Diminished Returns of Educational Attainment on Heart Disease among Black Americans

We analyzed data (n=25,659) from a nationally representative survey of American adults in 2013. The first wave of the Population Assessment of Tobacco and Health Adult (PATH-Adult) study was used. The independent variable was education (college graduate, high school graduate, less than a high school diploma). The dependent variable was any heart disease. Age and gender were the covariates. Race, as well as ethnicity, were the moderators. Logistic regressions were used to analyze the data.


INTRODUCTION
Although the health effects of socioeconomic status (SES) are well established [1 -5], such effects are unequal across racial and ethnic groups [6 -10]. While a higher education level predicts better health [3,4], recent research has documented weaker protective effects of SES indicators for racial and ethnic minority groups [11 -14]. Marginalization-related Diminished Returns (MDRs) refer to significantly weaker health effects of SES indicators, particularly education, for the members of racial and ethnic minorities, particularly Blacks and Hispanics, relative to non-Hispanic White Americans [6,7]. MDRs are being viewed as a paradigm shift as it investigates the mechanisms behind poor health of racial and ethnic minority people across SES levels [6,7].
In one study, highly educated Black and Hispanic people were found to be exposed to high levels of second-hand smoke exposure [33]. In addition, highly educated Blacks and Hispanics are more likely to drink alcohol [15,34], smoke cigarettes [18,35], and vape e-cigarettes [36]. This is probably why highly educated Black and Hispanic individuals may be at a higher risk of a wide range of non-communicable diseases such as hypertension [37], chronic obstructive pulmonary disease (COPD) [29] and asthma [28], compared with highly educated non-Hispanic Whites. Increased risk of heart disease may be particularly important and explain why highly educated Hispanics and Blacks stay at high risk of disability [30], hospitalization [31], and mortality [32], particularly for Black individuals.
Some of the mechanisms behind these MDRs include societal and structural processes and factors such as residential segregation, extra costs of moving upward in the society [38,39], and a higher level of exposure [8,40,41] and sensitivity to discrimination [42] in ethnic minorities, particularly Blacks. As a result of such societal processes, high SES will be followed with less tangible health effects for Black people than non-Hispanic White people. However, less is known about MDRs of education level for heart disease in general.
As a result of MDRs [6,7], the health impact of SES indicators including but not limited to education level, is diminished for racial and ethnic minority groups. Thus, diverse racial and ethnic groups not only vary in their SES [43 -45] but also how their SES relates to health outcomes [22,46].
Although high education reduces exposure to risk factors overall [47,48], highly educated Blacks and Hispanics report high levels of environmental risk factors such as stress [39], discrimination [38], and active and passive exposure to smoke [18,33,36]. While the very same SES indicators, such as educational attainment, show a strong impact on lowering non-Hispanic Whites' environmental risk exposures, highly educated Blacks and Hispanics are still at risk of factors that increase their risk to non-communicable illnesses such as hypertension [37], attention deficit hyperactivity disorder (ADHD) [19], asthma [28], and COPD [29].

AIMS
To better understand whether MDRs also applies to racial and ethnic disparities in heart disease, we compared Blacks, Hispanics, and non-Hispanic Whites for the negative effect of education level against heart disease. Although research has well-documented the effects of race/ethnicity [28,49] and SES (education level) [28, 50, 51] on heart disease, few studies have ever tested MDRs of SES resources on heart disease [28,29]. This study goes beyond the additive effect of race/ethnicity and SES and investigates the multiplicative effects of SES and race/ethnicity on heart disease [28,29]. This work is in line with the research question if it is race/ethnicity and SES or race/ethnicity or SES that cause health disparities. To generate results that are generalizable to the U.S. population of adults, we borrowed data from the Population Assessment of Tobacco and Health -Adult (PATH-Adult) study, a survey with a nationally representative sample of adults (18+ years old). In line with the MDRs [6,7] and previous relevant studies [28,29], we hypothesized smaller effects of SES (educational attainment) on heart disease for Blacks and Hispanics than non-Hispanic Whites.

Study Design and Setting
This cross-sectional survey borrowed existing data from the PATH 2013 [52 -55] study, a national survey of American adults sponsored by the National Institutes of Health (NIH) and the Food and Drug Administration (FDA). PATH is a landmark survey of tobacco use and associated diseases. The data collection was carried out in 2013. The PATH design, sampling, sample design, and measures are described elsewhere [52 -55].
The sample is limited to the civilian, non-institutionalized, adult (18+ years old) residents of the United States. The PATH data is a longitudinal study; however, we only used baseline data (wave 1). The sample design follows a multistage probability sample that recruited a representative sample of households and non-institutional people. The PATH study has used a multi-stage sampling strategy that involves survey weights. Due to the national sample and application of survey weights, the results of PATH are generalizable to the US population. In this study, 25,659 adults were analyzed.

Ethics
The PATH study protocol was approved by the Westat Institutional Review Board (IRB). All participating individuals signed informed written consent. All the PATH data were collected, stored, and analyzed anonymously.

Education level
Education level was operationalized as a categorical variable with 3 levels: less than high school graduate, completed high school, and college graduates.

Race
All participants self-identified their race. Race was treated as a dichotomous variable in the current study [Blacks = 1, non-Hispanic Whites = 0].

Ethnicity
All participants self-identified their ethnicity. Ethnicity was treated as a categorical variable in the current study [Hispanics = 1, Non-Hispanics = 0].

Outcome
Heart Disease. Heart diseases in this study included heart attack, congestive heart disease, and any other heart disease. Three items were used to measure lifetime history of heart disease. For example, participants were asked, "Have you ever been told by a doctor or other health professional that you had heart attack?", "Have you ever been told by a doctor or other health professional that you had congestive heart disease?", "Have you ever been told by a doctor or other health professional that you had any other heart disease?"

Data Analysis
We used the survey mode of SPSS 23.0 (IBM Inc, NY, USA) to analyze the data. Taylor series linearization was used to re-calculate the standard errors (SEs) that address the complex design of the PATH sample. Our estimates and inferences are generalizable to the US sample. We ran logistic regression models without and with interaction terms, all in the pooled sample. Model 1 had the main effects of race, ethnicity, education level, and covariates. Model 2 also included the following interaction terms: race × education level and ethnicity × education level. In all models, educational attainment and poverty level were the independent variables (IVs), heart disease was the dependent variable (DV), and age and gender were covariates. Race and ethnicity were the moderators. Publicly available data are available at the University of Michigan ICPSR Website available at: https://www.icpsr.umich.edu/icpsrweb/NAHDAP/studies/3623 1. As Black and Hispanic were both coded as 1, for the main effect of high educational attainment we expected a significant OR <1 (protective factor), and a significant OR >1 for the statistical interaction term between race/ethnicity and high educational attainment.

Descriptive Characteristics
This study included 25,659 American adults who were either White (82.9%), Black (17.1%), Non-Hispanic (83.9%), or Hispanic (16.1%). Table 1 provides a summary of the descriptive statistics of the study sample. Almost half of the participants were women.  Table 2 shows a summary of the output of 2 logistic regression models. In both models, educational attainment was the independent variable and any heart disease was the dependent variable. Both models were estimated in the total sample, which included Whites, Blacks, Hispanics, and non-Hispanics. Model 1 only entered the main effects of educational attainment as well as race, ethnicity, and covariates. Model 2, however, also added 2 interaction terms between race and ethnicity with educational attainment.

Multivariable Models
As shown in Model 1, high education predicted lower odds of heart disease. Model 2, however, showed significant interaction terms between the variables race and ethnicity with the independent variable educational attainment on the outcome variable heart disease, suggesting that high education may have a smaller protective effect against heart disease for Blacks and Hispanics than non-Hispanic Whites. This finding was reflected with a <1 and significant (protective factor) OR for the main effect of high educational attainment, and a >1 and  significant (risk factor) OR for the statistical interaction term between race/ethnicity and high educational attainment. The differences in the OR of college graduation relative to lowest education were 35% and 62% for Blacks compared with Whites and Hispanic compared with non-Hispanics, respectively ( Table 2).

DISCUSSION
Two findings were observed. First, education level was inversely associated with the prevalence of heart disease in the overall sample. Second, Blacks and Hispanics were at a relative disadvantage compared with non-Hispanic Whites regarding the protective effect of educational attainment on heart disease. To be more specific, a significant interaction was found between race/ethnicity and college-level education (but not with high school education), suggesting that the protective effect of college-level education was smaller for Blacks and Hispanics than non-Hispanic Whites. That is, the negative association between education and heart disease was weaker for Blacks and Hispanics than non-Hispanic Whites. As a result, highly educated Hispanics and Blacks remain at high risk of heart disease, compared with similarly educated non-Hispanic and White people.
The first finding on the inverse association between education and heart disease is in line with what is known about the protective influence of SES on health [1, 2], noncommunicable diseases [28,50,56], and heart diseases [57,58]. The protective health effect of SES is well known.
The second observation that the very same SES indicator (e.g. education) shows considerably weaker effects on heart disease for Blacks and Hispanics is relatively new. Previously, for Blacks and Hispanics, high education has shown weaker effects on obesity [26,20], asthma [28], COPD [33], hypertension [37], smoking, diet [25], exercise [16], hypertension [37], asthma [28] and COPD [29]. However, this was never shown for Hispanics. It was also not shown for heart disease overall.
We found that the protective effect of college-level education was smaller for Blacks and Hispanics than non-Hispanic Whites. The results may be due to the fact that highly educated Blacks and Hispanics use cigarettes [59], e-cigarettes [36], alcohol [15], have poor diets [25], live sedentary life styles [16], and have high BMI [26] at higher rates than expected based on their SES. Such a "more than expected" behavioral risk profile of high SES Blacks and Hispanics is not limited to tobacco as similar patterns are seen for impulse control [21], suicide [27], depression [22, 60 -63], anxiety [23], stress [38,39], and obesity [26]. These patterns are shown inside clinical settings [27] and in communities [23,64], as well as within [24] and between [21] generations. They are also shown for children [28], youth [23], adults [24], and older adults [17]. The universal nature of these diminished returns is suggestive of structural factors that reduce the health effects of SES. Some may attribute these patterns to structural racism and social stratification [6,7].
Another explanation for the lower health effects of educational attainment for Blacks and Hispanics compared with non-Hispanic Whites is the differences in the "quality" of education for Whites and Blacks and Hispanics may account for the lower effects of education on heart disease (and other measures of well-being) in Blacks and Hispanics. In other words, high-school and college education may not mean the same thing for Whites, Blacks and Hispanics. Hispanics and Blacks are more likely to attend urban public schools with lower resources [65]. Manley and others have shown that schooling differences may explain some of the differential effects of education for Whites and people of color [66,67].
Another explanation for this finding is that compared with non-Hispanic Whites, Blacks and Hispanics with high education are more likely to be new to the educational class. This is because children with highly educated parents are more likely to attain higher education as well. Future research should test recency to the middle class, parental education, and being a first-generation college graduate, as these factors may hint at mechanisms by which non-White families gain less health from their educational attainment.
There is also a need to evaluate how early in the life course SES disadvantage results in atherosclerotic changes among Blacks. Research has shown that one reason high SES during adulthood is still associated with poor health outcomes among Blacks is childhood poverty [68 -72]. That means early adversity results in some effects that cannot be undone with later availability of resources and human capital [73 -77]. According to the Barker hypothesis [78], pathological changes that mediate adversities may start in utero. Some behavioral mechanisms such as drinking [34], vaping [36], smoking [18,59], poor diet [25], poor health care use [79,80], and low exercise [16] that are all common in highly educated Blacks may mediate the additional cardiovascular risk of middle-class Blacks. Another mechanism is environmental exposures in highly educated Blacks [81 -83]. Genetics may also accelerate the process of atherosclerosis for Black people, and some research suggests that genes may increase the vulnerability of Black populations to hypertension [84 -87] and kidney disease [87 -91], which both have implications for development and outcomes of heart disease [92,93]. Diabetes may also be more common in highly educated Blacks through some genetic predisposition [94].

LIMITATIONS
The major limitation of this study is its cross-sectional design. As a result of such design, we do not infer causal effects from our observations. Most SES indicators and health outcomes have bidirectional links and reverse causal effect of health on SES cannot be ruled out. Poor health is one of the reasons behind downward social mobility. This study measured heart diseases such as COPD and asthma based on self-report rather than physical examination, laboratory testing, or administered data. Self-reports have been shown to generate valid and reliable measures of chronic diseases. However, future research may replicate these findings for other ethnic groups such as Asian Americans, Native Americans as well as immigrants. Future research may also use comprehensive measures. The study is at risk of omitted confounders such as wealth and area SES. We only described the MDRs of education on heart disease. Future research may test whether education quality, discrimination, segregation, labor market discrimination, or neighborhood poverty can explain why Blacks and Hispanics have less health gain from their available resources than non-Hispanic Whites. Finally, it is possible that the effect of education on heart disease may also be different for males and females among Blacks and Hispanics (and Whites). Future work may also include interactions between education, race/ethnicity, and gender. Still, this study had a large sample size and extends what we know regarding the non-linear effects of race, ethnicity, and SES on health outcomes, particularly heart disease.

CONCLUSION
In the United States, the inverse association between education and heart disease is weaker for Blacks and Hispanics than non-Hispanic Whites. More specifically, a significant interaction was found between race/ethnicity and college-level education (but not with high school education), suggesting that the protective effect of college-level education was smaller for Blacks and Hispanics than non-Hispanic Whites. As a result, a higher than expected prevalence of heart disease in highly educated Black and Hispanic people, a rate which is disproportionate to their SES, has been observed. Researchers should know that health disparities are not merely due to the additive but also the multiplicative effects of race/ethnicity and SES. Real solutions to racial and health disparities in heart disease would be based not only on increasing SES of Blacks and Hispanic but also empowering them to translate their available SES resources to health outcomes. Public policies need to go beyond equalizing SES and specifically address societal barriers, environmental risk factors, and structural factors that endanger the health and well-being of Blacks and Hispanics at all SES levels. Economic and public policies are needed to minimize diminished health returns of SES (i.e., MDRs). Unfortunately, MDRs are historically overlooked as contributors to racial and ethnic health disparities in the US.

AUTHORS' CONTRIBUTIONS
SA designed the conceptual model, secured funding, performed the analysis, wrote the first draft, and revised the manuscript. MB, MS and SC contributed to the interpretation of the findings, drafts, and revision. All authors approved the final version of the text.

ETHICS APPROVAL AND CONSENT TO PARTI-CIPATE
The PATH study protocol was approved by the Westat Institutional Review Board (IRB), the Westat Federal wide Assurance (FWA) number is FWA00005551.

HUMAN AND ANIMAL RIGHTS
Not applicable.

CONSENT FOR PUBLICATION
Written informed consent has been obtained from all the participants.

AVAILABILITY OF DATA AND MATERIALS
The data supporting the findings of the article is available in the ICPSR at https://www.icpsr.umich.edu/ icpsrweb/NAHDAP/ studies/36231 with reference number ICPSR36231.