Educational Attainment and Satisfaction With the Healthcare System: Racial Variation

Copyright © 2020 The Author(s). This is an open-access article distributed under the terms of the Creative Commons Attribution License (http:// creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 1. Background Research on the Minorities’ Diminished Returns (MDRs) phenomenon1 has proposed a new explanation for why ethnic health disparities exist,2,3 sustain,4-6 and grow7-11 over time. According to this framework, health inequalities are not limited to poor ethnic minorities, as the health of middle-class ethnic minorities also suffers.12,13 This view is supported by a large body of research showing that the health effects of socioeconomic status (SES), particularly education level, are weaker for ethnic minority groups relative to Whites.2,3 Similar patterns are shown for education level,14 income,15 employment,16 and marital status,17 suggesting that almost all SES indicators generate fewer tangible health outcomes for non-White than White Americans. MDRs research has attributed some of the ethnic inequalities in self-rated health,18-20 depression,21 anxiety,17 suicide,22 obesity,23 chronic diseases,15 chronic obstructive pulmonary disease,24 asthma,15 attentiondeficit/hyperactivity disorder25 hypertension,26 disability,27 hospitalization,28 and mortality29 in African Americans and Latinos to effect of SES indicators for ethnic minorities. This means there is a spillover effect of health inequalities to higher SES sections of society and the middle-class in marginalized groups.12,30-33 As these MDRs are well established and consistently replicated at national and local levels, researchers have recently made an initiative to explore if social stratification, economic markets,2,3,19,34 psychological processes,35,36 health behaviors,37-39 and health care systems40 have a role in explaining why high SES ethnic minorities still report poor health.18 Some studies have blamed the labor market by documenting a high risk of poverty among highly educated African American families.41 This view is further supported by evidence showing high levels of exposure to occupational stress and toxins in highly educated African Americans.42 Other studies have attributed MDRs to proximity to Whites and a higher risk of experiencing interpersonal discrimination in highly educated ethnic minorities.30-32 Although research has pointed to various mechanisms for such a phenomenon, one potential source of MDRS may be the healthcare system and behaviors related Abstract

to how ethnic minorities interact with the healthcare system. It has been suggested that highly educated African Americans and Latinos report less than expected health service utilization. 43 For example, highly educated and high income African Americans report less than expected usage of oral health exams, 44 breast exams, 45 and prostate exams. A similar pattern is shown for health risk behaviors, as African Americans and Latinos display weaker effects of education and income on health behaviors. [37][38][39][46][47][48][49] To give a few examples, highly educated African Americans and Latinos commonly display worse than expected diet, 46 exercise, 47 smoking, vaping, 39 alcohol use, 37,50 and binge drinking 37 behaviors. As shown by a recent study that followed individuals for 20 years, highly educated and high income African Americans show less than expected preventive health-seeking behaviors, suggesting that lower than expected health care use may be a reason for worse than expected health effects of ethnic minorities. 40,43,44

Objectives
There is a lack of previous study on MDRs of education level and income on satisfaction with healthcare across ethnic groups among adults. In response to this knowledge gap, this study investigated the MDRs of education level on satisfaction with healthcare among African American and Latino adults. We compared ethnic groups of adults for the association between education level and satisfaction with healthcare. In line with the MDRs theory, 2,3 and past research, 43 we expected a weaker link between education level and satisfaction with healthcare for African Americans and Latinos in comparison to their White and non-Latino counterparts. That is, we expected a low satisfaction with healthcare across all levels of education for African Americans and Latinos, a pattern opposite for Whites and non-Latinos. According to our search, this will be among the first, if not the first study, to test racial and ethnic variation in the association between educational attainment and satisfaction with the health care system.

Methods
The National Health Interview Survey (NHIS) was used for this cross-sectional study. NHIS is one of the central national health surveys of adults in the US. The NHIS data collection is run by the National Center for Health Statistics (NCHS) at the Centers for Disease Control and Prevention (CDC). The NHIS study uses a face-to-face interview format at participants' houses. For some individuals, a face-to-face interview was followed by a phone interview.

Participants and Sampling
The NHIS sample is limited to US residents who are noninstitutionalized, civilian adults. To enroll the sample, NHIS employed a multi-stage sampling strategy.

The Analytical Sample
A total of 24,835 adults were included in this analysis. This sample included all men and women who were 18+ years of age, participated in the NHIS, were non-Latino, Latino White, or African American, and provided valid data for all our variables (including use of healthcare services in the past 12 months). We did not include people from other ethnic groups or those who had not used health care services in the past 12 months.

Outcome
Satisfaction with healthcare. Participants' satisfaction with healthcare was asked using the following single item: "How satisfied were you with health care over the past 12 months?". Originally responses ranged from 1 to 4 (1 = very dissatisfied, 2 = somewhat dissatisfied, 3 = somewhat satisfied, and 4 = very satisfied). We operationalized this variable as a dichotomous variable: 0 = dissatisfied (very dissatisfied or somewhat dissatisfied) and 1 = satisfied (somewhat satisfied or very satisfied). Thus, our outcome was positive, not negative.

Predictor
Education level. Education was a four-level categorical variable with the following levels: less than high school diploma, high school graduate, some college, and college graduate.

Covariates
Psych Distress. Borrowed from the Kessler 6-item Psychological Distress Scale, 51-53 the following six items were used to measure psychological distress in the participants: 1) "How often you felt so sad nothing cheers you up during the past 30 days", 2) "How often you felt nervous during the past 30 days", 3) "How often you felt restless/fidgety during the past 30 days", 4) "How often you felt hopeless during the past 30 days", 5) "How often you felt everything was an effort during the past 30 days", and 6) "How often you felt worthless during the past 30 days". Responses ranged from 0 to 4 (none of the time to all of the time). The total sum score was calculated with a potential range from 0 to 24, with a higher score indicating higher psychological distress. For this study, we used this measure as a categorical variable after applying a cut-off point of 6. These items are based mainly on the measure, which is widely validated and used.
Body Mass Index (BMI). Participants height and weight were self-reported. Participants BMI was calculated and then classified based on a threshold of 30. BMI ≥ 30 was coded as 1 and BMI < 30 was coded as 0.
Self-rated health (SRH). NHIS applied the conventional single-item measure of SRH to measure self-rated health.
Participants were asked to report their overall health.
Possible responses were excellent, very good, good, fair, or poor. High SRH indicates worse health. We operationalized SRH as a dichotomous variable (0 = good health vs. 1 = poor health). Research has well-established high the validity and reliability of SRH as one of the strongest predictors of allcause risk of mortality. 54 Chronic Medical Conditions (CMCs). Participants were asked if a doctor has ever told them that they have any of the following chronic medical conditions: 1) diabetes mellitus, 2) emphysema/asthma, 3) stroke, 4) heart disease, 5) cancer, 6) arthritis, 7) hypertension, and/or 8) psychiatric conditions. We operationalized this variable as a dichotomous variable: 0 = no chronic disease vs. 1 = any chronic diseases.
Usual Place of Care. The NHIS used a single item to measure usual place of care. Participants were asked "Do you have a place you usually go when you get sick?". Responses were 1 (with a usual place of care) and 0 (without a usual place of care).
Demographic Characteristics. Demographics included age, marital status, and region. Age, an interval variable, was treated as a continuous measure in our analysis. Region was treated as a categorical variable, and included Northeast, Midwest, South, and West.

Statistics
The NHIS used a multi-stage sample design. Thus, we reestimated standard errors (SEs) by adjusting for the survey weights. We used weighted analysis/SPSS 23.0 (IBM Inc, New York, USA), to perform univariate (descriptive) and multivariable (analytical) analyses. We ran two logistic regression models, both in the pooled (overall) sample. In all models, education level (a 4-level categorical variable) was the independent variable. Healthcare satisfaction was the outcome/dependent variable. Ethnicity was the moderator. Model 1 did not enter ethnicity by education level interaction terms. Model 2, however, included ethnicity by education level interaction terms.

Descriptive Statistics
This study included 24,835 American adults. Participants were either Non-Latino, Latino, White (n = 20834), or African American (n = 4001). Of all of the participants, 1403 (5.6%) were dissatisfied and 23432 (94.4%) were satisfied with the healthcare they had received over the past 12 months. Table 1 provides a summary of the descriptive statistics in the study. Table 2 summarizes the results of two ethnic-stratified logistic regression models with education level as the predictor/independent variable and satisfaction with healthcare as the outcome/dependent variable. One model was estimated in Whites (Model 1) and another model was estimated in African Americans (Model 2). Model 1 showed that high education levels were associated with higher satisfaction with healthcare in Whites (OR = 1.24; 95% CI = 1.02-1.52; P = 0.034 for high school graduates compared to those with less than a diploma, OR = 1.25; 95% CI = 1.02-1.53; P = 0.029 for college graduates). Model 2 showed that higher education levels were not associated with higher satisfaction with healthcare in African Americans. Table 3 summarizes the results of two logistic regression models with education level as the predictor/independent variable and satisfaction with healthcare as the outcome/ dependent variable. Both of these models were estimated  in the overall sample with the following difference: Model 3 only entered the main effects, whereas Model 4 added three interaction terms between ethnicity and education levels. Based on Model 3, higher education levels were not associated with higher satisfaction with healthcare. Model 4 revealed interactions between African American ethnicity and education level on satisfaction with healthcare. These interactions suggested significantly smaller effects of high educational attainment on satisfaction with healthcare for African Americans than White adults (0.61; 95% CI = 0.38-1.00; P = 0.050 for AA x High School Graduate and OR = 0.60; 95% CI = 0.37-0.97; P = 0.036 for AA x Some College). Similar interactions could not be found for Latino ethnicity, suggesting that education similarly enhances healthcare satisfaction for Latino and non-Latino adults. Thus, the effect of high education level on increasing satisfaction with healthcare was smaller for African Americans than Whites, but similar for Hispanic and non-Hispanic adults (Table 3).

Discussion
This study showed an association between education level and satisfaction with healthcare in Whites but not African Americans in the US. This means highly educated Whites tend to be more satisfied with the healthcare system, whereas, African Americans tend to remain dissatisfied with the healthcare system at all levels of education. We found a similar association between education level and satisfaction with healthcare for Latino and non-Latino adults suggesting that the association between education and healthcare satisfaction is similar regardless of whether an individual is Latino or not. Our analysis successfully documented MDRs of education level for satisfaction with healthcare among African Americans relative to White adults. Previous research documented similar MDRs of various SES indicators in African Americans. 2,3 The unique aspect of this study was to replicate MDRs for a new outcome: satisfaction with healthcare. One previous study showed that in middle-aged and older adults, highly educated and high-income African Americans show poor health care use, preventive care use, and disease management. 43 This finding may be related to the high level of discrimination against African Americans in the healthcare system. 55 Social, economic, psychological, and behavioral mechanisms may explain why MDRs emerge. The psychological mechanism suggests that highly educated African Americans face more, not less, discriminatory experiences. [56][57][58][59][60] Similarly, an increase in SES may increase rather than decease African Americans' vulnerability to discrimination. 36 The social explanation discusses the role of segregation, and proximity to Whites when SES increases. Moving out of their communities to predominantly White neighborhoods, and working with predominantly White coworkers may increase perceived discrimination. [56][57][58][59][60] At least some of this is because high SES African Americans tend to move closer to White communities, which may increase their cross-ethnic interactions and encounters. [56][57][58][59][60] Discrimination, being a risk factor for many undesired health outcomes, impairs the expected gains of SES for African Americans. 14,38,44,[60][61][62][63] An economic explanation refers to the high likelihood of poverty in highly educated African Americans. 41 This is because education level has weaker effects on income 34 and job quality 42 of African Americans. Thus highly educated African Americans work in worse conditions and experience more stress than Whites. 64 Highly educated African Americans also live in worse conditions compared to highly educated Whites. 41 As a result of all these processes, highly educated African Americans do not similarly enjoy the expected effects of attaining additional education when compared to their White counterparts.
Overall, this study is the first to document MDRs of education on satisfaction with healthcare. Past research has shown MDRs of education and income for tobacco use, 38 drinking alcohol, 37,65 fruit and vegetable intake, 46 impulsivity 66 obesity, 23 chronic illness, 15 subjective health, 18 happiness, 34 depressed mood, 21 and suicidal ideation and attempt. 67 MDRs in healthcare satisfaction may be a mechanism by which MDRs in other health domains emerge.
Less than expected healthcare satisfaction of highly educated African Americans may be a consequence of differential treatment. An extensive body of research has shown that African Americans and other ethnic minority groups are often treated poorly in the healthcare system. 68 These differential treatments are shown for all types of healthcare needs and processes and are a rule rather than an exception. 68 Consequently, these differential treatments, combined with poor access, low trust, and low acceptability of healthcare interventions, may contribute to why African Americans experience worse healthcare satisfaction. [69][70][71] Ethnic minorities experience discrimination across all US institutions including the police, education system, labor market, banking, 72,73 and healthcare system. 68 Such discrimination may explain why we observed diminished effects of SES on the healthcare satisfaction of African Americans (MDRs). Due to worse treatment, ethnic minorities have developed lower levels of trust toward the healthcare system. 74 No matter what the diagnostic or treatment process is, across all types of conditions and illnesses, care is always of lower quality for African Americans than Whites. 68 These discriminatory experiences in the health care system predict poor outcomes for African Americans. 55 As such, for African Americans, high education does not generate the best health outcomes for African Americans in the US healthcare system. 34,41 At least some of the MDRs of education for African Americans are due to structural factors and residential segregation. Given the existing social stratification, African American and Latino individuals are more likely to live in underserved and marginalized urban areas that are very limited in a wide range of resources, services, and goods. Still, some highly educated African Americans and Latinos stay in their communities. Therefore, education for some Latinos and African Americans may not result in a change of neighborhood and greater access to resources. As such, highly educated Latinos and African Americans continue to live in underserved neighborhoods so they would have poor access to high-quality health care. Poor urban areas have few healthcare resources. 23,75 In the US, individual-level resources and assets are not enough for securing positive health outcomes. Such resources may generate positive health outcomes for Whites; however, they always generate less than expected positive health outcomes in any group that deviates from the center of society. This may be due to the high level of ethnocentrism in the US, and the fact that institutions are designed to function in a way that maximizes the gain of a particular group, White heterosexual men.
In a race-aware society, even if the individual is motivated to seek care, their experiences with institutions such as education, police, and the healthcare system are not always positive. At the same time, environmental conditions may not be conducive to a healthy lifestyle and health care use when needed. Neighborhoods with a high concentration of poverty, crime, violence, drugs, and social disorder may be a barrier against a healthy lifestyle. [76][77][78][79][80] Due to segregation, access to a high-quality healthcare system may be diminished for African Americans. There is a need to study how these societal and contextual constraints reduce the health of high SES ethnic minorities. 23 As shown by previous research, health disparities have sustained 4-6 and increased. [7][8][9][10][11] Ironically, education is not a real equalizer but instead may become a source of inequality. 81,82 For example, Zajacova, [82][83][84] Hayward, 85,86 Montez, 81,82 and others [86][87][88] have all shown that education does not result in the same level of health for African Americans and Whites. While education can be acquired in different settings, education can seldom generate the same health for diverse ethnic groups. As such, education becomes a source of health disparities across ethnic

What Is Already Known?
Education enhances satisfaction with the healthcare system. It is still unknown if the effect of education on healthcare satisfaction is similar or different across diverse ethnic groups.

What This Study Adds?
Although education enhances satisfaction with the healthcare system of Americans, this effect is unequal across ethnic groups. Education shows a diminishing effect on increasing satisfaction with the healthcare system for African Americans when compared with White Americans. Latino and non-Latino individuals, however, show similar effects of education on satisfaction with the healthcare system.

Research Highlights
groups. Thus, we need to enhance the quality of education and access to the healthcare system, as well as reduce discrimination for Latino and African American people. These should be done in addition to reducing the SES gap between ethnic minorities and Whites. 2,3

Implications
The results generated by this study may contribute to a reduction of ethnic health disparities, particularly those that are MDRs-related and impact middle-class African Americans. We may be able to enhance preventive behaviors and improve disease management of highly educated ethnic minorities through increasing their satisfaction with the healthcare system. This is important because low satisfaction is a predictor of non-adherence to preventive care, disease management, treatment, and medication. Such differential adherence and timing of diagnosis, missed preventive care and disease management visits are among the major causes of ethnic disparities in healthcare. Increasing the satisfaction of ethnic minorities with the healthcare system may be one of the strategies that may reduce ethnic disparities in health. That means, health disparities can be narrowed without re-distribution of resources. Thus, there is a need to increase trust and reduce discriminatory experiences of African Americans and Latinos in the healthcare system.

Limitations
This study had a few limitations. First, we only tested MDRs of education level. Some other SES indicators, such as wealth, assets, income, and occupational prestige, were not considered. Future research should also explore how neighborhood-level SES indicators alter or explain the observed MDRs. Both the physical and social aspects of the environment may have a role in reducing the return on education for some ethnic groups. For example, characteristics of the educational institutions may have some role. Financial worries, health care access and utilization, and health insurance coverage may also confound or mediate the effects of education and race on satisfaction with the healthcare system. Third, we only focused on ethnic differences. Fourth, while we used logistic regression, another option was to use ordinal regression. Results may vary based on the statistical approach being used. Future research may also explore how other sources of marginalization, such as immigration, nativity, language, religion, and insurance, result in MDRs. [89][90][91][92][93][94] Satisfaction with healthcare was measured using a single item. There is a need to study the experiences of individuals with the health care system. Such research may measure constructs and variables beyond self-reports. Administrative and claim data as well as direct observation of medical encounters may provide unique information.

Conclusion
In summary, ethnic differences in health are not all due to ethnic differences in exposure to risk factors and the availability of resources but also the differential magnitude of the effect of resources. This study, for example, showed that education level has a diminishing return on satisfaction with healthcare among African Americans than Whites. It is still unknown if health care discrimination, mistrust, quality of care, or types of interactions reduce the satisfaction of highly educated African Americans with the healthcare system.

Conflict of Interest Disclosures
The author declares that he has no conflicts of interest.

Ethical Approval
All NHIS participants provided written consent. The NHIS study protocol was approved by the CDC Institutional Review Board (IRB). This analysis, however, is exempt from a full IRB review. According to the NIH decision tool available at https://grants.nih.gov/policy/humansubjects/ hs-decision.htm, our paper was "Non-Human Subject Research".