Perceptions of Racial-Ethnic Inequities in COVID-19 Healthcare and Willingness to Receive the COVID-19 Vaccine

Abstract

Background:

Perceptions of the US healthcare system can impact individuals’ healthcare utilization, including vaccination intentions. This study examined the association between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine.

Methods:

This study used data from REACH-US, a nationally representative online survey of a large, diverse sample of U.S. adults (N=5145 January 26, 2021-March 3, 2021). Confirmatory factor and regression analyses examined a latent factor of perceived racial–ethnic inequities in COVID-19 healthcare, whether the factor was associated with willingness to receive the COVID-19 vaccine, and whether associations varied across racial–ethnic groups reported as probit estimates (B) and 95% confidence intervals (CIs).

Results:

Perceived racial–ethnic inequities in COVID-19 healthcare were highest among Black/African American adults (mean latent factor score: 0.65 ± 0.43) and lowest among White adults (mean latent factor score: 0.04 ± 0.67). Black/African American (B = −0.08; 95% CI = −0.19, 0.03) and Native Hawaiian/Pacific Islander (B = −0.08; 95% CI = −0.23, 0.07) adults who perceived greater racial–ethnic inequities in COVID-19 healthcare were less willing than participants who perceived lower inequities. In contrast, American Indian/Alaska Native (B = 0.15; 95% CI = −0.01, 0.30), Asian (B = 0.20; 95% CI = 0.08, 0.31), Hispanic/Latino (English language preference) (B = 0.22; 95% CI = 0.01, 0.43), Multiracial (B = 0.23; 95% CI = 0.09, 0.36), and White (B = 0.31; 95% CI = 0.19, 0.43) adults who perceived greater racial–ethnic inequities in COVID-19 healthcare were more willing to receive the COVID-19 vaccine than participants perceiving higher inequities.

Conclusions:

Greater perceived racial–ethnic inequities in COVID-19 healthcare were associated with less willingness to receive the COVID-19 vaccine among Black/African American and Native Hawaiian/Pacific Islander adults.

Throughout the SARS-CoV-2 (i.e., COVID-19) pandemic, there were well-documented US racial–ethnic disparities in COVID-19 vaccination. Non-Hispanic Black, Hispanic, and American Indian and Alaska Native populations, for example, were less likely to receive the COVID-19 vaccine compared to White adults.^1–3 Many studies attributed disparities to individual-level decisions (e.g., Black adults reported less confidence in the safety and efficacy of the COVID-19 vaccine than White adults.^4,5 Black and Hispanic adults were more likely to have concerns about missing work due to the side effects of COVID-19 vaccine uptake than White adults⁶). However, disparities can also arise from structural factors, which include a longstanding history of systemic racism throughout the US healthcare system.^7–10

There are documented examples of ways in which the US healthcare system can impact individuals’ healthcare utilization, including vaccination intentions. Previous studies found that individuals’ perceptions of the healthcare system were important drivers of intentions to use healthcare services¹¹ (e.g., when individuals perceived barriers to their healthcare and/or concerns over receiving equal treatment, they were less likely to engage with the healthcare system^12–14). In many cases, negative perceptions of the healthcare system decreased intentions to vaccinate.^15–17 Moreover, prior research found that individuals developed perceptions of healthcare system inequities through both direct and indirect experiences with the healthcare system (i.e., personal experiences as well as vicarious experiences such as learning about negative healthcare experiences from friends, family, or the media).^18,19 Given that racial–ethnic inequities in COVID-19 healthcare occurred throughout the pandemic (e.g., racial–ethnic disparities in access to COVID-19 testing, allocation of COVID-19 vaccine across racial–ethnic groups, and medical treatment for patients with COVID-19^20–22), individuals from marginalized racial–ethnic groups may have directly or indirectly developed negative perceptions of the healthcare system. Therefore, perceived racial–ethnic inequities in COVID-19 healthcare may partially explain racial–ethnic disparities in COVID-19 vaccination uptake.

Recent studies examined perceived healthcare system inequities in the COVID-19 context.^23–25 One nationally representative US study conducted during the COVID-19 pandemic found that compared to Non-Hispanic White adults, Non-Hispanic Black adults were more likely to report that “African Americans had a harder time accessing care than White Americans,” whereas Hispanic adults were less likely to report that “Latinos had a harder time accessing care than White Americans.”²³ Another nationally representative study of Black adults living in the United States found that greater perceived racial differences in medical care and treatment by the healthcare system were associated with greater mistrust of the COVID-19 vaccine and, in turn, lower intentions of being vaccinated for COVID-19.²⁴ In addition, a recent study conducted among a sample of US college students found that students who perceived higher racial–ethnic inequities in COVID-19 healthcare had greater COVID-19 vaccination intentions.²⁵ This study²⁵ measured perceived COVID-19 healthcare system inequities as an underlying or “latent” construct using three indicators of healthcare inequities.

Although these prior studies found evidence of racial–ethnic differences in perceived healthcare system inequities and their association with COVID-19 vaccination intentions, only one study²⁵ examined perceived racial–ethnic inequities in COVID-19 healthcare as a latent construct. The use of a latent construct model measured by multiple indicators is well suited for assessing perceived racial–ethnic inequities in COVID-19 healthcare given that perceptions of healthcare system inequity are multifaceted^26,27 and can be captured by various aspects of healthcare (e.g., interactions with providers, receipt of different health care services such as vaccinations or diagnostic tests). The multifaceted construct also represents underlying attitudes about the healthcare system, which provides a broader, well-rounded concept of perceived healthcare compared to single indicators. Given that attitudes are often difficult to measure using distinct, observable indicators, attitudes are typically conceptualized as latent constructs rather than using composite (i.e., average) scores of the indicators.²⁸ The use of latent constructs also provides greater explanatory power and accounts for measurement error across items compared to single observed indicators and/or composite scores.²⁹ In addition, previous studies^23–25 did not directly examine associations between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine. Finally, previous studies used relatively smaller samples and were not nationally representative across all racial–ethnic groups and/or age groups.

The present study addressed these gaps in the literature by using a large, nationally representative, diverse sample of US adults to first examine whether a latent factor of perceived racial–ethnic inequities in COVID-19 healthcare varied across racial–ethnic groups. Next, this study examined whether the latent factor was associated with willingness to receive the COVID-19 vaccine in the overall study population (Figure 1). Finally, the study examined whether the association between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine varied across racial–ethnic groups.

FIGURE 1.

Conceptual model of perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine.

METHODS

Data Source and Study Population

The Race-Related Experiences Associated with COVID-19 and Health in the United States (REACH-US) study³⁰ is a cross-sectional, online survey that was administered to a diverse, nationally representative sample of US adults between January 26, 2021 and March 3, 2021 using the YouGov, Inc. participant panel. YouGov Inc. is a nonpartisan research firm that recruits participants (~1.8 million US residents) via online advertising, email communication, and partner-sponsored solicitations.³¹ Participants received rewards or incentives for their participation.

The REACH-US study included 5500 adults, aged 18 years and older, from seven racial–ethnic groups (i.e., 500 American Indian/Alaska Native, 1000 Asian, 1000 Black/African American, 1000 Hispanic/Latino, 500 Multiracial, 500 Native Hawaiian/Pacific Islander, and 1000 White adults) living in the United States. YouGov panel members were proximity matched to a target sample of US adults generated from the 2018 American Community Survey 1-year data to achieve nationally representative estimates. Additional details about the recruitment methods, the stratified sampling approach to include a diverse set of racial–ethnic groups, the matching process, and propensity scoring to generate sample weights and nationally representative estimates within racial–ethnic groups are published elsewhere.³²

Sample Sizes and Inclusion Criteria

First, we excluded participants who had received at least one dose of the COVID-19 vaccine (n = 355) from the analysis to capture their willingness to receive the COVID-19 vaccine. This resulted in a sample size of 5145 adults. Next, given the measurement noninvariance of the latent factor, we excluded the Hispanic or Latino Spanish language preference subgroup (n = 489) (see Statistical Analysis and Results sections for additional details). In the adjusted latent regression analyses, we further excluded participants who were missing any of the covariates (n = 33) or because of small cell sizes for gender identity (i.e., self-identified as transgender [n = 15] or reported their gender as not listed [n = 19]), which resulted in a sample size of 4589 participants. De-identified data were provided to the study team. The study was considered exempt, nonhuman subject research as determined by the Institutional Review Board of the National Institutes of Health.

Measures

Willingness to Receive the COVID-19 Vaccine

Willingness to receive the COVID-19 vaccine was captured by asking participants, “Do you plan to get the COVID-19 vaccine once it becomes available?.” Response options included: (0) definitely not, (1) probably not, (2) probably yes, (3) definitely yes, (4) I have received one dose of the COVID-19 vaccine, and (5) I have received two doses of the COVID-19 vaccine. Consistent with prior literature,³³ these responses were treated as an ordinal variable in the analysis: (1) “definitely not,” (2) “probably not,” (3) “probably yes,” and (4) “definitely yes.” To capture willingness to receive the COVID-19 vaccine, participants who had received at least one dose of the COVID-19 vaccine (n = 419) were excluded from the analysis.

Indicators of Perceived Racial–Ethnic Inequities in COVID-19 Healthcare

We used three indicators to capture perceived racial–ethnic inequities in COVID-19 healthcare. Access to COVID-19 testing was measured by “How true is it that racial and ethnic minority groups have less access to COVID-19 testing compared to Whites?” Response options included: (1) very true, (2) somewhat true, and (3) not true. COVID-19 vaccine distribution was measured by “How confident are you that the COVID-19 vaccine will be distributed fairly across racial and ethnic groups?” Response options included: (1) not confident, (2) somewhat confident, and (3) very confident. Given that this question asked about fair distribution, we reverse-coded these items to reflect increasing perceptions of inequity. Treatment of COVID-19 patients was measured by “How often have racial and ethnic minority patients with COVID-19 been treated unfairly by the US health care system because of their race or ethnicity?” Response options included: (0) never, (1) somewhat often, and (2) very often.

Race–Ethnicity and Sociodemographic Covariates

Participants self-identified their race–ethnicity using three items. Participants self-identified whether they were of Hispanic, Latino, or Spanish origin (i.e., “no” or “yes”). If participants selected “yes,” we included them in a subgroup labeled “Hispanic/Latino,” which included participants from all gender identities (see gender items below). We further divided this Hispanic/Latino group by language preference (i.e., English or Spanish). Finally, participants self-identified their race using categories from the 2020 US Census. These three items were combined to include the race–ethnicity items of American Indian/Alaska Native, Asian, Black/African American, Hispanic/Latino English language preference, Hispanic/Latino Spanish language preference, Multiracial, Native Hawaiian/Pacific Islander, and White.

Sociodemographic characteristics included age in years (18–34, 35–49, 50–64, 65 and older), gender (man, woman, nonbinary, transgender, or not listed), education (high school or less, some college or 2-year college, 4-year college, or post-graduate), and annual household income (<20,000, 20,000–49,000, 50,000–100,000, or ≥100,000). In addition, we included sociodemographic items previously associated with COVID-19 vaccination as covariates^34–37 (i.e., political ideology [conservative, liberal, moderate, and not sure], health insurance coverage [not covered or covered], and high-risk chronic health condition).

Statistical Analyses

The present study consisted of three types of analyses: (1) descriptive analyses, (2) latent factor analyses, and (3) regression analyses. All analyses were weighted to be nationally representative within each racial–ethnic group. In sensitivity analyses, we reran all analyses when including the 355 participants who we previously excluded from the main analyses because they received at least one dose of the COVID-19 vaccine. These participants were included in the category of “definitely yes” in the sensitivity analyses.

Descriptive Analyses

We used descriptive analyses to examine the distribution of sociodemographic characteristics in the study population, indicators of perceived racial–ethnic inequities in COVID-19 healthcare, and willingness to receive the COVID-19 vaccine. We performed descriptive analyses in R Version 4.1.2. We used chi-square difference tests to assess differences in the distributions of sociodemographic characteristics, indicators of perceived racial–ethnic inequities in COVID-19 healthcare, and willingness to receive the COVID-19 vaccine when stratified by race–ethnicity.

Latent Factor Analyses

In the latent factor analyses, we used confirmatory factor analysis to examine perceived racial–ethnic inequities in COVID-19 healthcare as a latent factor and to examine the latent factor mean scores across racial–ethnic groups. The latent factor analyses were conducted in Mplus Version 8.6. The confirmatory factor analysis included probit regression and a diagonally weighted least squares estimator. Given the small number of missing responses for each of the indicators (i.e., <0.1%), a pairwise deletion approach was used with this estimator. This approach mitigated the loss of data by using all available cases in the data³⁸ and allowed for the use of the full weighted study population (N = 5145) when examining the presence of the latent factor.

We used confirmatory factor analysis to examine the factor loadings (i.e., associations between the three indicators of perceived racial–ethnic inequities in COVID-19 healthcare and the latent factor representing perceived racial–ethnic inequities in COVID-19 healthcare). Standardized coefficient cutoffs for the correlations between each item and the latent factor included ≥0.50 as acceptable and ≥0.70 as strong factor loadings.³⁹ Next, we assessed “measurement invariance” to confirm that the three indicators of perceived racial–ethnic inequities in COVID-19 healthcare measured the same underlying latent factor consistently across racial–ethnic groups.⁴⁰ Multigroup confirmatory factor analysis was used to test model fit when introducing increased constraints of the factor loadings, thresholds, and residuals across racial–ethnic groups (eAppendix Figure 1; http://links.lww.com/EDE/C123). Additional details on the measurement invariance testing methods are included in the eAppendix; http://links.lww.com/EDE/C123. Racial–ethnic groups with measurement noninvariance were excluded from the subsequent analysis comparing latent factor mean scores and latent factor regression.

Following the confirmatory factor analysis and measurement invariance testing, we estimated the scores for each latent factor for each participant using maximum a posteriori estimation.⁴¹ We used an analysis of variance test to examine whether the average latent factor scores of perceived racial–ethnic inequities in COVID-19 healthcare varied across racial–ethnic groups.

Regression Analyses

We conducted regression analyses to examine the association between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine in the overall study population and across racial–ethnic groups (see eAppendix Figure 2; http://links.lww.com/EDE/C123 for conceptual model). Latent factor regression was conducted in Mplus Version 8.6 and used an ordinal probit model with listwise deletion (i.e., excluding participants with missing data on any of the covariates [n = 33] or exclusion due to small cell sizes [gender identities of transgender {n = 15} and not listed {n = 19}] from the regression analyses). The final weighted sample used in the regression analyses included 4589 participants. We used probit estimates and 95% confidence intervals (CIs) to assess the association between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine in the overall study population, adjusting for race–ethnicity and sociodemographic covariates.

We conducted the multigroup analysis using ordinal probit regression to examine the association between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine in each racial–ethnic group. In multigroup analysis, we estimated the overall regression model (i.e., willingness to receive the COVID-19 vaccine regressed on perceived racial–ethnic inequities in COVID-19 healthcare) in each racial–ethnic group, adjusting for sociodemographic covariates. We used Wald tests to examine differences in the associations between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine across racial–ethnic groups. Additionally, we assessed the adjusted probabilities of reporting “definitely not” to receiving the COVID-19 vaccine based on perceived racial–ethnic inequities in COVID-19 healthcare using item characteristic curves generated in Mplus and exported them to R for plotting.

RESULTS

Descriptive Analyses

Participants’ sociodemographic characteristics varied across racial–ethnic groups (Table 1). Overall, most participants (36%) reported that they would definitely get the COVID-19 vaccine once it became available (Table 2). COVID-19 vaccination intentions also varied across racial–ethnic groups. Asian adults had the highest proportion of participants who would definitely get the COVID-19 vaccine (47%), while American Indian/Alaska Native (29%) and Black/African American adults (22%) had the highest proportions of participants who would definitely not get the COVID-19 vaccine.

TABLE 1.

Study Population Characteristics and Perceptions of Racial-Ethnic Inequities in COVID-19 Healthcare Across Racial–Ethnic Groups

	Total (N = 5145)	AI/AN (N = 458)	Asian (N = 926)	Black/AA (N = 951)	Hispanic/Latino ELP (N = 463)	Hispanic/Latino SLP (N = 489)	Multiracial (N = 473)	NH/PI (N = 461)	White (N = 925)
Age, %
18–34	36	36	35	35	44	36	48	38	26
35–49	27	24	30	24	27	37	25	34	22
50–64	24	26	23	27	14	24	18	22	30
65 and older	13	14	12	14	15	3	9	6	22
Gender, %
Man	44	35	46	46	50	39	47	34	49
Woman	54	62	53	53	49	61	46	64	50
Nonbinary	1	3	1	<1	<1	<1	5	<1	1
Transgender	<1	<1	<1	<1	<1	<1	<1	2	0.0
Not listed	<1	0.0	<1	0.0	<1.0	0.0	2	<1	<1
Education, %
≤High school	40	43	25	43	53	67	33	47	33
Some college or 2-year college	34	45	25	38	32	21	38	38	35
4-year college	16	8	30	12	8	9	19	12	20
Post-graduate	10	4	20	7	7	3	10	3	12
Annual household income, %
<20,000	31	43	19	42	31	32	25	39	21
20,000–49,000	31	32	26	30	35	48	28	27	30
50,000–100,000	24	18	29	19	24	16	32	25	30
≥100,000	14	7	26	9	10	4	15	9	19
Political ideology, %
Conservative	24	29	19	15	22	18	16	26	41
Liberal	29	22	33	31	33	26	42	19	25
Moderate	32	30	39	37	32	28	32	30	27
Not sure	15	19	9	17	13	28	10	25	7
Health insurance coverage, %
Not covered	16	13	10	16	18	42	12	17	10
Covered	84	87	90	84	82	58	88	83	90
High-risk chronic health conditions7, %
Not high-risk	59	50	67	53	64	70	59	59	55
High-risk	41	50	33	47	36	30	41	41	45

Weighted to be nationally representative within each racial–ethnic group (the weighted total for each racial-ethnic group is rounded and will not add to the weighted overall total). Less than 1% of participants were missing age (total population [0.1%] and Asian [0.1%]), gender (total population [0.1%], Multiracial [0.2%], White [0.3%]), annual household income (total population [0.2%], Asian [0.1%], Black/AA [0.3%], Multiracial [0.2%], White [0.5%]), political ideology (total population [0.1%], Black/AA [0.1%]), and health insurance coverage (total population [0.2%], AI/AN [0.4%], Asian [0.1%], Black/AA [0.1%], Hispanic/Latino SLP [0.2%], NH/PI [0.2%], White [0.2%]). Less than 2% of participants selected their gender as nonbinary (total population [1.2%], Asian [0.7%], Black/AA [0.6%], Hispanic/Latino ELP [0.6%], Hispanic/Latino SLP [0.2%], NH/PI [0.3%], White [0.9%]), transgender (total population [0.3%], Asian [0.6%], Hispanic/Latino ELP [0.3%], Multiracial [1.9%], NH/PI [0.5%], White [0.1%]) and not listed (total population [0.4%], Asian [0.6%], Hispanic/Latino ELP [0.3%], Multiracial [1.9%], NH/PI [0.5%], White [0.1%]).

All categories of sociodemographic characteristics varied across racial-ethnic groups (P values < 0.01).

We selected high-risk chronic health conditions using the Centers for Disease Control (CDC) guidelines on medical conditions (e.g., heart conditions, type 2 diabetes, asthma) that increase the risk for severe illness due to COVID-19. High-risk chronic health condition was included as a covariate and coded as 1 if participants reported at least one medical condition that was considered a high-risk chronic health condition. All other responses were coded as 0.

AI/AN indicates American Indian/Alaska Native; B/AA, Black/African American; ELP, English Language Preference; NH/PI, Native Hawaiian/Pacific Islander; SLP, Spanish Language Preference.

TABLE 2.

Perceived Racial-Ethnic Inequities in COVID-19 Healthcare Across Racial–Ethnic Groups

	Total (N = 5145)	AI/AN (N = 458)	Asian (N = 926)	Black/AA (N = 951)	Hispanic/Latino ELP (N = 463)	Hispanic/Latino SLP (N = 489)	Multiracial (N = 473)	NH/PI (N = 461)	White (N = 925)
Willingness to receive the COVID-19 vaccine, %
Definitely not	17	29	9	22	14	11	18	17	20
Probably not	18	16	13	23	16	13	19	28	16
Probably yes	29	28	31	29	34	36	24	29	24
Definitely yes	36	27	47	26	36	40	39	26	40
Perceived racial–ethnic inequities in access to COVID-19 testing, %
Not true	40	44	38	28	37	47	34	40	53
Somewhat true	34	30	36	35	39	34	33	41	29
Very true	26	26	26	37	24	19	33	19	18
Perceived racial–ethnic inequities in COVID-19 vaccine distribution, %
Very confident	29	34	29	19	31	29	22	30	40
Somewhat confident	39	36	45	41	39	32	43	39	33
Not confident	32	30	26	40	30	39	35	31	27
Perceived racial–ethnic inequities in treatment of COVID-19 patients, %
Never	38	41	37	22	35	53	29	35	50
Somewhat often	40	39	43	43	45	37	41	42	34
Very often	22	20	20	35	20	10	30	23	16

Weighted to be nationally representative within each racial-ethnic group. Less than 1% of participants perceived racial-ethnic inequities in access to COVID-19 testing (total population [0.1%], Asian [0.1%], Multiracial [0.4%], White [0.2%]), perceived racial-ethnic inequities in COVID-19 vaccine distribution (total population [0.1%], Black/AA [0.1%], White [0.2%]), perceived racial-ethnic inequities in treatment of COVID-19 patients (total population [0.1%], Asian [0.1%], Black/AA [0.1%], Multiracial [0.4%], White [0.2%]).

All categories of sociodemographic characteristics varied across racial-ethnic groups (P values <0.01).

AI/AN indicates American Indian/Alaska Native; Black/AA, Black/African American; ELP, English Language Preference; NH/PI, Native Hawaiian/Pacific Islander; SLP, Spanish Language Preference.

Overall, most participants (60%) reported that it was at least somewhat true or very true that marginalized racial–ethnic groups have less access to COVID-19 testing (34% and 26%, respectively) (Table 2). Most participants (71%) were not fully confident in the fair distribution of the COVID-19 vaccine (39% reported being somewhat confident and 32% reported not confident). Most participants (62%) reported that COVID-19 patients from marginalized racial–ethnic groups were treated unfairly by the healthcare system somewhat often (40%) to very often (22%). These responses to each indicator varied across racial–ethnic groups. Black/African American adults had the highest proportion of participants who reported that it was very true that marginalized racial–ethnic groups had less access to COVID-19 testing (37%), that they were not confident in the fair distribution of the COVID-19 vaccine across racial–ethnic groups (40%), and that COVID-19 patients from marginalized racial–ethnic groups very often get treated unfairly by the healthcare system (35%). White adults had the highest proportion of participants who perceived that it was not true that marginalized racial–ethnic groups have less access to COVID-19 testing (53%) and who reported that they were very confident that the COVID-19 vaccine was fairly distributed across racial–ethnic groups (40%). Hispanic/Latino adults with Spanish language preference had the highest proportion of participants who reported that COVID-19 patients from marginalized racial–ethnic groups were never treated unfairly by the healthcare system (53%).

Latent Factor Analyses

The three indicators of perceived racial–ethnic inequities in COVID-19 healthcare were well represented by the latent factor and were measured consistently across all racial–ethnic groups, except for the Hispanic/Latino Spanish language preference subgroup. Therefore, we excluded the Hispanic/Latino Spanish language preference subgroup from the latent factor regression analysis. Additional details on the confirmatory factor analysis and measurement invariance testing results are included in the eAppendix; http://links.lww.com/EDE/C123.

The average latent factor scores for perceived racial–ethnic inequities in COVID-19 healthcare varied across racial–ethnic groups (Table 3). Black/African American (mean: 0.65, standard deviation [SD] 0.43) and Multiracial adults (mean: 0.50, SD: 0.63) had the highest group means of perceived racial–ethnic inequities in COVID-19 healthcare, whereas White adults had the lowest group mean of perceived racial–ethnic inequities in COVID-19 healthcare (mean: 0.04, SD: 0.67).

TABLE 3.

Latent Factor Scores for Perceived Racial–Ethnic Inequities in COVID-19 Healthcare, by Race–Ethnicity

Racial-Ethnic group	Mean (SD)	Range
AI/AN (N = 458)	0.23 (0.64)	−0.69, 1.43
Asian (N = 926)	0.31 (0.57)	−0.55, 1.36
Black/AA (N = 951)	0.65 (0.43)	−0.23, 1.35
Hispanic/Latino ELP (N = 463)	0.33 (0.38)	−0.29, 1.10
Multiracial (N = 473)	0.50 (0.63)	−0.54, 1.50
NH/PI (N = 461)	0.30 (0.48)	−0.44, 1.23
White (N = 925)	0.04 (0.67)	−0.76, 1.40
Total (N = 4656)	0.34 (0.59)	−0.76, 1.50

AI/AN indicates American Indian/Alaska Native; Black/AA, Black/African American; ELP, English Language Preference; NH/PI, Native Hawaiian/Pacific Islander.

Regression Analyses

The regression analysis revealed that participants who perceived greater racial–ethnic inequities in COVID-19 healthcare were more willing to receive the COVID-19 vaccine in the adjusted model (B = 0.14; 95% CI = 0.08, 0.20) (Table 4). Standardized estimates are included in eAppendix Table 1; http://links.lww.com/EDE/C123.

TABLE 4.

Unstandardized Associations Between Perceived Racial-Ethnic Inequities in COVID-19 Healthcare and Willingness to Receive the COVID-19 Vaccine, by Race-Ethnicity

	Model II (adjusted)
Racial-Ethnic group	B	(SE B)	(95% CIs)
Total	0.14	(0.03)	0.08, 0.20
AI/AN	0.15	(0.08)	−0.01, 0.30
Asian	0.20	(0.06)	0.08, 0.31
Black/AA	−0.08	(0.06)	−0.19, 0.03
Hispanic/Latino ELP	0.22	(0.11)	0.01, 0.43
Multiracial	0.23	(0.07)	0.09, 0.36
NH/PI	−0.08	(0.08)	−0.23, 0.07
White	0.31	(0.06)	0.19, 0.43

Model II adjusted for age, gender, education, annual household income, political ideology, health insurance coverage, and high-risk chronic health conditions. Missing data on any of the covariates (n = 33) or exclusion due to small cell sizes (i.e., gender identities of transgender [n = 15] and not listed [n = 19]) resulted in a sample size of 4589 in the adjusted regression analyses.

AI/AN indicates American Indian/Alaska Native; Black/AA, Black/African American; ELP, English Language Preference; NH/PI, Native Hawaiian/Pacific Islander.

The multigroup regression analysis revealed racial–ethnic differences in the association between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine (Table 4). The adjusted probabilities of reporting “definitely not” to receiving the COVID-19 vaccine based on perceived racial–ethnic inequities in COVID-19 healthcare for each racial–ethnic group are presented in Figure 2.

FIGURE 2.

Adjusted probabilities of reporting “definitely not” to receiving the COVID-19 vaccine based on perceived racial–ethnic inequities in COVID-19 healthcare. Model adjusted for age, gender, education, annual household income, political ideology, health insurance coverage, and high-risk chronic health conditions. ELP indicates English Language Preference.

In the adjusted model, Black/African American (B = −0.08; 95% CI = –0.19, 0.03) and Native Hawaiian/Pacific Islander (B = –0.08; 95% CI = –0.23, 0.07) adults who perceived greater racial–ethnic inequities in COVID-19 healthcare were less willing to receive the COVID-19 vaccine (Table 4, although CIs were wide). By contrast, American Indian/Alaska Native (B = 0.15; 95% CI = –0.01, 0.30), Asian (B = 0.20; 95% CI = 0.08, 0.31), Hispanic/Latino English language preference (B = 0.22; 95% CI = 0.01, 0.43), Multiracial (B = 0.23; 95% CI = 0.09, 0.36), and White (B = 0.31; 95% CI = 0.19, 0.43) adults who perceived greater racial–ethnic inequities in COVID-19 healthcare were more willing to receive the COVID-19 vaccine (Table 4, although CIs were particularly wide for American Indian/Alaska Native adults). Wald tests revealed that the associations between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine were different in Black/African American adults compared to all other racial–ethnic groups, except for Native Hawaiian/Pacific Islander adults (Table 5). The associations between perceived racial–ethnic inequities in COVID-19 healthcare and willingness to receive the COVID-19 vaccine in Native Hawaiian/Pacific Islander adults were different compared to all other racial–ethnic groups, except for Black/African American adults.

TABLE 5.

Wald Chi-Square Tests for the Associations Between Perceptions of Racial–Ethnic Inequities in COVID-19 Healthcare and Willingness to Receive the COVID-19 Vaccine

	Asian Wald χ2 [df = 1]	Black/AA Wald χ2 [df = 1]	Hispanic/Latino ELP Wald χ2 [df = 1]	Multiracial Wald χ2 [df = 1]	NH/PI Wald χ2 [df = 1]	White Wald χ2 [df = 1]
AI/AN	0.28	5.60	0.32	0.64	4.12	2.74
Asian		11.56	0.04	0.13	7.87	1.82
Black/AA			6.10	11.68	0.003	21.25
Hispanic/Latino ELP				0.01	4.98	0.55
Multiracial					8.45	0.79
NH/PI						14.98

All models were adjusted for age, gender, education, annual household income, political ideology, health insurance coverage, and high-risk chronic health conditions. Missing data on any of the covariates (n = 33) or exclusion due to small cell sizes (i.e., gender identities of transgender [n = 15] and not listed [n = 19]) resulted in a sample size of 4589 in the adjusted regression analyses. Meaningful differences were observed between AI/AN and Black/AA, AI/AN and NH/PI; Asian and Black/AA, Asian and NH/PI; Black/AA and Hispanic/Latino ELP, Black/AA and multiracial, Black/AA and White; Hispanic/Latino ELP and NH/PI; Multiracial and NH/PI; and NH/PI and White groups (P values <0.05).

AI/AN indicates American Indian/Alaska Native; Black/AA, Black/African American; ELP, English Language Preference; NH/PI, Native Hawaiian/Pacific Islander.

Sensitivity Analyses

The results in the main analyses were consistent with those in the sensitivity analyses that included participants who received at least one dose of the COVID-19 vaccine (eAppendix, Tables 2–4; http://links.lww.com/EDE/C123).

DISCUSSION

This study used confirmatory factor and regression analyses to examine a latent factor of perceived racial–ethnic inequities in COVID-19 healthcare, whether the factor was associated with willingness to receive the COVID-19 vaccine, and whether associations varied across racial–ethnic groups. This study found that perceived racial–ethnic inequities in COVID-19 healthcare were well represented as a multifaceted latent factor, which is consistent with recent studies on perceived racial–ethnic inequities in COVID-19 healthcare using a latent construct.^25,42 However, the present study revealed differences in latent factor measurement for the Hispanic/Latino Spanish language preference subgroup compared to other racial–ethnic groups. This finding could reflect different experiences with the healthcare system and, in turn, different perceived racial–ethnic inequities in COVID-19 healthcare among Hispanic/Latino participants with Spanish language preference compared to other study participants. Alternatively, this finding could relate to previous evidence that a lower level of linguistic acculturation among Latino adults was associated with greater perceived healthcare system inequities⁴³ or differences due to the translation of the questionnaire.⁴⁴ Despite differences in measurement for the Hispanic/Latino Spanish language preference subgroup, the latent factor of perceived racial–ethnic inequities in COVID-19 healthcare was measured consistently across the remaining seven racial–ethnic groups, which allowed for comparisons of latent factor scores and the associations between the latent factor and willingness to receive the COVID-19 vaccine.

We observed racial–ethnic differences in perceived racial–ethnic inequities in COVID-19 healthcare in the present study, where perceived racial–ethnic inequities in COVID-19 healthcare were highest among Black/African American and Multiracial adults and lowest among White adults. These findings align with previous evidence that awareness of these healthcare system inequities was higher in Black/African American and Multiracial adults than in White adults.^45,46 Higher perceived racial–ethnic inequities in COVID-19 healthcare among Black/African American and Multiracial adults may be attributed to direct experiences with the healthcare system (e.g., healthcare provider discrimination among Black/African American adults are associated with more delays in healthcare seeking^45,47; Multiracial adults were more likely to report perceived discrimination in healthcare than White adults⁴⁶). In addition, there were many instances of COVID-19 healthcare system inequities throughout the COVID-19 pandemic among Black/African American and Multiracial adults.^48,49 Therefore, some Black/African American and Multiracial adults may have developed a greater awareness of perceived racial–ethnic inequities in COVID-19 healthcare through indirect experiences from social groups or media exposure reporting the disproportionate impact of COVID-19 among marginalized racial–ethnic groups.^50–53

The present study also found associations between greater perceived racial–ethnic inequities in COVID-19 healthcare and a greater willingness to receive the COVID-19 vaccine in the full study population, yet this association differed across racial–ethnic groups. Greater perceived racial–ethnic inequities in COVID-19 healthcare were associated with less willingness to receive the COVID-19 vaccine among Black/African American and Native Hawaiian/Pacific Islander adults although, the CIs were wide. Alternatively, among American Indian/Alaska Native, Asian, Hispanic/Latino English language preference, Multiracial, and White adults, greater perceived racial–ethnic inequities in COVID-19 healthcare were associated with greater willingness to receive the COVID-19 vaccine. These findings are consistent with the hypothesis that perceived racial–ethnic inequities in COVID-19 healthcare can differentially impact individuals based on whether they expect those inequities to affect their personal healthcare experiences. Black/African American and Native Hawaiian/Pacific Islander participants may have expected that they were more likely to experience COVID-19 healthcare inequities given the US history of mistreatment of Black/African American and Native Hawaiian/Pacific Islander groups (e.g., including, but not limited to the exploitation of Black or African American adults in the 1930’s US Public Health Service Study of Untreated Syphilis in the Negro Male⁵⁴ and the displacement of Native Hawaiian and Pacific Islander groups through US colonialism⁵⁵). Increased perceptions of racial–ethnic inequities in COVID-19 healthcare may have decreased willingness to receive the COVID-19 vaccine among Black/African American and Native Hawaiian/Pacific Islander study participants due to these expectations of potential harms by the US healthcare system and distrust in healthcare services. By contrast, American Indian/Alaska Native, Asian, Hispanic/Latino English language preference, Multiracial, and White adults may have been less concerned that they would personally experience COVID-19 healthcare inequities. It is possible that when American Indian/Alaska Native, Asian, Hispanic/Latino English language preference, Multiracial, and White adults learned more about the COVID-19 pandemic, their perceptions of racial–ethnic inequities in COVID-19 healthcare increased, as well as their willingness to receive the COVID-19 vaccine, particularly if they were less concerned that COVID-19 healthcare inequities would affect them personally.

Limitations should be considered in this study. With respect to the study sample, survey participants may have differed from those who did not participate in the online panel (e.g., those without access to the internet). In addition, participants who self-identified as transgender or reported their gender as not listed were excluded from the adjusted regression models due to small cell sizes, thereby limiting the generalizability of findings to adults of all genders. The survey was also only administered in English and Spanish languages, thus limiting the generalizability of these findings to non-English and non-Spanish speakers living in the United States. With respect to measurement, there may be unmeasured indicators of perceived racial–ethnic inequities in COVID-19 healthcare (e.g., perceived racial–ethnic disparities in clinical trials for COVID-19 treatment). Moreover, the study did not capture whether participants had personally experienced and/or anticipated they would be affected by perceived racial–ethnic inequities in COVID-19 healthcare which limited the ability to formally test whether greater perceived racial–ethnic inequities in COVID-19 healthcare among those who felt they would directly experience those inequities was associated with less willingness to receive the COVID-19 vaccine. Finally, the study was cross-sectional, which limited inferences on temporal order and causality about the impact of perceived racial–ethnic inequities in COVID-19 healthcare on willingness to receive the COVID-19 vaccine.

Despite these limitations, this study had several strengths. The study included a large nationally representative sample of diverse participants in the United States with substantial representation across seven racial–ethnic groups. Additionally, this study is one of the first to examine perceived racial–ethnic inequities in COVID-19 healthcare as a latent factor that captured the multifaceted nature of perceived racial–ethnic inequities in COVID-19 healthcare. Last, although the Hispanic/Latino Spanish language preference subgroup was not included in the regression analysis due to measurement noninvariance, their inclusion in the confirmatory factor analysis revealed important differences in how Hispanic/Latino participants with Spanish language preference may perceive racial–ethnic inequities in COVID-19 healthcare compared to adults who primarily speak English. This detection of measurement noninvariance is an important finding that extends current literature that neglected to assess differences by language preference.

There are multiple opportunities for future research. Given prior evidence that greater medical mistrust was associated with lower intent to vaccinate against COVID-19,⁵⁶ future studies could examine the association between healthcare inequities and willingness to receive the COVID-19 vaccine via trust in the healthcare system. Moreover, perceived racial–ethnic inequities in COVID-19 healthcare might be associated with additional COVID-19 behaviors (e.g., receipt of COVID-19 treatment), since these outcomes were also associated with trust in the healthcare system in prior studies.⁵⁷ Subsequent studies could examine perceived racial–ethnic inequities in COVID-19 healthcare with additional indicators to test a more robust measure of perceived racial–ethnic inequities in COVID-19 healthcare and examine other sociodemographic information that could influence racial–ethnic differences in perceived racial–ethnic inequities in COVID-19 healthcare (e.g., occupation type and geographic location). In addition, since there is limited evidence on whether perceived healthcare inequities differ based on language preference, future studies are warranted.

This study has public health implications for COVID-19 healthcare, as well as health services use and health equity more broadly. This study assessed perceived racial–ethnic inequities in COVID-19 healthcare, which can develop without directly engaging or interacting with the US healthcare system. In addition, perceptions of the US healthcare system can differentially drive healthcare behaviors, such as vaccination intentions, across racial–ethnic groups.

ACKNOWLEDGMENTS

The authors thank the YouGov, Inc staff and the participants of the Race-Related Experiences Associated with COVID-19 and Health in the United States (REACH-US) survey.

Supplementary Material

ede-35-377-s001.docx ^{(267KB, docx)}