Black–White Differences in Cardiovascular Disease Mortality: A Prospective US Study, 2003–2017

Abstract

Objectives. To determine factors that explain the higher Black:White cardiovascular disease (CVD) mortality rates among US adults.

Methods. We analyzed data from the Reasons for Geographic and Racial Differences in Stroke study from 2003 to 2017 to estimate Black:White hazard ratios (HRs) for CVD mortality within subgroups younger than 65 years and aged 65 years or older.

Results. Among 29 054 participants, 41.0% who were Black and 54.9% who were women, 1549 CVD deaths occurred. Among participants younger than 65 years, the demographic-adjusted Black:White CVD mortality HR was 2.23 (95% confidence interval [CI] = 1.87, 2.65) and 1.21 (95% CI = 1.00, 1.47) after full adjustment. Among participants aged 65 years or older, the demographic-adjusted Black:White CVD mortality HR was 1.58 (95% CI = 1.39, 1.79) and 1.12 (95% CI = 0.97, 1.29) after full adjustment. When we used mediation analysis, socioeconomic status explained 21.2% (95% CI = 13.6%, 31.4%) and 38.0% (95% CI = 20.9%, 61.7%) of the Black:White CVD mortality risk difference among participants younger than 65 years and aged 65 years or older, respectively. CVD risk factors explained 56.6% (95% CI = 42.0%, 77.2%) and 41.3% (95% CI = 22.9%, 65.3%) of the Black:White CVD mortality difference for participants younger than 65 years and aged 65 years or older, respectively.

Conclusions. The higher Black:White CVD mortality risk is primarily explained by racial differences in socioeconomic status and CVD risk factors.

Although cardiovascular disease (CVD) mortality rates have declined over the past several decades in the United States,^1,2 studies have consistently demonstrated higher rates of CVD mortality among Black compared with White adults.^2–4 Racial differences in factors associated with increased CVD risk exist in several domains including socioeconomic status (SES),⁵ psychosocial factors,⁶ CVD risk factors (i.e., systolic blood pressure, total and high-density lipoprotein cholesterol) included in the Pooled Cohort risk equations (used to determine 10-year risk for a CVD event),¹ and other clinical risk factors (e.g., body mass index) with a higher prevalence among Blacks compared with Whites.⁷ These differences may contribute to the disparity in CVD mortality.

Although previous studies have found higher rates of CVD mortality for Black compared with White adults,^2–4 they have not identified reasons for this difference. Understanding reasons for these differences is useful because race is a category that is socially and historically constructed, with little of the contribution to differences between Blacks and Whites in health outcomes resulting from genetic factors.⁸ Identifying modifiable factors contributing to higher Black:White CVD mortality rates has the potential to provide policymakers and clinicians information to prioritize directions for interventions aimed at eliminating this difference.

The Williams model of racial differences in health provides a useful framework to study the unequal Black:White CVD mortality rate.⁸ This model proposes that certain basic causes such as historical, political, and legal structures in the United States lead to racial differences in SES.⁸ SES then influences exposure to other nonbiological causes of poor health such as psychosocial factors and access to medical care. This leads to differences in biological processes (e.g., CVD risk factor development) and ultimately differences in health outcomes (e.g., CVD mortality).⁸ The purpose of this study was to compare CVD mortality rates in Black and White adults, and to determine whether the higher prevalence of low SES, psychosocial factors, and CVD and other clinical risk factors among Blacks compared with Whites explained the differences. Also, we determined the percentage of the difference in Black:White CVD mortality explained by each of these categories by using mediation analysis. For comparison, we also conducted analyses to determine if a racial difference in non-CVD mortality was present and to identify risk factors that might explain this difference.

METHODS

The Reasons for Geographic and Racial Differences in Stroke (REGARDS) study enrolled a population-based cohort of 30 239 community-dwelling, non-Hispanic, Black and White adults aged 45 years or older from across the United States between 2003 and 2007. It was designed to examine reasons for the higher rate of stroke mortality among Blacks compared with Whites and among residents of the Southeastern United States compared with the rest of the contiguous United States.^9,10 Residents of the Stroke Buckle (coastal plain region of North Carolina, South Carolina, and Georgia) and the Stroke Belt (remainder of North Carolina, South Carolina, and Georgia, and Alabama, Mississippi, Tennessee, Arkansas, and Louisiana) regions of the United States and Black adults were oversampled. The outcome of coronary heart disease is being investigated in an ancillary study. The final analytic sample included 29 054 participants. Exclusions are listed in Table A (available as a supplement to the online version of this article at http://www.ajph.org).

Baseline Data Collection

Information on medical history and health status were collected at baseline with a computer-assisted telephone survey. This was followed by an in-home examination during which blood pressure, height, and weight were measured; an electrocardiogram was performed; and fasting blood and urine samples were collected following standardized protocols. Race, age, sex, education, income, health insurance status, relationship status, current cigarette smoking, antihypertensive medication use, antihyperglycemic medication use, insulin use, and aspirin use were self-reported.⁹ The definitions and measurement protocols for history of CVD, depressive symptoms, stress, systolic and diastolic blood pressure, total and high-density lipoprotein cholesterol and triglycerides, high-sensitivity C-reactive protein, diabetes, body mass index (weight in kilograms divided by the square of height in meters), albumin-to-creatinine ratio, and estimated glomerular filtration rate are presented in Table B (available as a supplement to the online version of this article at http://www.ajph.org).

Identification and Adjudication of Deaths

Participant deaths were detected through online sources (e.g., the Social Security Death Index), the National Death Index, and by report of deceased study participants’ proxies or next of kin during semiannual follow-up calls. Cause of death was adjudicated with information from death certificates, medical records, and autopsy reports. All cases were reviewed independently by 2 clinician-adjudicators, and disagreements were resolved by committee (Table B).¹¹ The main underlying cause of death was defined as the 1 disease or injury that precipitated the events resulting in death. CVD-related mortality included sudden cardiac death, myocardial infarction, heart failure, stroke, other cardiac, not cardiac but other CVD (e.g., ruptured aortic aneurysm), and pulmonary embolism. Non–CVD-related mortality included all other causes of death (i.e., cancer; infection; chronic lung disease; other noncardiac or nonstroke death; unclassified death; dementia; accident, injury, suicide, or homicide; end-stage renal disease; and liver disease). For the current analysis, participants were followed from their baseline examination, 2003 to 2007, to their death or December 31, 2016, for those who did not die.

Statistical Analyses

We conducted all analyses for participants younger than 65 years and aged 65 years or older separately for 3 main reasons: (1) Black adults develop risk factors for CVD mortality at earlier ages compared with White adults,¹² (2) previous research reports a decrease in Black:White mortality disparities as age increases,^5,13 and (3) focusing on populations younger than 65 years and aged 65 years or older separately allows for different policy implications. We also tested the statistical interaction between race and age group (aged < 65 years and ≥ 65 years separately) in an overall model and found statistically significant racial differences in CVD and non-CVD mortality by age group (P < .001). We calculated participant characteristics and CVD and non-CVD mortality rates for Blacks and Whites separately. We calculated cumulative incidence curves for CVD and non-CVD mortality separately among adults younger than 65 years and those aged 65 years or older (Figure A, available as a supplement to the online version of this article at http://www.ajph.org).

Utilizing cause-specific hazards to account for competing risks, we calculated hazard ratios (HRs) for CVD and non-CVD mortality separately, comparing Black and White participants. To include participants with incomplete data (Table C, available as supplement to the online version of this article at http://www.ajph.org), we imputed missing covariates with 20 data sets by using chained equations.¹⁴ The base model included adjustment for sociodemographic covariates including age, sex, and region of residence. The fully adjusted model included adjustment for SES (education, income, and insurance status), psychosocial factors (relationship status, depressive symptoms, and stress), CVD risk factors used in the Pooled Cohort risk equations (systolic blood pressure, antihypertensive medication use, diabetes, cigarette smoking, total and high-density lipoprotein cholesterol), and additional clinical risk factor biological processes variables of disease severity (high-sensitivity C-reactive protein, albuminuria, body mass index, statin use, insulin use, aspirin use, and history of CVD). We repeated this analysis for the 5 most common causes of death in the current study and stratified by history of CVD. We also conducted a post hoc analysis, testing for evidence of a race-by-age (modeled as a continuous variable) interaction on CVD and non-CVD mortality separately among participants younger than 65 years and those aged 65 years or older.

We conducted mediation analyses for multiple mediators by using the inverse odds ratio weighting methodology.¹⁵ We created multiple mediator categories (SES, psychosocial factors, CVD risk factors, and additional clinical risk factors) by using the Williams model of racial differences in health as a guide.⁸ Briefly, this mediation utilized predicted probabilities of exposure (i.e., race) based on each mediator covariate group and age, sex, and region of residence.¹⁵ We estimated the β coefficient for the association of race with mortality from a Cox regression model with and without weighting. This allows for the decomposition of total effects of the exposure variable on the outcome into direct and indirect effects. We calculated the weights by the inverse of the odds of exposure.¹⁶ We estimated the percent mediation as the percent change between the β coefficients with and without weights applied. Confidence intervals (CIs) for these percent changes were nonparametrically estimated from the 2.5th and 97.5th percentiles from the distribution of 2000 bootstrapped samples.¹⁵ We conducted analyses with SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

Black participants were less likely to be male and to live in the Stroke Buckle compared with Whites (Table 1). Compared with White participants, Black participants were more likely to have less than a high-school education, a household income less than $20 000 per year, no health insurance, depressive symptoms, and high stress. Also, Black participants were more likely than were White participants to not be living with a significant other, to smoke cigarettes, to be taking antihypertensive medication, to have diabetes and be taking insulin, and to have a high-sensitivity C-reactive protein greater than 3 milligrams per liter, an albumin-to-creatinine ratio greater than 30 milligrams per gram, and an estimated glomerular filtration rate less than 60 milliliters per minute per 1.73 square meters. Mean systolic blood pressure was higher among Black compared with White participants.

TABLE 1—

Baseline Characteristics of Black and White Reasons for Geographic and Racial Differences in Stroke Study Participants Younger Than 65 Years and Aged 65 Years or Older: United States, 2003–2007

	Aged < 65 Years, Mean (SD) or %		Aged ≥ 65 Years, Mean (SD) or %
Characteristics	Black (n = 6432)	White (n = 8266)	Black (n = 5498)	White (n = 8867)
Age, y	57.1 (5.0)	57.4 (4.9)	72.3 (5.7)	72.9 (5.9)
Male	36.8	46.8	39.5	53.0
Region of residencea
Non–Stroke Belt or Buckle	43.9	40.7	54.6	42.0
Stroke Belt	36.2	35.7	29.5	35.4
Stroke Buckle	19.9	23.5	15.8	22.6
Socioeconomic status factors
Less than high-school education	12.8	5.3	28.0	9.0
Income < $20 000/y	25.6	10.0	37.8	17.1
No health insurance	16.2	8.1	2.0	0.8
Psychosocial factors
Not living with a significant other	48.0	22.7	56.8	36.1
Depressive symptoms	15.9	10.5	11.5	7.3
High stress	14.7	9.6	11.4	5.5
CVD risk factor biological processes
Cigarette smoking	21.7	16.6	12.4	8.6
SBP, mm Hg	129.1 (16.9)	122.3 (14.9)	132.7 (17.7)	128.2 (16.1)
Total cholesterol, mg/dL	194.8 (41.0)	196.5 (39.2)	190.9 (41.0)	186.7 (39.1)
HDL cholesterol, mg/dL	52.9 (15.8)	50.8 (16.2)	54.2 (16.1)	50.6 (16.3)
Antihypertensive medication use	59.0	35.3	69.1	51.4
Diabetes	28.2	13.7	33.0	17.6
Other clinical risk factor biological processes
Insulin use	10.3	4.9	12.3	5.4
Statin use	24.7	26.9	34.0	39.1
Aspirin use	32.9	39.1	44.7	53.8
BMI, kg/m2	31.6 (7.0)	29.0 (6.0)	29.8 (6.2)	27.6 (5.0)
History of CVD	22.9	18.3	35.5	36.1
High-sensitivity CRP > 3 mg/L	51.2	35.2	46.5	34.6
ACR > 30 mg/g	16.7	8.6	23.1	15.6
eGFR < 60 mL/min/1.73 m2	6.2	3.4	19.2	18.1

Note. ACR = albumin-to-creatinine; BMI = body mass index (weight in kilograms divided by the square of height in meters [kg/m²]); CRP = c-reactive protein; CVD = cardiovascular disease; eGFR = estimated glomerular filtration rate; HDL = high-density lipoprotein; mg/dL = milligrams per deciliter; mg/g = milligrams per gram; mg/L = milligrams per liter; mL/min = milliliters per minute; mm Hg = millimeters of mercury; SBP = systolic blood pressure. The sample size was n = 29 054.

^aThe Stroke Buckle comprises the coastal plain region of North Carolina, South Carolina, and Georgia; the Stroke Belt is the remainder of North Carolina, South Carolina, and Georgia, and Alabama, Mississippi, Tennessee, Arkansas, and Louisiana.

Disease Mortality Rates

There were 1549 CVD and 4108 non-CVD deaths in the current study (Table D, available as a supplement to the online version of this article at http://www.ajph.org). Mean follow-up time for CVD mortality among adults younger than 65 years and for non-CVD mortality among adults younger than 65 years and those aged 65 years or older was 8.9 years (maximum 13.9 years). Follow-up time for CVD mortality among adults aged 65 years or older was limited to 6.0 years (maximum 6.9 years) to avoid violating the proportional hazards assumption. Sudden death and myocardial infarction accounted for the majority of CVD deaths while cancer and infection accounted for the majority of non-CVD deaths. Among participants younger than 65 years, the CVD mortality rate was 5.6 (95% CI = 5.0, 6.2) and 2.7 (95% CI = 2.4, 3.1) per 1000 person-years among Black and White participants, respectively (Table 2). For participants aged 65 years or older, the CVD mortality rate was 14.2 (95% CI = 12.9, 15.5) and 10.5 (95% CI = 9.7, 11.4) among Black and White participants, respectively. Non-CVD related mortality rates were 9.0 (95% CI = 8.2, 9.8) and 6.4 (95% CI = 5.9, 7.0) per 1000 person-years among Black and White participants younger than 65 years, respectively, and 25.3 (95% CI = 23.9, 26.8) and 24.9 (95% CI = 23.8, 26.0) among Black and White participants aged 65 years or older, respectively. Table E (available as a supplement to the online version of this article at http://www.ajph.org) presents mortality rates for subtypes of CVD and non-CVD mortality.

TABLE 2—

Hazard Ratios for Cardiovascular and Noncardiovascular Disease Mortality Comparing Black With White Participants Younger Than 65 Years and Aged 65 Years or Older: United States, 2003–2017

	Sample Size		Mortality Rate per 1000 Person-Years (95% CI)		Black vs White, HR (95% CI)
Cause of Death	Black	White	Black	White	Sociodemographic-Adjusteda	Fully Adjustedb
CVDc
Aged < 65 y	6432	8266	5.6 (5.0, 6.2)	2.7 (2.4, 3.1)	2.23 (1.87, 2.65)	1.21 (1.00, 1.47)
Aged ≥ 65 yd	5498	8867	14.2 (12.9, 15.5)	10.5 (9.7, 11.4)	1.58 (1.39, 1.79)	1.12 (0.97, 1.29)
Non-CVDe
Aged < 65 y	6432	8266	9.0 (8.2, 9.8)	6.4 (5.9, 7.0)	1.51 (1.33, 1.71)	0.90 (0.78, 1.03)
Aged ≥ 65 y	5498	8867	25.3 (23.9, 26.8)	24.9 (23.8, 26.0)	1.16 (1.08, 1.25)	0.88 (0.81, 0.96)

Note. CI = confidence interval; CVD = cardiovascular disease; HR = hazard ratio. The sample size was n = 29 054.

^aSociodemographic-adjusted model included adjustment for age, sex, and region of residence.

^bFully adjusted model included sociodemographic covariates and additional adjustment for socioeconomic factors (less than high-school education, income < $20 000 per year, and insurance status), psychosocial factors (not living with a significant other, depressive symptoms, and stress), CVD risk factors (cigarette smoking, systolic blood pressure, total and high-density lipoprotein cholesterol, antihypertensive medication use, and diabetes), and other clinical risk factors (insulin use, statin use, aspirin use, body mass index, history of cardiovascular disease, high sensitivity C-reactive protein > 3 mg/L, albumin-to-creatinine > 30 mg/g, and estimated glomerular filtration rate < 60 mL/min/1.73 m²).

^cCVD-related mortality included sudden cardiac death, myocardial infarction, heart failure, stroke, other cardiac, not cardiac but other cardiovascular (e.g., ruptured aortic aneurysm), and pulmonary embolism.

^dFor CVD mortality among adults aged ≥ 65 y, the proportional hazards regression model was calculated with a mean follow-up time of 6.0 y (maximum 6.9 y). Mean follow-up time for CVD mortality among adults aged < 65 y and for non-CVD mortality among adults aged < 65 y and ≥ 65 y was 8.9 y (maximum 13.9 y).

^eNon–CVD-related mortality included cancer; infection; chronic lung disease; other noncardiac or nonstroke death; unclassified death; dementia; accident, injury, suicide, or homicide; end-stage renal disease; and liver disease.

Black:White Hazard Ratios for Mortality

Among participants younger than 65 years, the sociodemographic- (age, sex, and region) adjusted Black:White HR for CVD mortality was 2.23 (95% CI = 1.87, 2.65; Table 2). The fully adjusted Black:White HR for CVD mortality was 1.21 (95% CI = 1.00, 1.47). For participants aged 65 years or older, the sociodemographic- and fully adjusted Black:White HRs for CVD mortality were 1.58 (95% CI = 1.39, 1.79) and 1.12 (95% CI = 0.97, 1.29), respectively. For non-CVD mortality, among participants younger than 65 years, the sociodemographic-adjusted and fully adjusted Black:White HRs were 1.51 (95% CI = 1.33, 1.71) and 0.90 (95% CI = 0.78, 1.03; Table 2). Among participants aged 65 years or older, the sociodemographic and fully adjusted Black:White HRs for non-CVD mortality were 1.16 (95% CI = 1.08, 1.25) and 0.88 (95% CI = 0.81, 0.96), respectively. Table F (available as a supplement to the online version of this article at http://www.ajph.org) presents results progressively adjusted for sociodemographic factors, SES, psychosocial factors, CVD risk factors, and clinical risk factors. Tables G and H (available as supplements to the online version of this article at http://www.ajph.org) present HRs for the association of individual covariates from the fully adjusted models with CVD mortality and non-CVD mortality, respectively. Table I (available as a supplement to the online version of this article at http://www.ajph.org) presents HRs for CVD and non-CVD mortality stratified by history of CVD.

In posthoc analysis testing for evidence of a race-by-age (modeled as a continuous variable) interaction on CVD and non-CVD mortality separately, the association of Black compared with White race with increased CVD mortality risk was weaker as age increased among those aged 65 years or older (P < .001), but there was no evidence of effect modification for those younger than 65 years (P = .191). For non-CVD mortality, among adults younger than 65 years, the association of Black compared with White race with increased non-CVD mortality risk was weaker as age increased (P < .001), but there was no evidence of effect modification for those aged 65 years or older (P = .197).

For individual causes of CVD mortality, the largest sociodemographic-adjusted Black:White HR among adults younger than 65 years was for stroke (2.83; 95% CI = 1.78, 4.50; and fully adjusted HR of 1.61; 95% CI = 0.97, 2.69), and among adults aged 65 years or older it was for myocardial infarction (1.78; 95% CI = 1.46, 2.18; and fully adjusted HR of 1.26; 95% CI = 1.00, 1.58; Table 3). For non-CVD mortality, in both age groups, the largest sociodemographic-adjusted Black:White HR was for infection among adults younger than 65 years (1.80; 95% CI = 1.33, 2.44; and fully adjusted HR of 0.85; 95% CI = 0.61, 1.20) and aged 65 years or older (1.31; 95% CI = 1.11, 1.55; and fully adjusted HR of 0.89; 95% CI = 0.74, 1.08). Table J (available as a supplement to the online version of this article at http://www.ajph.org) presents results for individual causes of mortality progressively adjusted for sociodemographic factors, SES, psychosocial factors, CVD risk factors, and clinical risk factors.

TABLE 3—

Hazard Ratios for Leading Causes of Mortality Comparing Black With White Participants in the Reasons for Geographic and Racial Differences in Stroke Study by Subgroups of Participants Younger Than 65 Years and Aged 65 Years and Older: United States, 2003–2017

	Mortality Rate per 1000 Person-Years (95% CI)		Black vs White, HR (95% CI)
Cause of Death	Black	White	Sociodemographic-Adjusteda	Fully Adjustedb
Cancer
Aged < 65 y (n = 14 698)	4.5 (3.9, 5.0)	3.2 (2.8, 3.6)	1.50 (1.26, 1.79)	1.02 (0.84, 1.24)
Aged ≥ 65 y (n = 14 356)	10.0 (9.1, 10.9)	10.1 (9.4, 10.8)	1.10 (0.98, 1.24)	0.95 (0.83, 1.08)
Infection
Aged < 65 y (n = 14 698)	1.6 (1.3, 2.0)	1.0 (0.8, 1.2)	1.80 (1.33, 2.44)	0.85 (0.61, 1.20)
Aged ≥ 65 y (n = 14 356)	5.2 (4.6, 5.9)	4.7 (4.3, 5.2)	1.31 (1.11, 1.55)	0.89 (0.74, 1.08)
Sudden death
Aged < 65 y (n = 14 698)	1.8 (1.4, 2.1)	0.9 (0.7, 1.1)	2.08 (1.53, 2.81)	1.24 (0.88, 1.74)
Aged ≥ 65 yc (n = 14 356)	3.7 (3.0, 4.4)	2.7 (2.2, 3.1)	1.46 (1.19, 1.79)	1.04 (0.83, 1.32)
Myocardial infarction
Aged < 65 y (n = 14 698)	1.3 (1.0, 1.5)	0.8 (0.6, 1.0)	1.65 (1.18, 2.32)	0.79 (0.54, 1.16)
Aged ≥ 65 yc (n = 14 356)	3.9 (3.2, 4.6)	2.5 (2.0, 2.9)	1.78 (1.46, 2.18)	1.26 (1.00, 1.58)
Stroke
Aged < 65 y (n = 14 698)	0.9 (0.7, 1.2)	0.4 (0.2, 0.5)	2.83 (1.78, 4.50)	1.61 (0.97, 2.69)
Aged ≥ 65 yc (n = 14 356)	2.2 (1.7, 2.7)	1.9 (1.6, 2.3)	1.09 (0.85, 1.39)	0.89 (0.68, 1.18)

Note. CI = confidence interval; HR = hazard ratio. The sample size was n = 29 054.

^aSociodemographic-adjusted model included adjustment for age, sex, and region of residence.

^bFully adjusted model included sociodemographic covariates and additional adjustment for socioeconomic factors (less than high-school education, income < $20 000 per year, and insurance status), psychosocial factors (not living with a significant other, depressive symptoms, and stress), CVD risk factors (cigarette smoking, systolic blood pressure, total and high-density lipoprotein cholesterol, antihypertensive medication use, and diabetes), and other clinical risk factors (insulin use, statin use, aspirin use, body mass index, history of cardiovascular disease, high-sensitivity C-reactive protein > 3 mg/L, albumin-to-creatinine > 30 mg/g, and estimated glomerular filtration rate < 60 mL/min/1.73 m²).

^cProportional hazards regression models calculated with 6.0 y (maximum 6.9 y) of follow-up among adults aged ≥ 65 y for sudden death, myocardial infarction, and stroke. For all other mortality outcomes and age groups, mean follow-up time was 8.9 y (maximum 13.9 y).

Mediation Analysis Comparing Black:White Mortality

Among participants younger than 65 years and those aged 65 years and older, SES explained 21.2% (95% CI = 13.6%, 31.4%) and 38.0% (95% CI = 20.9%, 61.7%), respectively, of the Black:White difference in CVD mortality (Table 4). CVD risk factors explained 56.6% (95% CI = 42.0%, 77.2%) and 41.3% (95% CI = 22.9%, 65.3%) of the difference in CVD mortality for adults younger than 65 years and those aged 65 years or older, respectively. All risk factors combined explained 76.4% (95% CI = 56.5%, 107.5%) and 61.1% (95% CI = 31.5%, 97.8%) of the excess risk for Black:White CVD mortality among participants younger than 65 years and those aged 65 years or older, respectively. For both age groups, the excess risk for non-CVD mortality was completely explained when we assessed all risk factors investigated simultaneously (Table 4).

TABLE 4—

Percentage of the Excess Cardiovascular and Non–Cardiovascular Disease Mortality Experienced Among Blacks Compared With Whites Explained by Socioeconomic Factors, Psychosocial Factors, Cardiovascular Disease Risk Factors, and Other Clinical Risk Factors Among Participants Younger Than 65 Years and Aged 65 Years or Older: United States, 2003–2017

	Mediators
Age and Effect of Race on Mortality	Socioeconomic Factorsa	Psychosocial Factorsb	CVD Risk Factorsc	Other Clinical Risk Factorsd	All Risk Factors
CVD mortalitye
Age < 65 y
Indirect effect, HR (95% CI)	1.18 (1.12, 1.26)	1.15 (1.08, 1.22)	1.57 (1.43, 1.73)	1.40 (1.29, 1.52)	1.84 (1.61, 2.12)
Direct effect, HR (95% CI)	1.88 (1.57, 2.26)	1.94 (1.63, 2.36)	1.42 (1.16, 1.72)	1.59 (1.33, 1.92)	1.21 (0.95, 1.49)
Total effect, HR (95% CI)	2.23 (1.88, 2.69)	2.23 (1.88, 2.69)	2.23 (1.88, 2.69)	2.23 (1.88, 2.69)	2.23 (1.88, 2.69)
% mediated (95% CI)	21.2 (13.6, 31.4)	17.2 (9.8, 27.1)	56.6 (42.0, 77.2)	41.9 (30.3, 57.4)	76.4 (56.5, 107.5)
Age ≥ 65 yf
Indirect effect, HR (95% CI)	1.19 (1.10, 1.28)	1.13 (1.08, 1.19)	1.21 (1.12, 1.31)	1.18 (1.11, 1.27)	1.32 (1.16, 1.50)
Direct effect, HR (95% CI)	1.33 (1.15, 1.54)	1.40 (1.22, 1.61)	1.31 (1.12, 1.51)	1.34 (1.17, 1.53)	1.19 (1.01, 1.43)
Total effect, HR (95% CI)	1.58 (1.39, 1.79)	1.58 (1.39, 1.79)	1.58 (1.39, 1.79)	1.58 (1.39, 1.79)	1.58 (1.39, 1.79)
% mediated (95% CI)	38.0 (20.9, 61.7)	26.4 (14.6, 42.2)	41.3 (22.9, 65.3)	36.4 (21.4, 57.1)	61.1 (31.5, 97.8)
Non-CVD mortalityg
Age < 65 y
Indirect effect, HR (95% CI)	1.17 (1.12, 1.23)	1.14 (1.09, 1.20)	1.24 (1.15, 1.34)	1.15 (1.09, 1.22)	1.48 (1.33, 1.65)
Direct effect, HR (95% CI)	1.28 (1.13, 1.47)	1.32 (1.15, 1.51)	1.22 (1.06, 1.42)	1.31 (1.14, 1.51)	1.02 (0.86, 1.19)
Total effect, HR (95% CI)	1.51 (1.33, 1.71)	1.51 (1.33, 1.71)	1.51 (1.33, 1.71)	1.51 (1.33, 1.71)	1.51 (1.33, 1.71)
% mediated (95% CI)	39.1 (25.2, 59.2)	32.2 (18.2, 51.4)	51.8 (30.4, 82.5)	34.8 (19.8, 56.6)	95.8 (64.2, 147.6)
Age ≥ 65 y
Indirect effect, HR (95% CI)	1.13 (1.08, 1.19)	1.04 (1.01, 1.07)	1.11 (1.06, 1.16)	1.07 (1.03, 1.11)	1.20 (1.12, 1.30)
Direct effect, HR (95% CI)	1.02 (0.94, 1.11)	1.12 (1.03, 1.21)	1.04 (0.96, 1.14)	1.08 (1.00, 1.17)	0.96 (0.87, 1.06)
Total effect, HR (95% CI)	1.16 (1.08, 1.25)	1.16 (1.08, 1.25)	1.16 (1.08, 1.25)	1.16 (1.08, 1.25)	1.16 (1.08, 1.25)
% mediated (95% CI)	85.8 (47.6, 181.4)	25.3 (3.0, 63.6)	70.9 (34.3, 145.9)	45.9 (19.8, 104.0)	126.7 (68.5, 267.8)

Note. CI = confidence interval; CVD = cardiovascular disease; HR = hazard ratio.

^aSocioeconomic factors were less than high-school education, income < $20 000 per year, and insurance status.

^bPsychosocial factors were not living with a significant other, depressive symptoms, and stress.

^cCVD risk factors were cigarette smoking, systolic blood pressure, total and high-density lipoprotein cholesterol, antihypertensive medication use, and diabetes.

^dOther clinical risk factors were insulin use, statin use, aspirin use, body mass index, history of cardiovascular disease, high-sensitivity C-reactive protein > 3 mg/L, albumin-to-creatinine > 30 mg/g, and estimated glomerular filtration rate < 60 mL/min/1.73 m².

^eCVD-related mortality included sudden cardiac death, myocardial infarction, heart failure, stroke, other cardiac, not cardiac but other cardiovascular (e.g., ruptured aortic aneurysm), and pulmonary embolism.

^fFor CVD mortality among adults aged ≥ 65 y, the mediation analysis was calculated with a mean follow-up time of 6.0 y (maximum 6.9 y). Mean follow-up time for CVD mortality among adults aged < 65 y and for non-CVD mortality among adults aged < 65 years and ≥ 65 years was 8.9 y (maximum 13.9 y).

^gNon–CVD-related mortality included cancer; infection; chronic lung disease; other noncardiac or nonstroke death; unclassified death; dementia; accident, injury, suicide, or homicide; end-stage renal disease; and liver disease.

DISCUSSION

In the current study, CVD mortality rates were higher among Black compared with White adults. The Black:White HR for CVD mortality was larger for participants younger than 65 years compared with participants aged 65 years or older. SES and modifiable CVD risk factors explained a substantial amount of the excess CVD mortality among Black compared with White participants. There was also increased non-CVD mortality risk among Black compared with White participants.

Previous studies have reported an increased risk of mortality among Blacks compared with Whites for CVD mortality overall in the United States.^2–4 The current study adds to these findings by documenting the presence of a Black:White difference in CVD mortality in a contemporary cohort and identifying factors that explain this difference. Similar to previous studies,^10,13,17,18 Black:White differences were present for the most common subtypes of CVD mortality in the current study: sudden death, myocardial infarction, and stroke. The increased Black:White risk for sudden death was similar to a recent study that used data from the Atherosclerosis Risk in Communities (ARIC) study and may be explained by increased prevalence of CVD risk factors, particularly hypertension.¹⁷ However, factors related to SES such as longer ambulance response time, lower likelihood of bystander cardiopulmonary resuscitation, and limited access to high-performing hospitals have also been reported as reasons for the risk difference.¹⁷

The Black:White HR for CVD mortality was larger for participants younger than 65 years compared with those aged 65 years or older. Exposure to low SES and the earlier development of CVD risk factors among Black adults may explain this finding.^1,12 In the Coronary Artery Risk Development in Young Adults (CARDIA) study, the cumulative incidence of hypertension after 20 years of follow-up among Black women and men was 37.6% and 34.5%, respectively, and among White women and men was 12.3% and 21.4%, respectively.¹² The finding that the Black:White relative risk for CVD mortality was smaller at older age is consistent with previous studies.^5,13

Providing health insurance, a nonbiological cause of health disparities in the Williams model, among Black adults younger than 65 years may be a valuable strategy to reduce CVD mortality through the pathway of earlier detection and treatment of CVD risk factors.¹⁹ Implementation of the Affordable Care Act increased the number of Black adults with insurance and access to care in the United States by 7% between 2011 and 2014.²⁰ Although the US government provides Medicare insurance for adults aged 65 years or older in the United States, Black adults in this age group are more than twice as likely to report not being able to see a doctor because of cost.²¹ The prevalence of low income among Black adults in the current study was more than twice that of White participants. To decrease the Black:White difference in CVD mortality among adults aged 65 years or older, policymakers should consider providing subsidies for Medicare beneficiaries with low income who are not eligible for Medicaid.

Using mediation analysis, we identified CVD risk factors and SES as main contributors to the Black:White difference in CVD mortality risk. These data suggest that preventing the development of CVD risk factors could substantially reduce the Black:White disparity in CVD mortality. For diabetes prevention, cultural tailoring of the Diabetes Prevention Program, an effective weight loss and exercise intervention that lowers risk of diabetes development among adults with prediabetes, has shown promise, and the Diabetes Prevention Program has been effectively implemented in community-based locations including predominantly Black churches.²² For treatment, antihypertensive medication and statin therapy both reduce CVD risk^23,24 and have been shown to be cost-effective.^25,26 However, control of blood pressure and use of statin therapy is less common among Black adults compared with White adults.^27,28 Systems-level interventions may decrease this disparity in care. For example, the Kaiser Permanente health system in Southern California implemented an integrated health system model for hypertension treatment that resulted in control rates of greater than 80% among Black patients.²⁹ This is in stark contrast with national blood pressure control rates of 64.7% among Black adults taking antihypertensive medication.²⁷

Low SES is associated with an increased risk for development of CVD risk factors and CVD events.¹ In the current study, racial differences in SES explained a substantial proportion of the excess CVD mortality risk experienced among Blacks compared with Whites. This finding supports the pathway of the effect of race on health described in the Williams model wherein social systems of inequality contribute to lower SES for Blacks compared with Whites.⁸ Subsequently, limited resources and exposure to negative health risk factors among individuals with low SES have an effect on biological processes and ultimately health outcomes.⁸ Policy and intervention strategies such as housing voucher programs may result in improvements in health for individuals with low SES.³⁰ Voucher programs designed to allow relocation from high-poverty to low-poverty neighborhoods have demonstrated long-term impact on financial and educational attainment.³⁰ Also, CARDIA study participants who moved from high-poverty to low-poverty neighborhoods and did not move back had a 5-millimeters-of-mercury decrease in systolic blood pressure compared with those who remained in high-poverty neighborhoods.³¹ Previous studies reported that Black adults who live in integrated neighborhoods have similar prevalence of CVD risk factors to Whites.⁷ However, housing segregation and discrimination still occurs even for Black adults with financial means.³²

In the current study, risk of non-CVD mortality was also higher for Black compared with White participants. Two factors that explained a large percentage of the difference in CVD mortality, CVD risk factors and SES, also explained a large portion of the difference in non-CVD mortality. One explanation for this finding could be that the leading cause of non-CVD mortality in the current study, infection, is often a result of hospital admissions because of CVD comorbidities.³³ In the National Health and Nutrition Examination Survey Mortality Study, the adjusted relative risk for infection-related mortality was 2.0 and 2.8 for adults with versus without diabetes and with versus without heart failure, respectively.³³ Higher body mass index is associated with an increased risk of cancer.³⁴ Therefore, it is possible that preventing and treating the development of CVD risk factors could result in reductions in the Black:White difference in non-CVD mortality.

Strengths and Limitations

Strengths of the current study include use of adjudicated cause of death and the enrollment of Black and White adults from across the United States. Information on SES and clinical risk factors were collected via standardized procedures. A large number of mortality events allowed for investigation of CVD and non-CVD causes of mortality separately.

However, several limitations should be considered. Although the REGARDS study identified cause of death through a structured adjudication process, the gold standard for classifying events, it is possible that the cause of some mortality outcomes could have been misclassified. Risk factors were only available at baseline. The assessment of depressive symptoms and stress used in the current study is based on the previous week and month, respectively, before measurement and was only conducted once during baseline. Previous studies have reported stronger associations between psychological factors assessed over longer periods of time, rather than once, and outcomes.³⁵ Therefore, the association of these psychological measures with the racial difference in Black:White mortality outcomes may be underestimated. The REGARDS study contains participants from all 48 continental US states and Washington, DC; however, it is not a nationally representative sample because it only includes Whites and Blacks.

Public Health Implications

Higher CVD mortality rates for Black compared with White adults remain despite an overall decrease in CVD mortality rates over the past several decades in the United States.^2–4 In the current study, differences in SES and CVD risk factors explained a substantial proportion of the mortality difference. Additional efforts toward the prevention and treatment of modifiable CVD risk factors including hypertension, dyslipidemia, diabetes, and cigarette smoking among Black adults are needed to reduce the Black:White difference in CVD mortality. Implementing national policies aimed at addressing social determinants of health could also potentially decrease the Black:White difference in CVD mortality risk.