Barriers to Health Care Among Adults With Minoritized Identities in the United States, 2013–2017

Abstract

Objectives. To examine the relationship between minoritized identity and barriers to health care in the United States.

Methods. Nationally representative data collected from the 2013 to 2017 waves of the National Health Interview Survey were used to conduct descriptive and logistic regression analyses. Men and women were placed in 1 of 4 categories: no minoritized identities, minoritized identities of race/ethnicity (MIoRE), minoritized identities of sexuality (MIoS), or minoritized identities of both race/ethnicity and sexuality (MIoRES). Five barriers to health care were considered.

Results. Relative to heterosexual White adults and after controlling for socioeconomic status, adults with MIoRE were less likely to report barriers, adults with MIoS were more likely to report barriers, and adults with MIoRES were more likely to report barriers across 2 of the study measures.

Conclusions. Barriers to care varied according to gender, minoritized identity, and the measure of access to health care itself.

Public Health Implications. Approaching health disparities research using an intersectional lens moves the discussion from examining individual differences to examining the role of social structures such as the health care system in maintaining and reproducing inequality.

A significant body of health disparities literature suggests pervasive differences in the health of adults in the United States. In general, adults with minoritized racial, ethnic, and sexual identities are more likely to report poor health outcomes than are heterosexual non-Hispanic White adults.¹ Also, an extensive body of literature suggests that adults with minoritized identities are more likely to report barriers to health care than are their nonminoritized counterparts.^2,3 However, the reproduction of social hierarchies and structural policies and practices exacerbates disparities in access as a result of the norms of care developed around and delivered to traditional dominant groups (e.g., men, non-Hispanic Whites, and individuals of high socioeconomic status [SES]). Taking this historical and normative perspective into account in assessing disparities in health care access changes the focus from one of differences to one of inequalities, which are best addressed via an intersectional perspective.

Social identity is multifaceted, situated in historical and contemporary sociopolitical contexts and composed of several relational dimensions including but not limited to race, ethnicity, sexuality, ability, religious affiliation, nativity, gender, gender expression, and class.⁴ According to Zinn and Dill, these dimensions are “components of both social structure and social interaction.”^5(p327) Although race/ethnicity and sexuality are not the only dimensions of identity that influence an individual’s experience with the health care system, no study to our knowledge has quantitatively examined barriers to health care among both racial/ethnic and sexual minorities (multiple marginalization⁶).

To address this gap, we examined barriers to health care among adults with varying self-reported minoritized racial/ethnic and sexual social identities using an intersectionality perspective. The research question guiding the analysis was as follows: are adults with minoritized racial/ethnic and sexual identities more likely to report barriers to health care than are adults with nonminoritized racial/ethnic and sexual identities? On the basis of the literature just discussed, we hypothesized that net of controls, adults with minoritized identities would be more likely to report barriers to health care than would heterosexual White adults because of the ways in which structural racism and homophobia reinforce a social structure that limits and excludes some adults with minoritized identities from the health care system.

INTERSECTIONALITY IN PUBLIC HEALTH RESEARCH

Intersectionality, although unnamed until the 1980s, spans decades of academic and nonacademic research in Black feminism and race, class, and gender studies centered around social justice and addressing inequality.^7–9 Intersectionality describes the intersection of individual-level identity categories with interlocking structural-level systems of privilege and oppression that create new and unique experiences^5,8; however, this relational nature of intersectionality as a framework requires a careful exploration into the interconnectedness of power, social hierarchies, and their reproducibility.⁹

Bauer suggested that intersectionality can be used in population health research to advance health equity further when researchers thoughtfully design intersectional studies.⁶ This includes examination of not only marginalization but also privilege and the distinctions between social identities, positions, processes, policies, and historical context and their independent or joint impact on health. Bauer described social identity as an individual’s “personally held identity” (e.g., African American), social position as the “position one occupies in society” or “the way one is perceived and treated by others,” and social processes or policies as processes (e.g., racism and homophobia) or policies “that may generate, amplify or temper inequalities between groups.”^6(pp12–13)

Considering Bauer’s recommendations, we chose to operationalize intersectionality using an intercategorical complexity approach wherein the use of identity categories was purposeful and provisional.^4,6 We also decided to focus on social identity rather than social position because of the self-reported identity categories we used to construct our independent variable.

METHODS

We derived our data from the 2013 through 2017 waves of the National Health Interview Survey (NHIS), available from the IPUMS NHIS database.¹⁰ The NHIS is a cross-sectional, nationally representative survey conducted annually since 1957; a multistage area probability sampling design is used to monitor the health of the noninstitutionalized adult population of the United States.¹¹

Our variables, selected from the sample adult core, were measured in the same manner in the 2013 to 2017 NHIS waves; response rate for these waves (which included 164 696 sample adults) ranged from 53.0% to 61.2%.¹¹ Several exclusion criteria were used to create our final analytic sample. First, to construct the independent variable, minoritized identity, we excluded adults with missing information on sexual orientation (n = 4238). Second, we excluded adults whose race was reported as American Indian/Alaska Native, Asian, multiple races, or race group not releasable (n = 13 231) owing to large racial, ethnic, and socioeconomic heterogeneity among the excluded adults and our inability (because of the limited sample of adults with minoritized race/ethnicity and sexuality identities) to subset our data.

Third, we excluded adults 65 years or older (n = 37 514) as a result of factors associated with access to care in this age group (e.g., Medicare eligibility).³ Finally, adults with missing information on any of the dependent variables or covariates were excluded (n = 2636), resulting in a final analytic sample of 107 077 adults between 18 and 64 years of age (57 840 women and 49 237 men).

Measures

Barriers to health care during the preceding 12 months.

Dahlhamer et al. used 5 dichotomous measures to examine supply and demand barriers to health care among sexual minorities in the United States³; thus, we took a similar approach. Our barriers to health care measures were as follows: delayed or did not receive medical care as a result of cost, did not receive 1 or more specific services because of cost, delayed care for noncost reasons, trouble finding a provider, and no usual place of care. Dahlhamer et al. provided detailed descriptions of the questions used to construct each measure.³

We used self-reported racial/ethnic and sexual identity to construct our independent variable (minoritized identity). First, both dimensions of social identity were dichotomized (race/ethnicity as non-Hispanic White vs non-Hispanic Black and Hispanic and sexual orientation as heterosexual vs lesbian/gay, bisexual, something else, and do not know). The NHIS measures only 1 dimension of sexuality, self-identification, or sexual orientation. Measures of sexual behavior and sexual preference are not included.¹² Second, adults were placed in 1 of 4 categories: (1) adults who did not report minoritized identities of race/ethnicity or sexuality (reference, White, and heterosexual), (2) adults who reported minoritized identities of race/ethnicity and nonminoritized identities of sexuality (MIoRE; non-White and heterosexual), (3) adults who reported minoritized identities of sexuality and nonminoritized identities of race/ethnicity (MIoS; White and nonheterosexual),¹³ and (4) adults who reported minoritized identities of race/ethnicity and sexuality (MIoRES; non-White and nonheterosexual).

We aimed to use an intersectionality perspective to understand not only differences in access to health care but also the role of inequality and systems of power in reproducing these differences. To achieve this goal, we started by conceptualizing and operationalizing our independent variable using an intersectional lens. First, note that it is not identity itself that is a risk factor for experiencing barriers to health care. Identity categories such as race and sexuality can be used as markers to identify adults at risk for social structural exposure to racist and homophobic policies and practices that negatively affect access to health care and reproduce inequality.¹⁴ It is imperative to note that the health care system is a social institution, and the practices of health care professionals contribute to social structures that maintain and reproduce inequality.

Second, intersectionality suggests that identity is relational and focusing on only 1 axis is limiting. We chose to focus specifically on race/ethnicity and sexuality because of the limited literature. However, we acknowledge that these are not the only dimensions in which inequality can occur. Specifically, we identified gender and class as dimensions that contribute significantly to differences in access to health care. To account for gender differences, we stratified our analysis by gender. However, note that all women in our analysis, regardless of the group into which they were categorized, can experience minoritization based on gender.

We also included class by controlling for SES. We chose to control for SES and not include it as part of the independent variable because of our limited sample of adults with MIoRES. This limited sample also restricted our ability to subset our data on the basis of race/ethnicity or sexual orientation, and therefore we combined heterogeneous groups (e.g., African American vs Latinx and lesbian vs bisexual) under 1 homogeneous label. We intentionally set the reference category as White and heterosexual to situate our findings in the broader health disparities literature.

Covariates.

Several covariate characteristics identified as potential confounders in the health disparities literature¹⁵ were included and categorized into 3 groups. Sociodemographic characteristics were age, marital status, and region. Health characteristics were self-reported health, activity limitations, and serious psychological distress (assessed with the Kessler 6 nonspecific distress scale, described in detail elsewhere¹⁶). Socioeconomic characteristics (4 proxies for SES) were educational attainment, housing tenure, health insurance, and employment status in the preceding 12 months. Survey year was also included to account for potential period effects.

Statistical Analyses

We used the χ² test to examine bivariate associations between the covariates and the dependent and independent variables. We conducted gender-stratified analyses as a result of the significant differences in barriers to care by gender, whereby women were significantly more likely to report barriers. In our multivariable analysis, 3 logistic regression models were estimated for each of the 5 barriers to health care measures. The goal of this analytic approach was to tease out how SES affects barriers to care and acknowledge that SES profiles for racial/ethnic minorities are an important confounder. Among men, models for trouble finding a provider were not included owing to the small sample size with MIoRES and, therefore, unstable estimates. The analyses were conducted in Stata version 16 (StataCorp LLC, College Station, TX), with weights applied to adjust for the sampling design and provide representative statistics for US residents 18 to 64 years of age.

RESULTS

Tables A and B (available as supplements to the online version of this article at http://www.ajph.org) present the weighted percentages of women and men across minoritized identity. Among women, 64.7% reported no minoritized identities of race/ethnicity or sexuality, 31.2% reported MIoRE, 2.8% reported MIoS, and 1.3% reported MIoRES. Similarly, among men, 66.1% reported no minoritized identities, 30.6% reported MIoRE, 2.2% reported MIoS, and 1.2% reported MIoRES.

Table 1 presents the distribution of barriers to health care across minoritized identities for women and men. Higher percentages of women with MIoS and MIoRES than heterosexual White women and women with MIoRE reported barriers to care across all 5 measures. In addition, higher percentages of women with MIoRE, MIoS, and MIoRES reported no usual place of care than did heterosexual White women.

TABLE 1—

Barriers to Health Care Among Adults in the United States: National Health Interview Survey, 2013–2017

	Weighted %a
	No MIoRES	MIoRE	MIoS	MIoRES
Women
Delayed/did not receive care because of cost	12.3	13.6	20.2	17.9
Did not receive specific services because of cost	19.7	23.8	28.0	28.7
Delayed care for noncost reasons	12.0	13.7	17.3	20.4
Trouble finding provider	7.0	6.5	10.9	12.4
No usual source of care	10.4	16.4	16.8	26.4
Men
Delayed/did not receive care because of cost	10.0	11.5	14.7	12.9
Did not receive specific services because of cost	13.5	17.8	20.7	23.0
Delayed care for noncost reasons	8.0	9.6	14.4	22.1
No usual source of care	18.3	29.0	15.9	26.2

Note. MIoRE = minoritized identities of race/ethnicity; MIoRES = minoritized identities of race/ethnicity and sexuality; MIoS = minoritized identities of sexuality. Women, n = 57 840; men, n = 49 237. Percentages may not sum to 100.0 as a result of rounding.

^aStatistically significant differences across categories of minoritized identity (P < .001).

Higher percentages of men with MIoS and MIoRES reported 4 of the 5 barriers (delaying or not receiving care because of cost, not receiving specific services because of cost, delaying care for noncost reasons, and trouble finding a provider) than did heterosexual White men and men with MIoRE. However, relative to men with minoritized sexual identities, higher percentages of heterosexual White men, men with MIoRE, and men with MIoRES reported no usual source of care. Interestingly, adults with minoritized racial/ethnicity identities (MIoRE and MIoRES) were more likely than were heterosexual White adults and adults with MIoS (White and nonheterosexual) to report lower levels of SES.

Table 2 presents the results of the logistic regression models for women and men separately after controlling for minoritized identity, survey year, and all of the covariates (model 3). Only the odds ratios for the independent variable, minoritized identity, are shown in Table 2. Table C (available as a supplement to the online version of this article at http://www.ajph.org) provides minoritized identity odds ratios for the 3 models for men and women separately. Before controlling for SES (models 1 and 2), adults with minoritized identities were generally more likely to report barriers to care. After controlling for SES (model 3), adults with MIoRE were generally less likely to report barriers to care, adults with MIoS were more likely to report barriers to care, and adults with MIoRES were more likely to report barriers across 2 measures only than were heterosexual White adults. Delaying care for noncost reasons was the only measure in which all adults with minoritized identities (MIoRE, MIoS, and MIoRES) were more likely than were heterosexual White adults to report a barrier across each of the 3 models.

TABLE 2—

Logistic Regression Model Results for Barriers to Health Care Among Adults in the United States: National Health Interview Survey, 2013–2017

	Women, AOR (95% CI)	Men, AOR (95% CI)
Delayed/did not receive care because of cost
No MIoRES (Ref)	1	1
MIoRE	0.7 (0.7, 0.8)	0.7 (0.7, 0.8)
MIoS	1.4 (1.2, 1.7)	1.3 (1.1, 1.6)
MIoRES	0.9 (0.7, 1.2)	0.8 (0.6, 1.2)
Did not receive specific services because of cost
No MIoRES (Ref)	1	1
MIoRE	0.9 (0.8, 0.9)	0.9 (0.9, 1.0)
MIoS	1.3 (1.1, 1.5)	1.5 (1.2, 1.8)
MIoRES	1.0 (0.8, 1.2)	1.3 (1.0, 1.8)
Delayed care for noncost reasons
No MIoRES (Ref)	1	1
MIoRE	1.2 (1.1, 1.2)	1.3 (1.1, 1.4)
MIoS	1.2 (1.0, 1.5)	1.6 (1.3, 2.0)
MIoRES	1.6 (1.2, 2.0)	3.0 (2.1, 4.2)
Trouble finding provider
No MIoRES (Ref)	1	. . .a
MIoRE	0.8 (0.7, 0.9)	. . .a
MIoS	1.3 (1.0, 1.6)	. . .a
MIoRES	1.4 (1.0, 2.0)	. . .a
No usual source of care
No MIoRES (Ref)	1	1
MIoRE	0.9 (0.8, 0.9)	1.0 (0.9, 1.1)
MIoS	1.3 (1.1, 1.6)	0.7 (0.6, 0.9)
MIoRES	1.5 (1.2, 2.0)	0.9 (0.6, 1.2)

Note. AOR = adjusted odds ratio; CI = confidence interval; MIoRE = minoritized identities of race/ethnicity; MIoRES = minoritized identities of race/ethnicity and sexuality; MIoS = minoritized identities of sexuality. Women, n = 57 840; men, n = 49 237. Models included minoritized identity, survey year, sociodemographic characteristics, health characteristics, and socioeconomic characteristics.

^aEstimates are not provided because of the small sample of men with MIoRES.

DISCUSSION

Building on the Dahlhamer et al. study³ and using data from the 2013 through 2017 waves of the NHIS, we examined the relationship between 2 dimensions of social identity using an intersectional framework and barriers to health care among US adults 18 to 64 years of age. Three broad patterns emerged. Relative to heterosexual White adults and after controlling for SES, adults with MIoRE were less likely to experience barriers to care, adults with MIoS were more likely to experience barriers, and adults with MIoRES were both more and less likely to report barriers depending on the specific measure. Therefore, our hypothesis that, net of controls, adults with minoritized identities would be more likely to report barriers to health care than adults with no minoritized identities was partially supported. Barriers to care varied by gender, minoritized identity, and the measure of access to health care itself.

Our results suggest differences in barriers to health care across categories of minoritized identity. How does intersectionality move the discussion past identifying differences and toward examining inequality? We began our study by conceptualizing our independent variable, minoritized identity, using an intersectional lens. We noted that identity itself is not a risk factor for experiencing barriers to care but rather a marker for potential detrimental exposures to inequality through racist, homophobic, and sexist policies and practices experienced in not only the health care system but other social institutions as well.

Consider, for example, men and women with minoritized sexual identities. The finding that these men and women were more likely to report barriers to health care net of SES controls was not unexpected and was supported by previous studies.^3,17–19 For example, Heck et al. suggested that barriers to health care may be amplified by structural discrimination experienced among women in same-sex relationships.¹⁷ It is important to note that all women, regardless of race/ethnicity and sexuality, can experience minoritization based on gender. For example, women may experience gender discrimination when interacting with the health care system because of unconscious forms of bias such as stereotyping and moral rationing among physicians.²⁰

It was surprising to find that men with MIoS were less likely to report no usual source of care than were heterosexual White men and men with MIoRE and MIoRES. In this instance, minoritized sexual identity among non-Hispanic white men was protective. Heck et al. suggested that the HIV epidemic may have had a role in increasing the frequency of interaction between health care providers and some men in same-sex relationships.¹⁷ The need for HIV testing and care suggests that men with MIoS may be more likely to have a regular provider¹⁷ and therefore less likely to report no usual source of care. Although not possible with the NHIS data, controlling for HIV status may help further explain this association. Targeted initiatives with a focus on diversity and inclusion in health care professions would likely help minoritized adults establish a usual source of care.¹⁷

The finding that adults with both minoritized identities were more likely to report barriers, but only for 2 measures among women and 1 measure among men, also was unexpected. Adults with MIoRE, MIoS, and MIoRES were all more likely than heterosexual White adults to report delaying care for noncost reasons. This finding certainly warrants more research on the related intersections of race/ethnicity, sexuality, gender, and social position²¹ and intersecting social structures independent of SES. For adults with MIoRES, this unique set of intersectional identities is difficult to study empirically but raises many questions on how inequality is perpetuated along these intersectional axes.

Finally, adults with minoritized racial/ethnic identities were less likely to report barriers to health care than were heterosexual White adults after controlling for sociodemographic, health, and socioeconomic characteristics. This result was unexpected given the general trend in the health disparities literature that racial/ethnic minorities, and specifically African Americans, are more likely to report barriers to health care.¹ This implies that barriers to health care among African Americans or Hispanics might have been worse without controls for sociodemographic, socioeconomic, and health characteristics (models 1 and 2). In the United States, the overall lower SES profile of adults with minoritized racial/ethnic identities relative to Whites is well documented; however, there is a large amount of variability in measures of educational attainment, home ownership, health insurance coverage, and employment status across and within racial/ethnic groups as well. Tables A and B highlight these discrepancies for the MIoRE and MIoRES subgroups.

This study contributes to the growing literature on access to health care among adults with MIoRES by employing intersectionality to identify individual differences and examine the role of inequality. Instead of focusing on only 1 dimension of social identity, we were able to compare barriers to care among 4 minoritized identity categories. This unique operationalization of minoritized identity allowed us to present a nationally representative quantitative profile of adults across various sociodemographic, socioeconomic, and health characteristics and their interaction, or lack thereof, with the health care system.

Limitations

Although we were able to operationalize minoritized identity differently, our study was not without limitations. First, adults with MIoS and MIoRES made up less than 4% of the total analytic sample, and there were further limitations in the overall SES diversity present in these groups of adults. One advantage of using the NHIS is that the data collected are representative of the civilian noninstitutionalized adult population.¹¹ The United States has the world’s highest incarceration rate.²² Among those incarcerated, 56% are African American or Hispanic. Adults with MIoS are also disproportionately incarcerated. Meyer et al. found that, among prison inmates, 9.3% of men and 42% of women identified as sexual minorities.²³ The disproportionate incarceration of adults with minoritized identities and their experiences with the civilian and prison health care systems require further examination.

Second, although our use of categorical intersectionality was purposeful, identity is complex and rational and many dimensions are susceptible to change. When we dichotomized race/ethnicity and sexual orientation to construct our minoritized identity variable, we deliberately grouped adults with diverse experiences into 1 of 2 categories in the interest of comparing access to health care across the 4 minoritized identity categories. Although we were able to examine heterogeneity in access to health care between 4 groups of adults, a limitation was our inability to measure what we expected to be substantial heterogeneity within groups.

On the basis of previous literature,^1,3 we expected adults with diverse racial and ethnic identities (e.g., native-born African American vs foreign-born Mexican American) and sexual identities (e.g., lesbian vs bisexual) to experience the health care system differently as a result of experiences rooted in time and place, and thus we expected them to report different barriers to health care.^24,25 Further integrating these types of identities, along with contextual and historical experiences of inequality, into future analyses could help in examining specific identities.

Given these limitations, we identified 3 areas for future research. First, issues of participation and underrepresentation of adults with minoritized identities in nationally representative surveys need to be acknowledged and addressed. When minoritized adults are excluded from or underrepresented in data such as those derived from large, nationally representative surveys, we risk silencing or diminishing their experiences. As funding and opportunities grow in the realm of big data, we suggest the use of sampling methodologies and data sets that intentionally capture the unique experiences of individuals with a diverse set of minoritized identities at different and intersectional axes of inequality to better inform decision-makers on the issues these populations face.

Second, future research should examine the relationship between other dimensions of identities, beyond the social, and barriers to health care. Particularly of interest, given the intensive health disparities literature, are dimensions of gender, class, and inequality. Finally, intersectionality refers to the intersection of individual-level minoritized identities within structural-level systems of oppression.⁸ Thus, we suggest the use of structural-level data and multilevel models when possible.

Public Health Implications

Approaching health disparities research using an intersectional lens moves the discussion forward from examining individual differences to examining the role of social structures such as the health care system in maintaining and reproducing inequality. We found that adults with MIoRE, MIoS, and MIoRES were more likely to report delaying care for noncost reasons, including not being able to schedule an appointment soon enough, waiting too long to see a doctor, inconvenient office hours, and lacking transportation to get to the doctor’s office or clinic. In much the same way as disparities by race/ethnicity have been studied, acknowledged, and incorporated into medical training, similar attention should be given to how sexual minority groups interact with and relate to the medical profession.

According to Collins, “because public health remains connected to the practices of health care professionals, the challenge for this field is integrating intersectional frameworks into both health care delivery practices and public policies.”^9(p16) This integration will require difficult and sensitive discussions on the part of health care users and providers but could have the benefits of resulting in more appropriate and sensitive care for minoritized adults and further reducing health care access disparities,²⁶ especially as they relate to noncost reasons. Another requirement is the deliberate inclusion of minoritized adults and their voices in data collection efforts and at the center of all stages of programming, curriculum and policy development, and decision-making. Education, acceptance, and empathy are necessary for this transformation in health care to occur.