmueller and Carpenter, 2010; Heck et al., 2006), which increases risk for adverse health outcomes. Research suggests that compared to heterosexuals, sexual minority adults have lower rates of private health insurance coverage and are more likely to be uninsured (Charlton et al., 2018; Lunn et al., 2017; Gonzales and Blewett, 2014). Similarly, sexual minorities utilize routine healthcare less frequently than heterosexuals (Buchmueller and Carpenter, 2010; Austin and Irwin, 2010), and they are more likely to report an unmet need for care due to cost (Jackson et al., 2016; Buchmueller and Carpenter, 2010; Heck et al., 2006).

To date, we have limited knowledge on other mechanisms of healthcare access inequity. For example, our understandings of insurance disparities are largely attributable to pre-Affordable Care Act (ACA) samples. The ACA, through multiple mechanisms including Medicaid expansion, has expanded insurance access for many U.S. adults (Baker et al., 2014), though emerging research suggests healthcare access disparities persist for sexual minorities with insurance coverage (Nguyen et al., 2018; Hsieh and Ruther, 2017; Skopec and Long, 2015). Thus, more post-ACA estimates on multi-dimensional metrics of healthcare access are needed in order to better understand current disparities. For example, insurance coverage stability among sexual minorities is not well-understood and could be a critical disparity area as it has been linked to high medical debt and unmet care needs (Schoen and DesRoches, 2000). Research also suggests sexual minorities are more likely to experience an unmet need for care than heterosexuals (Jackson et al., 2016; Buchmueller and Carpenter, 2010; Heck et al., 2006), but reasons for delaying care are less understood apart from socioeconomic (Jackson et al., 2016; Buchmueller and Carpenter, 2010; Heck et al., 2006; Hsieh and Ruther, 2017) and provider discrimination barriers (Macapagal et al., 2016; Hirsch et al., 2016). The Minority Stress Model (Meyer, 2003) posits that heterosexist experiences lead to worse psychological states and subsequent health outcomes for sexual minorities. From a complementary lens, the Andersen Healthcare Utilization Model (Andersen, 1995) posits that individual factors like sexual orientation can influence contextual determinants of health services use, like perceptions of care or provider availability (Bradley et al., 2002). Therefore, it is possible that discrimination from providers and other aspects of the healthcare environment may influence both social and health perceptions that can, in turn, influence healthcare utilization.

A final area in need of further research is the interaction of sex and sexual orientation in relation to healthcare access. Several national samples of U.S. adults have used sex-stratified models that suggest differences exist by both sex and sexual orientation (Jackson et al., 2016; Buchmueller and Carpenter, 2010; Heck et al., 2006; Charlton et al., 2018; Lunn et al., 2017; Gonzales and Blewett, 2014; Hsieh and Ruther, 2017; Gonzales and Henning-Smith, 2017). For example, disparities in Medicaid and other government assistance plans persist for sexual minority men when accounting for employment status, while similar disparities are attenuated by employment status for sexual minority women (Gonzales and Blewett, 2014). However, such studies have not examined the formal interaction between sex and sexual orientation on healthcare access. As heterosexual and sexual minority people are theorized to engage in different health behaviors due to differences in gendered constructions of masculinity, femininity, and health-related beliefs (Courtenay, 2000), their perceptions for having unmet needs for care may also differ. For example, masculine gender role norms pertaining to decreased help-seeking may contribute to heterosexual men having lower rates of healthcare access compared to sexual minority men, while disparities among women may be more directly driven by minority stress and contextual healthcare environment factors. Meanwhile, healthcare services targeted toward gay and bisexual men (like the Ryan White HIV/AIDS Program) (Health Resources and Services Administration, 2019) may also uniquely increase healthcare availability for many gay and bisexual men, while no similar mechanism exists for the healthcare environments of sexual minority women.

Understanding both socioeconomic and psychosocial barriers to accessing timely care will demonstrate multiple areas in which healthcare access can be improved beyond socioeconomic mechanisms alone. The proposed research addresses important gaps in the literature by examining sexual orientation-based differences in a large sample of U.S. adults. We hypothesized that 1.) Compared to heterosexuals, sexual minorities will be more likely to: use government-sponsored insurance, use emergency/urgent care services, experience an unmet need for care for any reason, and experience an unmet need for care due to past perceived barriers related to cost, insurance, or past negative healthcare experiences. Further, we hypothesized that 2.) These differences will interact with sex. For example, we expected sexual minority men to be less likely to delay care than heterosexual men, while sexual minority women will be more likely to delay care than heterosexual women.

2. Methods

2.1. Study population

Participants were selected from three cohorts of young and middle-age adults from the Growing Up Today Study (GUTS) 1 and 2 and the Nurses’ Health Study 3 (NHS3) between the follow-up period of 1996–2019 (N = 41,583) across all 50 U.S. states. GUTS and NHS3 are unique in their inclusion of both multidimensional sexual orientation and nuanced healthcare access measures. The NHS3 is an open cohort of nurses and nursing students and its questionnaires were administered in a chronological sequence beginning from the participant’s baseline date (2010–2019). In contrast, GUTS 1 and 2 cohorts receive annual questionnaires and consist of children of nurses from the Nurses’ Health 2 Study. Data on healthcare access outcomes were drawn from the most recent questionnaire (GUTS [2016]; NHS3 [Questionnaire 8; 2015–2019]). Participants were excluded from the analytic sample if they were: 1.) Missing/reported an unsure sexual orientation (n = 5918); 2.) Their gender identity differed from their sex assigned at birth (n = 75); or 3.) Missing all information on healthcare access outcomes (n = 4418). The final analytical sample consisted of 31,172 adults. This study was approved by the Brigham and Women’s Hospital Institutional Review Board.

2.2. Measures

2.2.1. Sexual orientation

The sexual orientation item was adapted from the Minnesota Adolescent Health Survey (Remafedi et al., 1992), which asks about feelings of attraction and identity with six mutually exclusive response options (completely heterosexual, mostly heterosexual, bisexual, mostly homosexual, completely homosexual, and unsure). We used report of sexual orientation from the same year as the healthcare access outcomes (GUTS 2016). NHS3, which is an open cohort with sequential questionnaires, began collecting sexual orientation data in Questionnaire 5 (2013–2019). Sexual orientation groups were modeled as: completely heterosexual (reference), mostly heterosexual, bisexual, and gay/lesbian (combination of the mostly homosexual and completely homosexual categories due to small cell sizes). If data were missing from the GUTS cohorts, we carried forward the most recent previous questionnaire response (which has been collected on every questionnaire starting in 1999 for GUTS1 and 2008 for GUTS2).

2.2.2. Healthcare access factors

2.2.2.1. Insurance access.

Type of insurance coverage, gaps in coverage, and having any insurance coverage were assessed in GUTS 2016 and NHS3 Questionnaire 8 (2015–2019). The most recent data for having any coverage were modeled using a dichotomous variable for some kind of coverage (yes; no), any lapse in coverage (yes, no); and type of coverage (private, government, single-service [plans only cover one type of service, like hospitalization], other, uninsured, not specified). If there were missing data from the any insurance coverage item in 2016 for GUTS participants, the most recent response was carried forward (2013–2015).

2.2.2.2. Healthcare utilization.

The most recent data (GUTS 2016, NHS3 Questionnaire 8 [2015–2019]) on time since last routine physical exam was modeled as < 1; 1–2; or > 2 years. Data on usual place of care was modeled as: (Jackson et al., 2016) private practice or HMO (Health Maintenance Organization) (Sears and Mallory, 2011), public clinic (community health center or public health clinic (Badgett et al., 2013), family planning clinic (e.g., Planned Parenthood, school or school-based clinic), (Buchmueller and Carpenter, 2010) urgent care center or hospital emergency room, (Heck et al., 2006) other (employer or company clinic, hospital outpatient clinic, other setting). For GUTS participants, if time since last routine physical exam was missing (2016), the most recent response was carried forward (2005–2015).

2.2.2.3. Unmet needs for care.

Experiencing an unmet need for care was assessed in GUTS 2016 and NHS Questionnaire 8 (2015–2019) using a method from Schroeder and colleagues (Schroeder et al., 2000) that asked participants about the last time they needed non-routine healthcare (e.g., sprained ankle, bad cut) but did not seek care. Participants were then asked to indicate all reasons for not receiving care: (Jackson et al., 2016) Symptoms were not serious enough; (Sears and Mallory, 2011) Did not want to bother my healthcare provider; (Badgett et al., 2013) Concern for cost or lack of insurance; (Buchmueller and Carpenter, 2010) Bad prior experience with hospitals/healthcare providers; (Heck et al., 2006) Did not believe there was anything to help me; (Charlton et al., 2018) Could not get an appointment (e.g., due to geographic distance or scheduling difficulties); and (Lunn et al., 2017) Does not apply/always gotten care. Data were modeled using a dichotomous variable for any delays (yes, no) and dichotomous variables for each of the six categories.

2.3. Statistical analyses

Data from the three cohorts were combined and demographic characteristics were reported (Table 1). We used cross-sectional multivariable regression from log-binomial models to estimate risk ratios (RR) and 95% confidence intervals (CI) for all hypotheses (Table 2). Log-Poisson regression estimation was used if a model did not converge. In order to account for sibling clusters, we estimated variance using generalized estimating equations (GEE) with a compound symmetry working correlation matrix (Fitzmaurice et al., 2004). All models were adjusted for age in years, study cohort (GUTS1, GUTS2, NHS3), and race/ethnicity (white, another race/ethnicity). To account for potential changes in insurance due to local or federal healthcare policy, the any insurance coverage analysis that used data carried forward from previous years was adjusted for questionnaire year. To examine whether sex modified associations between sexual orientation and healthcare access outcomes, models were run with interaction terms between sex and sexual orientation (Table 3; only significant interaction models reported). Analyses were conducted using SAS 9.4 (SAS Institute, Cary, NC) (SAS Statistical Software, Release 9.4, n.d.).

Table 1.

Demographic characteristics by sexual orientation in three cohorts¹ of U.S. men and women (N = 31,172) collected from 1996 to 2019.

	Completely heterosexual		Mostly heterosexual		Bisexual			Lesbian/gay
	(n = 25,308, 81%)		(n = 4434, 14%)		p4	(n = 675, 2%)		p4	(n = 755, 2%)		p4
Current age mean years (SD), range: 20–54	33.3	(7.2)	31.9	(6.2)	< 0.0001	32.0	(6.7)	< 0.0001	32.7	(7.4)	0.16
White race/ethnicity, % (n)	95	(23,807)	94	(4151)	0.12	96	(644)	0.08	93	(693)	0.007
Female sex,2% (n)	76	(19,325)	86	(3796)	< 0.0001	92	(618)	< 0.0001	64	(484)	< 0.0001
Health insurance in last year
Some kind of coverage,3% (n)	98	(22,288)	97	(4035)	< 0.0001	95	(602)	< 0.0001	96	(694)	0.0009
Any lapse in coverage, % (n)	6	(1163)	11	(411)	< 0.0001	12	(66)	< 0.0001	9	(56)	0.002
Type of plan5											< 0.0001
Private	92	(18,788)	90	(3480)	< 0.0001	86	(488)	< 0.0001	91	(594)	0.19
Government6	5	(979)	5	(213)	0.06	9	(50)	< 0.0001	5	(33)	0.78
Other6	2	(333)	2	(83)	0.02	2	(12)	0.38	2	(15)	0.19
Single-service	11	(2207)	13	(510)	< 0.0001	14	(77)	0.04	14	(91)	0.01
Not specified	1	(119)	1	(33)	–*	1	(4)	–*	< 1	(1)	–*
Uninsured/no plan7	2	(473)	3	(128)	0.0003	4	(25)	0.001	3	(22)	0.08
Healthcare use, % (n)
Time since last routine physical exam, % (n)					< 0.0001			0.32			0.04
<1	63	(15,935)	60	(2670)		60	(406)		59	(446)
1–2	23	(5763)	26	(1146)		24	(163)		24	(179)
>2	14	(3548)	14	(614)		15	(104)		17	(128)
Place of usual care,5% (n)
Private practice or HMO+	86	(17,484)	83	(3211)	< 0.0001	79	(451)	< 0.0001	82	(541)	0.01
Public clinic8	5	(1094)	12	(451)	< 0.0001	17	(97)	< 0.0001	14	(92)	< 0.0001
Urgent care or hospital emergency room	14	(2806)	17	(642)	< 0.0001	19	(107)	0.0006	14	(94)	0.67
Other8	15	(3101)	16	(607)	0.43	18	(105)	0.03	16	(103)	0.72
No usual place of care	2	(472)	3	(113)	0.02	3	(18)	0.18	3	(17)	0.64
Delayed care, % (n)
Delayed care for at least one reason, % (n)	74	(14,976)	79	(3071)	< 0.0001	82	(464)	< 0.0001	74	(483)	0.78
Reasons for delay5
Symptoms not serious enough	67	(13,576)	70	(2700)	0.0007	69	(393)	0.28	66	(429)	0.62
Avoid bothering healthcare provider	7	(1466)	8	(318)	0.03	10	(56)	0.02	10	(64)	0.01
Nothing can help	17	(3509)	24	(915)	< 0.0001	26	(147)	< 0.0001	20	(131)	0.06
Concerns over cost/insurance	15	(3054)	24	(919)	< 0.0001	27	(155)	< 0.0001	19	(123)	0.007
Bad prior healthcare experiences	3	(691)	7	(257)	< 0.0001	11	(60)	< 0.0001	8	(54)	< 0.0001
Could not get appointment	6	(1145)	9	(353)	< 0.0001	13	(74)	< 0.0001	8	(54)	0.004

¹GUTS1 (n = 11,477) participants were born 1982–1987, GUTS2 (n = 7084) 1989–1993, and NHS3 (n = 12,611) 1965–1995.

²Male sex was included only in GUTS1 and 2; NHS3 sample is all female.

³“Any coverage in past year” has been carried forward from previous years if missing data from 2016, and percentages will not total 100 when combined with “type of plan – Uninsured/no plan” variable from 2016; the following uninsured category was exclusive and coded such that no insurance plan was specified and the participant reported they were not covered in the last year (2016 only).

⁴p calculated using chi-square tests for categorical outcomes and Welch’s ANOVA (with Tukey’s honestly significant difference test as a post-hoc test to examine mean differences between subgroups) for continuous outcomes. P-values indicate comparisons between completely heterosexuals (reference group) and each sexual minority group.

⁵Participants could check all that apply for “type of plan,” “place of usual care,” and “reasons for delay;” total percentages are > 100.

⁶Government health insurance includes Children’s Health Insurance Program, Medicaid, Medicare (NHS3 only), and military plans.

⁷Uninsured/no plan includes participants who did not report coverage, or participants who reported Indian health services or single-services plans with no other coverage options selected.

⁸Public clinics include community health centers/public health clinics, family planning clinics (e.g., Planned Parenthood), school or school-based clinics. Other place of care includes employer or company clinic, hospital outpatient clinic, or “other setting”.

^*Cell sizes too small for inferential analyses.

⁺HMO is “Health Maintenance Organization.”

Table 2.

Adjusted¹ log binomial models of sexual orientation differences in healthcare utilization, reasons for having an unmet need for care, and insurance access in three cohorts of U.S. men and women (N = 31,172) collected from 1996 to 2019.

	Mostly heterosexual vs. completely heterosexual		Bisexual vs. completely heterosexual		Gay/lesbian vs. completely heterosexual
	RR2	(95% CI)2	RR2	(95% CI)2	RR2	(95% CI)2
Health insurance in last year
Some kind of coverage3	0.99	(0.98, 0.99)	0.98	(0.96, 0.99)	0.98	(0.97, 1.00)
Any lapse in coverage	1.73	(1.55, 1.93)	2.01	(1.60, 2.53)	1.35	(1.04, 1.76)
Type of plan
Private	0.98	(0.97, 0.99)	0.94	(0.91, 0.97)	1.00	(0.97, 1.02)
Government	1.13	(0.98, 1.30)	1.84	(1.40, 2.42)	0.98	(0.69, 1.39)
Single-service	1.14	(1.04, 1.25)	1.19	(0.96, 1.47)	1.23	(1.01, 1.50)
Other	1.24	(0.97, 1.57)	1.10	(0.61, 2.01)	1.20	(0.71, 2.03)
Healthcare use
Over 1 year since last routine physical exam3	1.11	(1.07, 1.16)	1.15	(1.04, 1.26)	1.01	(0.93, 1.11)
Place of usual care
Private practice or HMO+	0.96	(0.94, 0.97)	0.91	(0.87, 0.95)	0.97	(0.94, 1.01)
Public clinic4	1.92	(1.72, 2.13)	2.72	(2.25, 3.30)	2.32	(1.90, 2.82)
Urgent care/hospital emergency room	1.19	(1.10, 1.29)	1.36	(1.15, 1.62)	1.04	(0.86, 1.25)
Other	1.09	(1.00, 1.18)	1.32	(1.11, 1.57)	1.03	(0.86, 1.23)
Delayed care
Delayed care for at least 1 reason3	1.09	(1.07, 1.11)	1.11	(1.07, 1.16)	1.01	(0.96, 1.05)
Reasons for delay
Symptoms not serious enough	1.06	(1.03, 1.08)	1.05	(0.99, 1.11)	0.99	(0.93, 1.04)
Avoid bothering healthcare provider	1.17	(1.04, 1.31)	1.40	(1.08, 1.80)	1.46	(1.15, 1.86)
Nothing can help	1.33	(1.25, 1.42)	1.48	(1.28, 1.71)	1.14	(0.97, 1.33)
Concerns over cost/insurance	1.48	(1.38, 1.58)	1.75	(1.52, 2.00)	1.19	(1.01, 1.41)
Bad prior healthcare experiences	1.82	(1.58, 2.10)	2.96	(2.31, 3.81)	2.31	(1.76, 3.03)
Could not get appointment	1.59	(1.42, 1.79)	2.23	(1.79, 2.78)	1.55	(1.19, 2.01)

¹Models adjusted for sex, race/ethnicity, study cohort, and age; “some kind of coverage” model also adjusted for questionnaire year; reference group is completely heterosexuals.

²RR is “risk ratio,” CI is “confidence interval.”

³Sensitivity analyses were performed for the “over 1 year since last routine physical exam” and “some kind of coverage” models with outcome variables that were not imputed; only small coefficient discrepancies emerged between the imputed and non-imputed models. Another sensitivity analysis was conducted by adding “any insurance coverage” as a covariate in the healthcare utilization and unmet needs for care models, but associations were not attenuated.

⁴The “public clinic” model used Poisson distribution estimation.

⁺HMO is “Health Maintenance Organization.”

Table 3.

Adjusted¹ log binomial models of the interaction between male sex and sexual orientation differences in usual place of care and unmet needs for care in three cohorts of U.S. men and women (N = 31,172) collected from 1996 to 2019.²

	Over 1 year since last exam		Delayed care for at least 1 reason		Delayed care since symptoms not serious enough		Public clinic as place of usual care		Ever lapse in insurance coverage
	RR3	(95% CI)3	RR3	(95% CI)3	RR3	(95% CI)3	RR3	(95% CI)3	RR3	(95% CI)3
Male (ref: female)	1.69	(1.51, 1.77)	1.10	(1.08, 1.13)	1.12	(1.09, 1.15)	1.01	(0.88, 1.16)	1.16	(1.01, 1.33)
Mostly heterosexual	1.16	(1.11, 1.22)	1.10	(1.07, 1.12)	1.07	(1.04, 1.09)	2.00	(1.78, 2.24)	1.76	(1.56, 1.99)
Bisexual	1.10	(1.08, 1.34)	1.12	(1.07, 1.17)	1.05	(0.99, 1.12)	2.85	(2.34, 3.48)	2.12	(1.67, 2.69)
Gay/lesbian	1.15	(1.02, 1.31)	1.08	(1.03, 1.14)	1.06	(0.99, 1.12)	1.74	(1.28, 2.36)	1.68	(1.23, 2.29)
Mostly heterosexual * male	0.87	(0.79, 0.95)	0.96	(0.91, 1.01)	0.95	(0.89, 1.01)	0.79	(0.60, 1.04)	0.92	(0.69, 1.21)
Bisexual * male	0.84	(0.65, 1.11)	0.87	(0.71, 1.07)	1.00	(0.81, 1.24)	0.55	(0.24, 1.24)	0.50	(0.16, 1.52)
Gay/lesbian * male	0.78	(0.65, 0.93)	0.79	(0.70, 0.88)	0.81	(0.71, 0.92)	1.74	(1.17, 2.58	0.54	(0.30, 0.96)

¹Models adjusted for race/ethnicity, study cohort, and age.

²Outcomes only included in table for models that contained a significant interaction between sex and sexual orientation. All other outcomes did not contain a significant interaction (ps > 0.05). Due to model convergence issues due to small sample size distribution, the “bad prior experiences” model was not tested with an interaction term, thus its omission does not represent lack of significance.

³Reference group is completely heterosexuals and female sex; RR is “risk ratio,” CI is “confidence interval.”

3. Results

Significant differences emerged in the models adjusted for sex that examined the main effect of sexual orientation on healthcare access (Table 2). Completely heterosexuals were more likely than all sexual minority groups to report some insurance coverage in the past year. Among types of insurance coverage, both mostly heterosexual and bisexual participants were less likely than completely heterosexuals to be enrolled in a private insurance plan. Finally, compared to completely heterosexuals, bisexual individuals were more likely to be enrolled in a government insurance plan, and both gay/lesbian and mostly heterosexual individuals were more likely to be enrolled in a single-service insurance plan.

Both mostly heterosexual and bisexual individuals were less likely than completely heterosexuals to have a private practitioner as their place of usual care, and they were more likely to have emergency departments/urgent care settings as their place of usual care. Mostly heterosexual and bisexual individuals were also more likely than completely heterosexual to have an “other” setting as their usual place of care.

All sexual minority groups were more likely than completely heterosexuals to delay care due to concerns over cost/insurance, previous bad healthcare experiences, being unable to get an appointment, and in order to avoid bothering a provider. Finally, mostly heterosexual and bisexual individuals were more likely than completely heterosexuals to delay care due to believing nothing can help.

Next, interaction terms between sexual orientation and sex were added to healthcare access models, and significant models were reported in Table 3. Examination of predicted probabilities showed that mostly heterosexual, bisexual, and gay/lesbian women were more likely than completely heterosexual women to have over 1 year since their last routine physical exam (Fig. 1), while sexual minority men did not significantly differ from completely heterosexual men. Additionally, all women were less likely than their male counterparts to have more than 1 year since their last routine physical exam.

Fig. 1.

Bar chart of predicted probabilities of the interaction between sexual orientation and sex on likelihood of having over 1 year since last routine exam in a sample of three longitudinal cohorts of U.S. adults (N = 31,172) collected from 1996 to 2019.

There was also a significant interaction of sexual orientation and sex in delaying care for at least one reason. Gay men were less likely than completely heterosexual men to delay care for at least one reason, while gay/lesbian women were more likely than completely heterosexual women to delay. Further, completely heterosexual women were less likely than completely heterosexual men to delay, while gay men were less likely to delay than gay/lesbian women.

Delaying care due to perceiving symptoms as not serious enough followed a similar pattern, such that gay men were less likely than completely heterosexual men to delay care due symptoms not being serious enough, while mostly heterosexual women were more likely than completely heterosexual women to delay for this reason. Finally, completely heterosexual men were more likely to delay than completely heterosexual women, but gay/lesbian women were actually more likely than gay men to delay due to perceptions of severity.

Interactions also emerged for whether or not someone used a public clinic for routine care. When examining predicted probabilities, gay men were 3 times as likely as completely heterosexual men and 2.5 times as likely as gay/lesbian women to use a public clinic for routine care. Sexual minority women were also more 2–3 times more likely than completely heterosexual women to use a public clinic. Other differences among subgroups were much smaller, with completely heterosexual women slightly less likely than men to use a public clinic.

Finally, there was a significant interaction between sexual orientation and sex on having any lapse in insurance coverage. All sexual minority women were more likely to report a lapse in coverage than completely heterosexual women, and mostly heterosexual men were slightly more likely than completely heterosexual men to report a lapse. Finally, completely heterosexual women were less likely to report a lapse than men.

4. Discussion

Sexual minorities are at great risk for both acute and chronic adverse health outcomes (Jackson et al., 2016; Operario et al., 2015). In order to improve health equity, it is critical to improve healthcare access by identifying deficits and barriers in healthcare utilization, insurance access, and unmet needs for care. Within this study, differences in routine healthcare utilization and delaying needed care emerged related to both sex and sexual orientation, where gay men and completely heterosexual women had more positive outcomes, and sexual minority women and heterosexual men worse outcomes. Further, among both men and women, many sexual minorities were more likely than completely heterosexuals to be uninsured, use emergency departments/urgent care for usual care, and delay care due to a variety of reasons ranging from socioeconomic constraints to perceptions about their own health or anticipatory interactions with providers. This study is primarily novel in its inclusion of multiple sexual orientation identity categories (like mostly heterosexual) in investigating aspects of healthcare utilization not previously studied, like unmet needs for care pertaining to patient perceptions or expectations.

Sexual minorities’ increased risk of foregoing annual routine physical exams, delaying care due to financial barriers, and being uninsured is consistent with extant research from both pre- (Charlton et al., 2018; Gonzales and Blewett, 2014) and post-ACA (Lunn et al., 2017; Nguyen et al., 2018; Hsieh and Ruther, 2017; Skopec and Long, 2015; Gonzales and Henning-Smith, 2017) samples of U.S. adults, which indicates healthcare access disparities are enduring and remain a critical area of public health attention. Interestingly, though lesbian women were more likely than completely heterosexual women to have missed their annual routine physical, they were still less likely than gay men to miss this type of care. This finding is in contrast to a previous community convenience sample of sexual and gender minority adults where those assigned male at birth were more likely than those assigned female at birth to report easier access to healthcare (Macapagal et al., 2016). The current study also extends upon past findings by newly identifying insurance disparities related to sexual minority women’s increased risk of experiencing an insurance gap, which places them at risk for greater financial and subsequent health vulnerabilities in the event of health emergency (Schoen and DesRoches, 2000). These vulnerabilities are likely exacerbated by other disparities found in this study, including a multitude of reasons for delaying care when experiencing an acute injury or other non-routine need for medical attention.

Strikingly, sexual minorities were 2–3 times more likely than completely heterosexuals to delay care due to past negative healthcare experiences, which is in line with other research on provider discrimination and healthcare access (Macapagal et al., 2016; Hirsch et al., 2016). It is important to note that our measure only examines the likelihood of delaying care, but not the frequency, type, or timing of past negative healthcare encounters. However, what our results indicate is that sexual minorities are much more likely to delay needed care due to past negative experiences, and steps should be taken by medical professionals to be more affirming of sexual minority patients. Also, of particular note is the novel finding of certain sexual minorities being more likely than completely heterosexuals to delay care due to perceptions of their own health condition’s severity or potential to be helped, as well as factors related to patient-provider relationships, like not wanting to bother a provider. Extant research has primarily found that sexual minorities are more likely than heterosexuals to delay care due to concerns over cost (Jackson et al., 2016; Buchmueller and Carpenter, 2010; Heck et al., 2006; Hsieh and Ruther, 2017), though socioeconomic barriers have failed to fully explain reduced healthcare access among sexual minorities (Charlton et al., 2018; Gonzales and Blewett, 2014; Nguyen et al., 2018; Hsieh and Ruther, 2017; Skopec and Long, 2015; Blosnich, 2017). Thus, it is clear a variety of psychosocial mechanisms in addition to socioeconomic barriers play a role in sexual minorities’ unmet needs for care. Our study also indicates that such mechanisms differ by sex: delaying care due to perceptions that a health concern is not serious enough was more likely for mostly heterosexual women than completely heterosexual women, while gay men were less likely than lesbians to delay for this reason. Similarly, the finding that gay men used public clinics more than both completely heterosexual men and lesbians is novel, as past studies have primarily focused on general routine healthcare access disparities (Jackson et al., 2016; Charlton et al., 2018; Lunn et al., 2017) or emergency department use (Hsieh and Ruther, 2017).

Sex- and sexual orientation-based differences in care utilization are likely explained by several mechanisms not assessed in our study. From the lenses of the Minority Stress Model (Meyer, 2003) and Andersen Model of Healthcare Utilization (Andersen, 1995), one potential explanation is that aspects of provider-patient encounters like reactions to sexual orientation disclosure (Hirsch et al., 2016), discrimination (Li et al., 2015), and difficulty obtaining specialist appointments (Clift and Kirby, 2012) predispose sexual minorities to delay care until symptoms are severe enough to warrant emergency services. Theorized differences in health awareness and consciousness of physical health on the basis of structural oppression and gender role ideology between gay and heterosexual men as well as men and women (Courtenay, 2000) potentially contribute to these patterns. Similarly, sex-specific patterns in sexual minorities’ public clinic use may be related to differences in healthcare needs: gay men may be specifically attending HIV-focused clinics served by programs like the Ryan White HIV/AIDS Program (Health Resources and Services Administration, 2019) at higher rates, while lesbian women are seeking lesbian-affirmative care at either sexual minority-focused clinics or private practitioners (Saulnier, 2002). Additionally, bisexuals’ and mostly heterosexuals’ decreased likelihood of having private insurance, decreased use of private healthcare sources, and increased use of emergency departments/urgent care centers, combined with the absence of such differences between completely heterosexual and lesbian adults, indicates that healthcare utilization and insurance access disparities are not homogenous for sexual minorities. Bisexual stigma and invisibility (Ebin, 2012) are likely unique factors that modify the healthcare access experiences of sexual minorities with attractions to multiple genders. Alternatively, as adults in unmarried other-sex relationships are less likely to be privately insured than those in same-sex relationships (Gonzales and Blewett, 2014), interactions by relationship status and current legal partner’s sex may be partially attributable to these patterns.

4.1. Limitations and future directions

First, data were primarily limited to one questionnaire year, thus analyses are cross-sectional and cannot determine causality, and we were unable to conduct pre-/post-ACA analyses. Future research should examine longitudinal patterns in healthcare access by sexual orientation. Second, the study used a convenience sample of nurses, nursing students, and children of nurses and is not fully generalizable to the U.S. population in regards to employment, education, socioeconomic status, and potentially greater healthcare access. Third, the sample is primarily white and analyses lacked appropriate statistical power to examine the intersection of race/ethnicity and sexual orientation, thus this paper’s scope focused on general healthcare access disparities. Similarly, data were insufficient to examine the intersection of gender identity and sexual orientation. Future research should prioritize intersectional designs. Fourth, future studies should also include both legal same-sex partnership status and sexual orientation identity measures in the assessment of healthcare access disparities. Though we found unique differences in healthcare access based on sexual orientation and sex, we did not have the data to contrast this with present legal partner status and sex (e.g., marriage/civil union). This limitation is shared in other research on healthcare access disparities, which instead assesses sexual orientation through same-sex partnership (Buchmueller and Carpenter, 2010; Heck et al., 2006; Gonzales and Blewett, 2014; Clift and Kirby, 2012). As a result, we were unable to examine whether same-sex legal partnership attenuated certain insurance access disparities in our models, which only used self-reported sexual orientation identity. Fifth, insurance rates were much higher in our sample compared to rates found in other studies, thus results may not be generalizable to all sexual minorities. Finally, due to limited questionnaire space, private insurance options were not disaggregated into type of plan (employer-sponsored, individually purchased), thus we were unable to examine differences in type of private insurance plan coverage.

5. Conclusions

Results indicate that improving the healthcare access experiences of sexual minorities must occur along several different trajectories. First and key is the prioritization of improving the healthcare experiences of sexual minorities within healthcare environments. Improving provider and staff cultural competency and humility (i.e., the iterative process of self-reflection and self-critique of multicultural communication practices with patients) (Tervalon and Murray-García, 1998) would greatly reduce one of the largest barriers to care that we identified in this study: having a past adverse healthcare experience. Adverse healthcare experiences can also reduce sexual minorities’ healthcare utilization (Li et al., 2015). Any reduction in delays due to these experiences would likely have a cascading effect and thus enable sexual minorities to utilize more affordable care. Second, mostly heterosexual and bisexual adults were both more likely to rely on emergency department, urgent care services, and government insurance plans, and mostly heterosexual and bisexual in particular experienced instability in annual insurance coverage and socioeconomic-related reasons for delaying care. These disparities indicate mostly heterosexual and bisexual adults are important populations for socioeconomic-based healthcare interventions that are sensitive to bisexual stigma in healthcare systems. Cultural competency trainings of social workers should make sure to include sexual minority competencies in training and acknowledge sexual orientation-specific barriers to care when working with any patients in general healthcare settings. Finally, given the sex- and sexual orientation-based disparities identified, mostly heterosexual, bisexual, and gay/lesbian men and women must be acknowledged by healthcare policymakers and practitioners as having different healthcare barriers and addressed as unique sub-populations within sexual minority public health research and practice.