Perceived Discrimination and Incident Cardiovascular Events

Abstract

Perceived discrimination is positively related to cardiovascular disease (CVD) risk factors; its relationship with incident CVD is unknown. Using data from the Multi-Ethnic Study of Atherosclerosis, a population-based multiethnic cohort study of 6,508 adults aged 45–84 years who were initially free of clinical CVD, we examined lifetime discrimination (experiences of unfair treatment in 6 life domains) and everyday discrimination (frequency of day-to-day occurrences of perceived unfair treatment) in relation to incident CVD. During a median 10.1 years of follow-up (2000–2011), 604 incident events occurred. Persons reporting lifetime discrimination in ≥2 domains (versus none) had increased CVD risk, after adjustment for race/ethnicity and sociodemographic factors, behaviors, and traditional CVD risk factors (hazard ratio (HR) = 1.36, 95% confidence interval (CI): 1.09, 1.70) and after control for chronic stress and depressive symptoms (HR = 1.28, 95% CI: 1.01, 1.60). Reported discrimination in 1 domain was unrelated to CVD (HR = 1.05, 95% CI: 0.86, 1.30). There were no differences by race/ethnicity, age, or sex. In contrast, everyday discrimination interacted with sex (P = 0.03). Stratified models showed increased risk only among men (for each 1–standard deviation increase in score, adjusted HR = 1.14, 95% CI: 1.03, 1.27); controlling for chronic stress and depressive symptoms slightly reduced this association (HR = 1.11, 95% CI: 0.99, 1.25). This study suggests that perceived discrimination is adversely related to CVD risk in middle-aged and older adults.

Racial/ethnic disparities in cardiovascular disease (CVD) risk in the United States are well-documented (1) but not clearly understood. Experiences of ethnic discrimination and racism have long been considered to contribute to racial/ethnic disparities in CVD rates (2), but evidence for this is scarce. As several reviews have highlighted (3–6), much of the extant literature on perceived discrimination and racism has focused on mental health consequences; more recently, studies have linked discrimination to physical health outcomes, including several indicators of cardiovascular health and functioning, such as higher prevalence of carotid artery atherosclerosis and coronary artery calcification, higher blood pressure, greater visceral fat and central adiposity, and higher levels of C-reactive protein and E-selectin (7–16).

Evidence linking discrimination to CVD risk factors is inconsistent, with positive, null, and conditional associations that vary by race/ethnicity and/or sex being observed (see Lewis et al. (2) and Williams et al. (17) for reviews). Moreover, there is debate in the literature as to whether racial/ethnic discrimination is more deleterious to health than other forms of discrimination (18). Few studies examining discrimination and CVD risk factors have clearly distinguished between race-based and nonracial discrimination. Studies that also evaluated racial/ethnic discrimination and CVD risk factors have generally not shown a stronger relationship for racial/ethnic discrimination than for overall perceptions of discrimination (7–9). Notably, no prior studies have investigated perceived discrimination or racial/ethnic discrimination in relation to incident CVD. Two studies have examined perceived discrimination and mortality (19, 20). Among community-dwelling older adults, perceived discrimination without regard to attribution was related to increased all-cause mortality risk, especially among whites (19), whereas in a study of black women, no association was found between racial/ethnic discrimination and all-cause mortality or death due to CVD or cancer (20).

In a recent review, Lewis et al. (2) called for large prospective epidemiologic studies of experiences of discrimination and objectively measured, clinically relevant endpoints, including clinical CVD outcomes, to more clearly examine the relationship between exposure to discrimination and CVD risk. There also is a need to evaluate whether the experience of unfair treatment in general or racial/ethnic discrimination specifically is related to CVD risk. In the present study, we examined self-reported perceived discrimination in relation to clinically adjudicated incident cardiovascular events in a population-based cohort of over 6,000 adults. We investigated experiences of discrimination or unfair treatment without regard to attribution as well as perceived racial/ethnic discrimination. Available data on health behaviors, traditional CVD risk factors, and other psychosocial stressors allowed us to evaluate whether associations were independent of established risk factors. Given reports in the literature showing that the relationship between adverse health and discrimination may vary by demographic characteristics, we also examined potential differences in the associations of discrimination with incident CVD by race/ethnicity, age, and sex.

METHODS

Study design and participants

The Multi-Ethnic Study of Atherosclerosis (MESA) is an ongoing population-based prospective cohort study of risk factors for subclinical atherosclerosis conducted at 6 US field centers (Baltimore, Maryland; Chicago, Illinois; St. Paul, Minnesota; Los Angeles, California; New York, New York; and Forsyth County, North Carolina) (21). Between July 2000 and August 2002, a total of 6,814 persons (60% of those eligible) aged 45–84 years who were asymptomatic with regard to clinical CVD were enrolled in MESA. All provided written informed consent, and institutional review boards at each participating institution approved the study.

For these analyses, participants were excluded if they were discovered to have had a prebaseline CVD event (n = 5), to have had no follow-up after baseline (n = 27), to have not completed the questionnaire for either discrimination measure (n =24), or to have missing data on family income (n = 250); this resulted in 6,508 persons in the analytical sample. Another 39 persons with missing data on the everyday discrimination measure only were excluded from analyses of that measure.

Outcome

The primary outcome was incident cardiovascular events, clinically adjudicated per a standard MESA protocol (22), with follow-up complete through December 31, 2011. Cardiovascular events included incident myocardial infarction, resuscitated cardiac arrest, coronary revascularization (percutaneous coronary intervention or coronary artery bypass graft), definite angina, fatal or nonfatal stroke, and death due to CVD, as defined by the MESA protocol. Follow-up time was defined as time (in days) between the baseline visit and either an incident CVD event, death, or the last follow-up contact, whichever occurred first. A total of 604 confirmed incident CVD events occurred during follow-up (median, 10.1 years; range, 0.02–11.5 years).

Discrimination measures

Lifetime discrimination

Two scales with documented reliability and validity in diverse samples (12, 23–27) were used to measure perceived discrimination at baseline. The Lifetime Discrimination Scale, adapted from the Detroit Area Study (23), asked respondents to report whether they had ever been treated unfairly (yes/no) in 6 domains: being fired from a job or denied a promotion; not being hired for a job; being treated unfairly by the police; being discouraged by a teacher/advisor from continuing one's education; being prevented from moving into a neighborhood by a landlord or realtor; and having life made difficult by neighbors. For each “yes” response, participants indicated the perceived reason for the unfair treatment (race/ethnicity, sex, age, religion, physical appearance, sexual orientation, income level/social class, or other). For our primary analysis, a summary score for lifetime discrimination without regard to attributed reason was created, with 1 point being assigned for each “yes” response (range, 0–6 points) (Cronbach's α = 0.61). Additionally, 3 groups were formed on the basis of responses indicating an experience of lifetime discrimination in 0 (57.9%), 1 (22.2%), or ≥2 (19.9%) domains. A summary score for lifetime discrimination attributed specifically to race/ethnicity was also created, with 1 point being assigned for each “yes” response attributed to race/ethnicity (range, 0–6 points).

Everyday discrimination

The Everyday Discrimination Scale, a 9-item scale developed for use in the Detroit Area Study (24), assessed everyday occurrences of unfair treatment. Respondents indicate the frequency with which certain experiences of unfair treatment occur in their day-to-day life, without reference to race/ethnicity, age, sex, or other demographic characteristics; sample items are: “You receive poorer service than other people at restaurants or stores”; “People act as if they think you are not smart”; and “You are called names or insulted.” Response options were expanded from the original 4-point scale to a 6-point scale: 1 = almost every day; 2 = at least once a week; 3 = a few times a month; 4 = a few times a year; 5 = less than once a year; and 6 = never. Responses to individual items were reverse-coded and averaged for an overall score (range, 1–6 points); a higher score indicated a greater frequency of unfair treatment (Cronbach's α = 0.88). Four groups were created based on the distribution of scores, representing levels of reported everyday unfair treatment: none (26.8%; score = 1), low (27.0%; score >1 and <1.5), moderate (24.8%; score 1.5–2), or high (21.4%; score >2).

Covariates

Information on covariates was taken from the baseline MESA visit. Sociodemographic variables included age, self-reported race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, or Chinese), sex, gross family income (categorized as low (<$20,000/year), medium ($20,000/year–<$50,000/year), or high (≥$50,000/year)), and education (highest attained level of completed education, modeled in 3 groups: less than high school; high school, technical school, or associate's degree; or college or more). Self-reported health behaviors included smoking (never, former, or current smoker), alcohol use (number of drinks per week, modeled in 4 groups: nondrinker, light drinker (1–7 drinks/week), moderate drinker (8–14 drinks/week), or heavy drinker (>14 drinks/week)), and physical activity, assessed by means of the MESA Typical Week Physical Activity Survey (28).

CVD risk factors included resting systolic blood pressure (average of the last 2 of 3 blood pressure measurements obtained after a 5-minute seated rest using an automated oscillometric sphygmomanometer (Dinamap Pro 100; Critikon, Tampa, Florida)), body mass index (weight (kg)/height (m)²), height (in centimeters), total and high-density lipoprotein cholesterol and triglyceride levels (measured from blood specimens obtained after a 12-hour fast using standard assays), low-density lipoprotein cholesterol level (calculated from the Friedewald equation (29)), self-reported use of antihypertensive medication (confirmed by medication review), and diabetes status based on the 2003 American Diabetes Association fasting criteria (30).

Chronic stress and depressive symptoms also were selected as covariates based on their association with perceived discrimination and/or CVD risk in prior studies (4, 5, 7, 31). Chronic stress, assessed with the Chronic Burden Scale (32), is a measure of the presence and severity of ongoing stress related to one's own health problems, the health problems of close friends or family members, employment/ability to work, relationships, and finances. The chronic stress score was the sum of the number of domains in which ongoing moderately or very stressful problems were reported (range, 0–5). Depressive symptoms were measured with the 20-item Center for Epidemiologic Studies Depression Scale (33); higher scores indicate more depressive symptoms (range, 0–60). A variable representing MESA field center (site) was included in all models.

Data analysis

Descriptive statistics were computed for all participants and by categories of lifetime discrimination. Cox proportional hazards regression was used to calculate adjusted hazard ratios and 95% confidence intervals for the associations of each discrimination variable with incident CVD events in separate models. Models were minimally adjusted for age, race/ethnicity, sex, education, family income, marital status, and MESA field center. Risk factor–adjusted models included covariates for health behaviors (model 2) and additionally body mass index, height, systolic blood pressure, use of antihypertensive medication, prevalent diabetes, lipid levels (total, high-density lipoprotein, and low-density lipoprotein cholesterol), triglyceride levels, and use of lipid-lowering medication (model 3). A final model adjusted for chronic stress and depressive symptoms (model 4). (Chronic stress and depressive symptoms were moderately correlated (r = 0.39) and showed small correlations with both measures of discrimination (r ≤ 0.25), suggesting no evidence of multicollinearity.) To assess interactions between discrimination measures and age, sex, and race/ethnicity, we repeated the analyses including a cross-product term for each potential interaction in the models. Proportional hazards assumptions were tested for all models, with no evidence of violation. All statistical analyses were performed using SAS, version 9.3 (SAS Institute, Inc., Cary, North Carolina). Because of missing values for some covariates, numbers in the Cox models varied slightly (from 6,271 to 6,508, with the number of events ranging from 579 to 604). Sensitivity analyses limiting all models to participants with complete data on all covariates showed results essentially identical to those presented.

RESULTS

Baseline characteristics are shown in Table 1 for all participants and by categories of lifetime discrimination. Nearly all demographic characteristics, behaviors, CVD risk factors, and psychosocial factors varied by reported discrimination. Persons reporting discrimination in ≥2 domains were younger, more educated, and had higher family incomes; were more likely to be male, black, currently unmarried, and smokers; reported more alcohol use; had a higher body mass index but lower blood pressure, cholesterol, and triglyceride levels; and were more physically active. They had more exposure to race-based discrimination and higher reported levels of everyday discrimination, chronic stress, and depressive symptoms. This pattern of characteristics for demographic factors, health behaviors, CVD risk factors, and psychosocial characteristics was essentially the same across all 4 everyday discrimination groups (data not shown).

Table 1.

Baseline Characteristics of Participants in the Multi-Ethnic Study of Atherosclerosis, 2000–2002

	All Participants (n = 6,508a)		Lifetime Discrimination Category						P Valueb
			None (n = 3,767)		1 Domain (n = 1,446)		≥2 Domains (n = 1,295)
	Mean (SD)	%	Mean (SD)	%	Mean (SD)	%	Mean (SD)	%
Demographic Factors
Age, years	62.0 (10.2)		63.3 (10.3)		60.8 (9.9)		59.4 (9.7)		<0.0001
Race/ethnicity									<0.0001c
Non-Hispanic white		39.0		42.0		41.5		27.7
Chinese		12.2		16.8		6.8		4.8
Black		26.4		17.8		30.2		47.3
Hispanic		22.3		23.4		21.5		20.2
Sex									<0.0001c
Female		52.8		56.5		50.1		45.1
Male		47.2		43.5		49.9		54.9
Education									<0.0001c
Less than high school diploma		17.7		22.0		14.4		8.9
High school diploma or some college/technical training		46.4		46.5		46.2		46.1
College degree or higher		35.9		31.5		39.4		45.0
Marital status									<0.0001c
Married/living with partner		61.1		64.3		58.2		54.8
Widowed		12.7		14.6		11.3		8.6
Divorced		13.6		10.8		16.6		18.1
Separated		3.6		2.6		5.0		4.8
Never married		8.4		7.2		8.0		12.3
Preferred not to answer		0.7		0.5		0.9		1.3
Annual family income									<0.0001c
<$20,000		23.8		26.7		20.2		19.5
$20,000–$49,999		36.5		36.2		38.7		35.1
≥$50,000		39.6		37.1		41.1		45.4
Health Behaviors
Alcohol use									<0.0001c
Nondrinker		47.9		51.6		44.1		41.4
Light (1–7 drinks/week)		36.5		34.1		39.6		40.1
Moderate (8–14 drinks/week)		8.6		7.8		9.1		10.5
Heavy (>14 drinks/week)		6.9		6.5		7.2		8.0
Smoking status									<0.0001c
Never smoker		50.5		54.6		45.9		43.9
Former smoker		36.5		34.4		40.4		38.3
Current smoker		13.0		11.0		13.7		17.8
Moderate/vigorous physical activity, MET-minutes/week	5,789 (5,922)		5,354 (5,486)		6,197 (5,793)		6,602 (7,065)		<0.0001
Cardiovascular Disease Risk Factors
Body mass indexd	28.3 (5.5)		27.8 (5.3)		28.7 (5.5)		29.3 (5.8)		<0.0001
Height, cm	166.4 (10)		165.2 (10)		167.4 (9.9)		168.6 (9.6)		<0.0001
Systolic blood pressure, mm Hg	126.2 (21.4)		127.1 (21.7)		125.3 (21.1)		124.9 (20.6)		0.0003
Use of antihypertensive medication		36.5		37.0		37.8		33.9	0.10c
Total cholesterol, mg/dL	194.3 (35.8)		195.6 (36.1)		193.6 (35.5)		191.4 (35.2)		0.0002
High-density lipoprotein cholesterol, mg/dL	51.0 (14.8)		51.3 (14.8)		50.8 (14.9)		50.0 (14.5)		0.006
Low-density lipoprotein cholesterol, mg/dL	117.3 (31.5)		117.6 (31.5)		116.9 (31.4)		116.6 (31.7)		0.27
Triglycerides, mg/dL	132.0 (87.7)		134.4 (90.1)		130.5 (84.2)		126.7 (83.9)		0.005
Use of lipid-lowering medication		16.1		16.8		17.1		13.1	0.006c
Fasting glucose/diabetes status
Normal		75.0		73.9		75.0		70.2	0.06c
Impaired fasting glucose		13.1		13.7		13.2		16.0
Untreated diabetes		2.5		2.6		2.6		2.8
Treated diabetes		9.4		9.8		9.5		11.0
Psychosocial Factors
Lifetime discriminatione	0.7 (1.1)		0.0 (0.0)		1.0 (0.0)		2.6 (0.9)		<0.0001
Racial/ethnic lifetime discriminationf	0.3 (0.8)		0.0 (0.0)		0.3 (0.5)		1.2 (1.3)		<0.0001
% reporting racial/ethnic lifetime discrimination		18.7		0		29.1		61.5	<0.0001
Everyday discriminationg	1.6 (0.7)		1.4 (0.5)		1.7 (0.6)		2.1 (0.8)		<0.0001
Chronic stress burdenh	1.2 (1.2)		1.0 (1.1)		1.4 (1.2)		1.7 (1.3)		<0.0001
Depressive symptomsi	7.6 (7.6)		7.0 (7.1)		7.4 (7.1)		9.5 (9.2)		<0.0001

Abbreviations: MET, metabolic equivalent of task; SD, standard deviation.

^a Numbers of participants varied slightly because of missing data on some covariates.

^b Unless otherwise specified, P values reflect tests for linear trend over lifetime discrimination categories from linear regression Wald χ² tests (for continuous variables) or Cochran-Mantel-Haenszel tests (for categorical variables).

^c Pearson's χ² test for independence.

^d Weight (kg)/height (m)².

^e Lifetime Discrimination Scale (23) score (range, 0–6).

^f Score for racial/ethnic discrimination on the Lifetime Discrimination Scale (range, 0–6).

^g Everyday Discrimination Scale (24) score (range, 1–6).

^h Chronic stress was assessed with the Chronic Burden Scale (32). The chronic stress score was the sum of the number of domains (personal health, friend/family health, employment, relationships, and finances) in which ongoing moderately or very stressful problems were reported (range, 0–5).

ⁱ Depressive symptoms were measured with the 20-item Center for Epidemiologic Studies Depression Scale (33). Higher scores indicate more depressive symptoms (range, 0–60).

Lifetime discrimination and incident CVD events

Table 2 presents results for lifetime discrimination modeled continuously. Experiences across more domains of lifetime discrimination were related to excess risk of incident CVD. Each 1–standard deviation increase in score on the lifetime discrimination measure was associated with 11% greater CVD risk (model 1). This finding was unchanged after adjustment for health behaviors (model 2) and traditional CVD risk factors (model 3) and was slightly reduced with adjustment for chronic stress burden and depressive symptoms (model 4). The depressive symptoms score was significant (hazard ratio (HR) = 1.01, 95% confidence interval (CI): 1.00, 1.02; P = 0.042) and chronic stress was marginally significant (HR = 1.08, 95% CI: 0.996, 1.17; P = 0.062) in model 4.

Table 2.

Results From Cox Models Examining the Association Between Experiences of Lifetime Discrimination^a and Incident Cardiovascular Disease Events (n = 6,508), Multi-Ethnic Study of Atherosclerosis, 2000–2011^b

Modelc	No. of Participants	No. of Events	HRd	95% CI
1	6,508	604	1.11	1.02, 1.21
2	6,454	602	1.12	1.03, 1.22
3	6,340	589	1.10	1.01, 1.20
4	6,302	582	1.07	0.98, 1.17

Abbreviations: CI, confidence interval; HR, hazard ratio.

^a Lifetime Discrimination Scale (23) score (range, 0–6).

^b Numbers of participants and events varied in models 2–4 because of missing data on some covariates.

^c Model 1 included covariates for age, race/ethnicity, sex, education, family income, marital status, and study field center. Model 2 additionally adjusted for alcohol use, smoking status, and moderate and vigorous physical activity. Model 3 additionally adjusted for body mass index, height, systolic blood pressure, use of antihypertensive medication, diabetes/fasting blood glucose status, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides, and use of lipid-lowering medication. Model 4 additionally adjusted for depressive symptoms and chronic stress burden.

^d Hazard ratio for a 1–standard deviation increment (1.08) in Lifetime Discrimination Scale score.

Analyses of lifetime discrimination modeled in 3 categories showed a threshold effect (Figure 1). Persons who experienced discrimination in ≥2 domains had 38% greater risk of incident CVD than those reporting no lifetime discrimination (model 1), a finding that was little changed after controlling for health behaviors (model 2) or CVD risk factors (model 3). The hazard ratio was attenuated somewhat but remained statistically significant in model 4 after further adjustment for depressive symptoms and chronic stress burden.

Figure 1.

Hazard ratios (Cox proportional hazards models) for the association of categorized lifetime discrimination (none (♦), 1 domain (●), or ≥2 domains (▴)) with risk of incident cardiovascular disease events over a median 10.1-year follow-up period, Multi-Ethnic Study of Atherosclerosis, 2000–2011. Model 1 included covariates for age, sex, race/ethnicity, education, family income, marital status, and study field center. Model 2 additionally adjusted for alcohol use, smoking status, and moderate and vigorous physical activity. Body mass index, height, systolic blood pressure, use of antihypertensive medication, diabetes/fasting blood glucose status, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides, and use of lipid-lowering medication were added to model 3; and model 4 was further adjusted for depressive symptoms and chronic stress burden. Numbers of participants and events in each discrimination group for model 1 were: none—n = 3,767 and 344 events; 1 domain—n = 1,446 and 129 events; and ≥2 domains—n = 1,295 and 131 events. Numbers of participants and events varied in models 2–4 because of missing data on some covariates. Bars, 95% confidence intervals.

None of the observed associations varied by age, sex, or race/ethnicity (P > 0.15). Additional analyses limited to perceptions of race-based discrimination showed no significant associations with incident CVD (HR = 1.04, 95% CI: 0.96, 1.14) per 1–standard deviation increase; this finding was unchanged across risk factor–adjusted models (data not shown). In subsequent exploratory analyses, we grouped participants on the basis of exposure to only race-based discrimination (11.9%), only non–race-based discrimination (23.4%), or both types of discrimination (6.8%). The adjusted hazard ratio for incident CVD was larger for persons exposed to both types of discrimination (HR = 1.56, 95% CI: 1.12, 2.17) than for those exposed to either non–race-based (HR = 1.14, 95% CI: 0.93, 1.40) or race-based (HR = 1.06, 95% CI: 0.79, 1.42) discrimination, in comparison with those reporting no discrimination. Among participants reporting both types of discrimination, the mean Lifetime Discrimination Scale score was 2.8, whereas among those reporting only 1 type of discrimination, the mean scores were 1.7 (race-based) and 1.4 (non–race-based), respectively, suggesting an overall higher level of exposure to discrimination among persons reporting both types.

Everyday discrimination and incident CVD events

Each 1–standard deviation increase in Everyday Discrimination Scale score was related to a 9% higher risk of CVD (HR = 1.09, 95% CI: 1.00, 1.19), adjusted for sociodemographic factors, and this finding was unchanged after further adjustment for health behaviors (HR = 1.09, 95% CI: 1.00, 1.19) and traditional CVD risk factors (HR = 1.09, 95% CI: 1.00, 1.19). With chronic stress burden and depressive symptoms added to the model, the association was reduced (HR = 1.04, 95% CI: 0.95, 1.14). There was no interaction between everyday discrimination and age or race/ethnicity (P > 0.07), but an interaction with sex was noted (P = 0.033). Subsequent sex-stratified models showed that everyday discrimination was associated with incident CVD in men only (Table 3). Men experienced approximately 15% increased risk of incident CVD, which was slightly reduced in model 4; for men, chronic stress was a significant covariate (HR = 1.12, 95% CI: 1.01, 1.24), whereas the depressive symptoms score was not (HR = 1.00, 95% CI: 0.98, 1.02).

Table 3.

Results From Sex-Stratified Cox Models Examining the Association Between Experiences of Everyday Discrimination^a and Incident Cardiovascular Disease Events (n = 6,469), Multi-Ethnic Study of Atherosclerosis, 2000–2011^b

Modelc	No. of Participants	No. of Events	HRd	95% CI
Men
1	3,052	360	1.15	1.04, 1.28
2	3,032	359	1.15	1.03, 1.27
3	2,974	351	1.14	1.03, 1.27
4	2,954	348	1.11	0.99, 1.25
Women
1	3,417	240	0.97	0.83, 1.14
2	3,385	239	0.98	0.83, 1.15
3	3,331	234	0.99	0.83, 1.17
4	3,317	231	0.92	0.77, 1.10

Abbreviations: CI, confidence interval; HR, hazard ratio.

^a Everyday Discrimination Scale (24) score (range, 1–6).

^b Numbers of participants and events varied in models 2–4 because of missing data on some covariates.

^c Model 1 included covariates for age, race/ethnicity, education, family income, marital status, and study field center. Model 2 additionally adjusted for alcohol use, smoking status, and moderate and vigorous physical activity. Model 3 additionally adjusted for body mass index, height, systolic blood pressure, use of antihypertensive medication, diabetes/fasting blood glucose status, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides, and use of lipid-lowering medication. Model 4 additionally adjusted for depressive symptoms and chronic stress burden.

^d Hazard ratio for a 1–standard deviation increment (0.67 for men, 0.65 for women) in Everyday Discrimination Scale score.

Modeling the Everyday Discrimination Scale score as 3 levels of reported unfair treatment (low, moderate, and high), with “none” as the referent group, showed no increase in risk for low (HR = 1.04, 95% CI: 0.84, 1.28) or moderate (HR = 0.99, 95% CI: 0.78, 1.25) levels of unfair treatment. Persons reporting the highest levels of unfair treatment had 26% greater risk (HR = 1.26, 95% CI: 0.98, 1.63), which was little changed after adjustment for health behaviors (HR = 1.25, 95% CI: 0.98, 1.62) or CVD risk factors (HR = 1.27, 95% CI: 0.98, 1.64) but was attenuated after controlling for chronic stress and depressive symptoms (HR = 1.11, 95% CI: 0.84, 1.46); this increase in risk was not statistically significant. There were no interactions between Everyday Discrimination Scale scores, modeled in groups, and sex, age, or race/ethnicity (P > 0.12).

DISCUSSION

In this population-based study of middle-aged and older adults, excess risk of incident CVD was related to reported experiences of discrimination. Observed associations were largely independent of age, race/ethnicity, sex, education, family income, marital status, and study site, as well as health behaviors and traditional risk factors for CVD. Reported levels of chronic stress and depressive symptoms appeared to account for some of this relationship, although persons who reported lifetime discrimination in at least 2 life domains (nearly 20% of our sample) remained at increased risk of CVD after we controlled for these psychosocial factors. Similar, albeit weaker, associations were observed when discrimination was assessed using everyday experiences of unfair treatment, but for men only. Perceptions of lifetime discrimination attributed specifically to race/ethnicity were not independently associated with increased risk of incident CVD, though persons reporting both race-based and non–race-based discrimination did show elevated risk.

There are several ways in which experiences of discrimination may be adversely associated with health. Discrimination has been associated with more depressive symptoms (34), poorer mental health (4–6, 17), poorer sleep (35, 36), increased adiposity (12–14, 37), higher blood pressure (9–11), inflammatory biomarkers (15, 16), and subclinical CVD (7, 8). However, studies of discrimination and CVD risk factors are not unequivocal (2, 17). This probably reflects differences in how discrimination is measured, as well as the complex interplay of personal and environmental factors that affect susceptibility to both CVD and experiences of discrimination (2). The current study adds to the literature on discrimination and CVD because it is, to our knowledge, the first large population-based study to link discrimination with increased risk of clinically adjudicated incident CVD. Both lifetime discrimination and everyday discrimination were related to increased CVD risk, with the associations for everyday discrimination limited to men. Our data suggest that repeated instances of discrimination experienced in significant life domains—typically more blatant types of discrimination—have long-lasting health effects, and that for men, everyday occurrences of unfair treatment are similarly related to CVD risk. This study did not allow us to determine when during the life course the reported instances of discrimination occurred. Presumably, cumulative experiences of blatant as well as more subtle forms of discrimination in multiple domains could have adverse socioeconomic, emotional, and behavioral consequences that over time could elevate risk of CVD and have long-lasting effects. In a recent report from the Jackson Heart Study, Sims et al. (38) found that reported lifetime discrimination, assessed across 9 important life domains, was related to prevalent hypertension regardless of racial or nonracial attributions, whereas everyday discrimination was not; they did not observe sex differences.

In a recent meta-analysis, Pascoe and Smart Richman (5) evaluated several plausible pathways between perceived discrimination and health outcomes and concluded that perceived discrimination is associated with heightened physiological responses and more negative psychological stress responses, as well as more adverse behavioral risk factor profiles. However, available evidence comes from a limited number of studies, with some methodological limitations. In our study, the associations of discrimination with CVD risk remained after adjustment for health behaviors and CVD risk factors, indicating that these factors did not account for the relationship of discrimination with incident CVD. MESA did not have baseline data on sleep quality or duration. Accruing evidence suggests a consistent and strong negative association of reported discrimination with disrupted sleep (35, 36), making this an important behavior to include in future studies.

Psychological responses to discrimination may be important to consider. In our analysis, associations between lifetime discrimination and incident CVD were reduced slightly when chronic stress burden and depressive symptoms were added to the models. In the sex-stratified analyses conducted for everyday discrimination, chronic stress burden was a stronger covariate for men and depressive symptom score was a stronger covariate for women (data not shown). This may reflect sex differences in psychological coping and symptom reporting, although we were unable to disentangle this in the current study. Specific psychological responses to discrimination were not ascertained in MESA, so we cannot discern whether reported stress or depressive symptoms were specifically linked to prior experiences of discrimination. Participants were asked 2 questions about hypothetical occurrences of unfair treatment: 1) whether they would accept it as a fact of life or try to do something about it, and 2) whether they would talk to others or keep it to themselves. Over 62% of men and women said they would try to do something about it, and over 84% of women and nearly 78% of men said they would talk to others, indicating that most respondents would use an active approach to coping; but whether coping styles affect the association of discrimination with CVD risk is unknown.

In our study, the prevalence of reported experiences of discrimination varied by race/ethnicity, but there were no racial/ethnic differences in the relationship of perceived discrimination with incident CVD. These data suggest that the levels of CVD risk conferred by exposure to discrimination itself may be similar across racial/ethnic groups, but our power to detect differences by race/ethnicity was limited. When we limited the exposure to reported racial/ethnic discrimination, no clear pattern of increased CVD risk was noted, although in exploratory analyses persons reporting both racial and nonracial discrimination had elevated CVD risk. This suggests that an overall higher level of exposure to discrimination matters more for health than does a specific type of discrimination. In a previous report from MESA, Borrell et al. (39) found that a higher prevalence of unhealthy behaviors (smoking and heavy alcohol consumption) was related to reported racial/ethnic discrimination, but other studies have not found that racial discrimination has worse health consequences than nonracial discrimination (3, 7–9, 35, 36). Black respondents in our study were more likely to report perceived discrimination and to be in the “high” exposure categories for lifetime discrimination (Table 1), which is consistent with prior work (7, 14, 40). In terms of absolute risk exposure, it is clear that the burden of discrimination is highest among African Americans compared with other racial/ethnic groups.

There are limitations to acknowledge. Although the MESA cohort is well-characterized, with a sample of more than 6,000 persons, the numbers of racial/ethnic minority (especially Chinese and Hispanic) participants are relatively small, making in-depth analysis of the types of discrimination reported by the various racial/ethnic groups difficult and limiting our ability to investigate potential heterogeneity of associations by race/ethnicity. MESA did not collect detailed information about experiences of discrimination and did not obtain repeat assessments of discrimination; further, information about coping strategies used when discrimination occurred was extremely limited. Such information would be valuable to better understand whether more recent or more distant experiences have greater health impacts and to aid our understanding of the somewhat inconsistent patterns of relationships between baseline characteristics and reported levels of lifetime discrimination (Table 1). Persons reporting discrimination in more domains had generally better sociodemographic and CVD risk profiles than their counterparts who reported no discrimination. For example, we observed that persons who reported more discrimination were younger and had more education and higher family incomes, which is consistent with prior research (41, 42) and is largely attributed to the fact that middle– and higher–socioeconomic status individuals often live and work in more integrated environments than their counterparts of lower socioeconomic status and consequently are more likely to be exposed to discrimination (42, 43). Persons reporting the highest level of discrimination also had lower blood pressure and healthier lipid profiles but had a higher body mass index, drank more alcohol, and were more likely to be current smokers. This suggests that behavioral coping strategies used in response to discrimination may be important to consider in future work.

Several strengths offset these limitations. MESA is a multiethnic, population-based cohort study, which enhances the generalizability of findings. Both lifetime experiences of discrimination in significant life domains and more day-to-day types of unfair treatment were assessed, enabling us to investigate the relationships between perceptions of both blatant and subtle types of discrimination and CVD risk. Incident CVD events were clinically adjudicated by an expert panel, allowing us to investigate discrimination in relation to a well-established and meaningful clinical cardiovascular outcome. Numerous demographic, behavioral, and CVD risk factor measures were obtained from participants, and thus we were able to investigate the relationship of discrimination with incident CVD independently of important sociodemographic characteristics and risk factors.

In sum, this study offers new epidemiologic evidence that experiences of discrimination are importantly related to incident clinical CVD outcomes in middle-aged and older adults. Moreover, our data suggest that the number of domains in which a person experiences discrimination or unfair treatment may be more important than the type of discrimination experienced. With the recurring instances of discriminatory experiences and incivilities that are prevalent in contemporary society, our findings point to the public health importance of evaluating best practices for mitigating both discrimination and its associated adverse consequences.