Race Differences in Patient Experience by Hispanic Ethnicity Among Veteran Health Administration Users

INTRODUCTION

Race and ethnicity are complex social constructs whose measurement and subsequent classification have ramifications for how health systems track and address racial/ethnic health disparities.^{1, 2} Racial/ethnic classifications are proxies for social contexts, lived-experiences, and relationships that influence healthcare experiences including experiences with racism, cultural proficiency, and communication barriers.³ Some research suggests that racial/ethnic minorities, including Hispanics, report worse provider satisfaction and communication, and posit that cultural and language barriers in healthcare settings contribute to these disparities.^3–5

Research on racial/ethnic differences in patient experiences often combines race and ethnicity into a single measure. However, the relationship between Hispanic ethnicity and patient experience for patients of different races is unknown, which has implications for whether combined race/ethnicity categories accurately capture differences. We examined whether the relationship between race and patient experience differed by Hispanic ethnicity among Veteran Health Administration (VHA) users of different races.

METHODS

We used data from the 2014 and 2015 Survey of Healthcare Experiences of Patients (SHEP) Patient-Centered Medical Home Survey, conducted by the VHA Office of Reporting, Analytics, Performance, Improvement and Deployment. SHEP, based on the Consumer Assessment of Healthcare Providers & Systems (CAHPS) survey, includes validated patient healthcare experience measures.⁶ Our sample included a stratified random sample of VHA users with at least one VHA primary care visit in 2013–2015 (n = 551,992) drawn from VA care facilities nationwide. Self-reported patient race (White, Black, American Indian/Alaskan Native (AI/AN), Native Hawaiian and Other Pacific Islander (NH/OPI)) and ethnicity (Hispanic, non-Hispanic) came from SHEP. Outcomes were two domains of patient experience with primary care: a six-item composite provider communication measure (percent of items receiving top rating versus not) and a single-item overall provider rating measure (high [≥ 9] on 0-to-10 scale versus not high [≤ 8]).

We assessed whether patient experiences differed between Hispanics and non-Hispanics within each race group with multivariate linear and logistic regression models that included a product term between patient race and Hispanic ethnicity, while controlling for patient age, sex, self-rated health, and education. For each race group, we used the fitted model to compute (1) predicted probability of a high provider rating for Hispanic and non-Hispanic patients separately; (2) mean provider communication rating for Hispanic and non-Hispanic patients separately; and (3) difference in probability or mean rating between Hispanic and non-Hispanic patients.

RESULTS

Within race groups, the proportion identifying as Hispanic ranged from 2.4% (Blacks) to 17.0% (AI/ANs) (Table 1). After controlling for individual-level characteristics, Hispanic Whites were more likely to report high provider ratings (predicted probability difference = 4.4%, p value < 0.001) and better provider communication (mean difference = 2.2%, p value < 0.001) than non-Hispanic Whites (Fig. 1). By contrast, Hispanic NH/OPIs were less likely to report higher provider rating than non-Hispanic NH/OPIs (predicted probability difference = 3.6%, p value = 0.088). We did not find differences by Hispanic ethnicity for provider rating or provider communication quality in other race groups (all p values ≥ 0.280).

Table 1.

VHA Sample Characteristics

	White		AI/AN		Black		NH/OPI
	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic
Sample size	21,859	439,515	942	4,596	1,349	55,021	377	1,972
Demographics
Age, mean (SD)	65.9 (13.1)	69.4 (11.4)	62.3 (11.6)	64.2 (10.9)	62.6 (13.4)	62.5 (11.9)	62.1 (11.8)	61.9 (11.9)
Sex, %
Female	4.9	4.1	7.0	6.7	9.2	9.9	4.8	5.9
Self-rated health, %
Excellent	5.3	4.9	5.3	4.1	6.9	4.1	5.6	4.1
Very good	17.9	21.6	15.0	16.8	18.6	16.1	15.3	17.0
Good	34.9	38.3	31.2	33.0	30.2	36.6	33.5	35.5
Fair	32.2	27.5	33.8	33.8	35.3	34.6	32.7	31.1
Poor	9.7	7.7	14.6	12.4	9.1	8.6	12.9	12.4
Educational attainment, %
≤ 8th grade	4.2	3.0	3.0	2.7	1.9	1.7	2.4	0.7
Some high school	7.1	6.6	6.3	6.5	8.2	6.0	7.7	4.2
High school graduate or GED	29.2	35.3	28.6	29.8	29.6	31.7	34.4	33.0
Some college or 2-year degree	40.2	36.2	46.1	43.4	35.2	43.5	41.1	45.2
4-year college graduate	10.8	9.5	8.8	8.0	15.0	9.0	7.2	9.8
> 4-year college degree	8.6	9.5	7.2	9.6	10.2	8.1	7.2	7.1
Patient experience
Provider rating1, %
High rating	68.5	69.2	64.4	63.3	70.9	66.0	65.7	65.3
< High rating	31.5	30.8	35.6	36.7	29.1	34.1	34.3	34.7
Provider communication quality2, mean % (SD)	72.3 (31.2)	74.4 (36.2)	66.8 (40.3)	67.4 (39.8)	75.4 (35.1)	72.5 (37.4)	70.7 (38.4)	69.1 (39.5)

Notes. AI/AN denotes American Indian/Alaska Native; NH/OPI denotes Native Hawaiian and Other Pacific Islander

¹Provider rating based on a 0 (worst) to 10 (best) scale, dichotomized as high rating of 9–10 versus < high rating of 0–8.

²Percent of questions assessing provider communication quality to which respondents answered always (versus never, sometimes, usually): how well provider listened, showed respect, explained well, spent enough time, gave information that was easy to understand, and knew about patients’ medical history

Figure 1.

High quality patient experience by race and Hispanic ethnicity—adjusted predicted probabilities and means. Notes: *p < 0.1; **p < 0.05 for ethnicity-by-race interaction. Provider rating based on a 0 (worst) to 10 (best) scale, dichotomized as high rating of 9–10 versus < high rating of 0–8. Provider communication quality based on percent of questions to which respondents answered always (versus never, sometimes, usually): how well provider listened, showed respect, explained well, spent enough time, gave information that was easy to understand, and knew about patients’ medical history. Models adjusted for age, sex, self-rated health, and educational attainment. Predicted probabilities and contrasts calculated at mean.

DISCUSSION

We found few differences in patient experience by Hispanic ethnicity within race groups. However, there were exceptions. Among Whites, those of Hispanic ethnicity reported higher provider ratings and better provider communication quality. Non-Hispanic NH/OPIs reported greater provider satisfaction than their Hispanic counterparts. While previous research notes that self-identified Hispanic ethnicity may indicate acculturation and language barriers,⁴ our findings among White and NH/OPI VHA users suggest that Hispanic identity’s influence on patient experience is more complex for these groups. Specifically, the Hispanic vs. non-Hispanic association with patient experience differed across race groups. Additionally, in our sample of VHA users, NH/OPI had the second largest proportion of individuals who identified as Hispanic (16%), so understanding the multifaceted interplay between race and Hispanic ethnicity may be particularly important for this group.

Study limitations included having too few Asian and multiple race patients to study in this sample, limited power to detect differences by Hispanic ethnicity for smaller race groups, and limited generalizability to non-VHA users (e.g., VHA users may have better healthcare experiences and higher English proficiency). Additionally, these differences are small, and their clinical meaning is uncertain.

VHA and other federal agencies should consider whether and how to combine racial and ethnic categories in ways to accurately capture patients’ experiences. While current categories may suffice for reporting and monitoring purposes,² it is also important to qualitatively understand how individuals self-identify by and experience race and ethnicity, and how these identities shape their healthcare experiences.

Compliance with Ethical Standards

Conflict of Interest

The authors declare that they do not have a conflict of interest.