Racial Differences in Sudden Cardiac Death: The Atherosclerosis Risk in Communities Study (ARIC)

Abstract

Background:

African Americans (AAs) have a higher incidence of out-of-hospital sudden cardiac death (SCD) compared with Whites. However, the racial differences in the cumulative risk of SCD and the reasons for these differences have not been assessed in large-scale community-based cohorts. The objective of this study is to compare the lifetime cumulative risk of SCD among AAs and Whites, and to evaluate the risk factors that may explain racial differences in SCD risk in the general population.

Methods:

Cohort study of 3,832 AAs and 11,237 Whites participating in the Atherosclerosis Risk in Communities (ARIC) Study. Race was self-reported. SCD was defined as a sudden pulseless condition from a cardiac cause in a previously stable individual and SCD cases were adjudicated by an expert committee. Cumulative incidence was computed using competing risk models. Potential mediators included demographic and socioeconomic factors, cardiovascular risk factors, presence of coronary heart disease and electrocardiographic parameters as time-varying factors.

Results:

The mean (SD) age was 53.6 (5.8) for AAs and 54.4 (5.7) years for Whites. During 27.4 years of follow-up, 215 AAs and 332 Whites experienced SCD. The lifetime cumulative incidence of SCD at age 85 was 9.6, 6.6, 6.5, and 2.3% for AA men, AA women, white men and white women, respectively. The sex-adjusted hazard ratio (HR) for SCD comparing AAs with Whites was 2.12 (95% CI 1.79, 2.51). The association was attenuated but still statistically significant in fully adjusted models (HR 1.38, 95% CI 1.11, 1.71). In mediation analysis, known factors explained 65.3% (37.9, 92.8%) of the excess risk of SCD in AAs compared to Whites. The single most important factor explaining this difference was income (50.5%), followed by education (19.1%), hypertension (22.1%), diabetes (19.6%). Racial differences were evident in both genders, but stronger in women than in men.

Conclusions:

AAs had a much higher risk for SCD compared to Whites, particularly among women. Income, education, and traditional risk factors explained approximately 65% of the race difference in SCD. The high burden of SCD and the racial-gender disparities observed in our study represent a major public health and clinical problem.

Introduction

Despite the decline in cardiovascular mortality over the past 5 decades, sudden cardiac death (SCD) is still a leading cause of mortality in developed countries, and is often the first manifestation of cardiovascular disease.^1–3 African Americans (AAs) have a higher incidence of out-of-hospital sudden cardiac arrest compared with Whites, as evidenced in emergency medical service systems data in several US cities.^4–8 In addition, in some but not all studies, survival from sudden cardiac arrest is worse in AAs compared with Whites.^9–11 The association between race and SCD has been further examined in clinical studies of participants with pre-existing conditions, such as hypertension.¹² Nevertheless, the racial differences in the cumulative risk of SCD and the reasons for these differences have not been assessed in large-scale community-based cohorts.

The main objective of this paper was to estimate the lifetime cumulative risk of SCD among AAs and Whites in the Atherosclerosis Risk in Communities (ARIC) study, a long-term community-based cohort study of cardiovascular disease and its risk factors. In addition, we used detailed information collected repeatedly in ARIC’s in-person visits at baseline and over follow-up to identify factors that could explain the differences in SCD between AAs and Whites.

Methods

Study population

The ARIC Study data are available through the National Center of Biotechnology Information’s database of Genotypes and Phenotypes (dbGaP)¹³ and the National Heart, Lung, and Blood Institute’s Biological Specimen and Data Repository Information Coordinating Center (BioLINCC).¹⁴ The ARIC Study is a multicenter prospective cohort study designed to investigate risk factors for atherosclerosis and cardiovascular disease. Details of the study design and procedures have been described previously.¹⁵ Briefly, 15,792 men and women aged 45 to 64 years were randomly selected from four US communities at baseline (Visit 1, 1987–1989): Forsyth County, NC; Jackson, MS; Minneapolis suburbs, MN; and Washington County, MD. Follow-up clinic visits have occurred in 1990–1992 (Visit 2), 1993–1995 (Visit 3), 1996–1998 (Visit 4) and 2011–2013 (Visit 5). The present study excluded participants with self-reported race other than AA or White (n = 48), and AA participants from the Minneapolis and Washington County centers as the numbers are too small for adequate adjustment or within-community comparisons (n = 55). We further excluded participants whose outcome or relevant covariates were missing (n = 620), including one participant with missing SCD status. The final study population included 15,069 participants (Supplemental Figure 1). The ARIC Study has been approved by the institutional review boards at all centers, and all participants provided written informed consent.

SCD ascertainment

Participants were followed for SCD events from the baseline visit through December 31, 2012. SCD was defined as death from a sudden pulseless condition occurring out of hospital or in the emergency room from a cardiac cause in a previously stable individual without a non-cardiac cause of cardiac arrest.¹⁶ For witnessed cases, SCD was operationally defined as a sudden collapse (pulseless condition), without evidence of a non-cardiac cause of cardiac arrest. For unwitnessed cases, patients had to be known to be in a stable condition in the 24 hours prior to cardiac arrest without evidence of a non-cardiac cause of cardiac arrest.

The ARIC study conducted continuous surveillance for all cardiovascular related events and collected data from death certificates, annual phone calls, next-of-kin interviews, physician questionnaires, coroner information, and hospital discharges. To identify SCD cases, a committee of physicians first adjudicated all deaths attributable to coronary heart disease (CHD).¹⁷ Then, all fatal CHD cases (including definite fatal MI, definite fatal CHD, and possible fatal CHD) occurring out-of-hospital or in the emergency room were further adjudicated by two physicians and classified as definite sudden arrhythmic death, possible sudden arrhythmic death, not sudden arrhythmic death, or unclassifiable.¹⁸ Discordant classifications were resolved by a third physician. In this study, SCD was defined as an adjudicated definite or possible sudden arrhythmic death. Interviewer agreement was 83.2%.

Risk factor assessment

At the baseline exam, each participant was asked by a trained interviewer how they would identify their race/ethnicity, and then categorized into AA, White or other. Demographic factors, insurance status, education, and income were also collected at the baseline visit, as previously described.¹⁵ Prevalent CHD at baseline was defined as a self-reported history of CHD or electrocardiographic evidence of myocardial infarction. Prevalent CHD at each subsequent clinic visit was defined as the presence of prevalent CHD at baseline or the development of non-fatal incident CHD from the last clinic visit.

Cardiovascular risk factors and a standard 12-lead electrocardiogram were collected at baseline and at each follow-up visit, as previously described.^{15, 17} Sports physical activities were assessed via a modified Baecke questionnaire. Heart rhythm medication use was defined if participant reported taking beta-blockers (non-selective, cardio-selective, or combination), or medication for abnormal heart rhythm. Weight and height were measured and body mass index (BMI) was calculated as weight (kilograms) divided by height (meters) squared. Sitting blood pressure was calculated as the average of the second and third measurements. Diabetes was defined by self-report of a physician diagnosis, a fasting blood glucose level ≥ 126 mg/dl, a non-fasting blood glucose level ≥ 200 mg/dl, or use of insulin. Estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration equations using serum creatinine,¹⁹ and chronic kidney disease was defined as eGFR < 60 mL/min/1.73m². QRS duration, QT interval, Cornell voltage for left ventricular (LV) hypertrophy (QRS duration times the Cornell voltage) and heart rate were also obtained from standard 12-lead electrocardiograms.

Statistical analysis

We first obtained non-parametric estimates of the cumulative incidence function for SCD using age as time scale and taking into account competing risks due to other causes of death.²⁰ We calculated separate cumulative incidence functions for AA women, White women, AA men, and White men. We then calculated multivariable adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for SCD comparing AAs with Whites using competing risks survival analysis models with time-dependent covariates using age as time scale. We used four models with progressive degrees of adjustment. Model 1 adjusted for sex. Model 2 further adjusted for education, insurance, annual household income, smoking status, alcohol intake and physical activity. Model 3 further adjusted for body mass index, systolic blood pressure, anti-hypertensive medication use, diabetes, total cholesterol, HDL-cholesterol, use of lipid lowering medications, estimated GFR, and heart rhythm medication use. Model 4 further adjusted for heart rate, QT interval, QRS duration, LV hypertrophy, and prevalent CHD. Sex, education, insurance, and annual household income were only measured at baseline and used as time-fixed variables. Smoking status, alcohol intake, and all variables additionally adjusted for in Models 3 and 4 were measured at each clinic visit and used as time-varying variables. Physical activity was collected at baseline, visit 3 and visit 5 and it was also used as a time-varying variable.

We also estimated the mediation effect of each variable on the association between race and SCD as the percent change in sex-adjusted proportional hazards regression coefficients for race after adjusting for the variable of interest. 95% CIs were calculated using bootstrapping. The findings using Aalen additive hazard models were similar (not shown).

Finally, we evaluated the association of race with SCD in pre-specified groups defined at baseline by age, sex, smoking status, alcohol consumption, BMI categories, hypertension, diabetes, kidney disease, and prevalent CHD. The interactions between race and other participant characteristics on SCD risk were tested using likelihood ratio tests comparing nested models with and without interaction terms.

All reported P values were two-sided and the significance level was set at 0.05. All statistical analyses were performed using STATA version 15 (StataCorp LP, College Station, Texas).

Results

The number of AA women, White women, AA men and White men in our study was 2,366 (15.7%), 5,941 (39.4%), 1,466 (9.7%), and 5,296 (35.1%), respectively (Table 1). Compared with Whites, AAs were younger, more likely to be uninsured, current smokers, to have a higher prevalence of hypertension, diabetes, BMI, and LV hypertrophy, to have lower levels of education, alcohol consumption, triglycerides, and eGFR.

Table 1.

Characteristics of study participants at baseline by race and sex: the ARIC cohort (1987–1989).

	Women			Men
	AAs	Whites	p-value	AAs	Whites	p-value
N	2,366	5,941		1,466	5,296
Age (years)	53.4 (5.7)	54.0 (5.7)	<0.001	53.9 (6.0)	54.8 (5.7)	<0.001
Education			<0.001			<0.001
<High school	945 (39.9)	966 (16.3)		644 (43.9)	951 (18.0)
High school / vocational school	697 (29.5)	3,028 (51.0)		383 (26.1)	2,076 (39.2)
College, graduate, or professional school	724 (30.6)	1,947 (32.8)		439 (29.9)	2,269 (42.8)
Have insurance	1,814 (76.7)	5,646 (95.0)	<0.001	1,136 (77.5)	5,057 (95.5)	<0.001
Income (US$)			<0.001			<0.001
<16,000	1,256 (53.1)	889 (15.0)		567 (38.7)	443 (8.4)
16,000–<25,000	371 (15.7)	897 (15.1)		253 (17.3)	629 (11.9)
25,000–<35,000	236 (10.0)	1,117 (18.8)		192 (13.1)	1,008 (19.0)
35,000–<50,000	167 (7.1)	1,202 (20.2)		166 (11.3)	1,243 (23.5)
≥50,000	110 (4.6)	1,540 (25.9)		136 (9.3)	1,786 (33.7)
Total Cholesterol (mg/dl)	217.3 (45.5)	218.4 (42.3)	0.29	210.9 (44.0)	211.3 (38.7)	0.73
HDL (mg/dl)	40.3 (11.5)	40.5 (10.6)	0.44	36.7 (11.5)	32.5 (9.7)	<0.001
LDL (mg/dl)	137.7 (43.3)	135.5 (39.6)	0.02	137.4 (42.0)	140.2 (35.8)	0.01
Triglycerides (mg/dl)	109.8 (70.6)	128.8 (85.1)	<0.001	120.0 (94.8)	148.0 (98.0)	<0.001
Body mass index (kg/m2)	30.8 (6.5)	26.6 (5.5)	<0.001	27.6 (4.9)	27.5 (4.0)	0.22
Physical activity index (score)	2.1 (0.7)	2.4 (0.8)	<0.001	2.2 (0.7)	2.7 (0.8)	<0.001
Smoking Status			<0.001			<0.001
Never	1,347 (56.9)	3,008 (50.6)		412 (28.1)	1,476 (27.9)
Former	430 (18.2)	1,450 (24.4)		496 (33.8)	2,522 (47.6)
Current	589 (24.9)	1,483 (25.0)		558 (38.1)	1,298 (24.5)
Current alcohol consumption	490 (20.7)	3,611 (60.8)	<0.001	726 (49.5)	3,660 (69.1)	<0.001
Diabetes	478 (20.2)	478 (8.0)		263 (17.9)	537 (10.1)
Systolic BP (mm Hg)	127.8 (21.2)	117.0 (17.7)	<0.001	130.1 (21.3)	120.2 (16.2)	<0.001
Hypertension	1,326 (56.0)	1,550 (26.1)		790 (53.9)	1,507 (28.5)
Estimated GFR (ml/min/1.73 m2)			<0.001			<0.001
≥90	2,083 (88.0)	5,108 (86.0)		1,211 (82.6)	4,176 (78.9)
60–<90	228 (9.6)	783 (13.2)		223 (15.2)	1,060 (20.0)
<60	55 (2.3)	50 (0.8)		32 (2.2)	60 (1.1)
Prevalent CHD	69 (2.9)	108 (1.8)	0.002	85 (5.8)	468 (8.9)	<0.001
LV hypertrophy	136 (5.8)	66 (1.1)	<0.001	100 (6.9)	63 (1.2)	<0.001
Heart rhythm medication	322 (13.6)	654 (11.0)	<0.001	182 (12.4)	747 (14.1)	0.10
Heart rate (beat/min)	68.0 (11.0)	68.0 (9.6)	1.00	65.2 (10.9)	65.1 (10.1)	0.80
QT interval (sec)	412.1 (31.2)	410.6 (27.0)	0.03	405.1 (29.4)	407.3 (27.8)	0.01
QRS duration (sec)	88.7 (11.3)	88.3 (10.5)	0.13	96.1 (14.2)	97.3 (12.7)	0.002

Data in the Table are mean (SD) or number (percent).

AA: African American; BP: blood pressure; CHD: coronary heart disease; GFR: glomerular filtration rate; HDL: high-density lipoproteins; LDL: low-density lipoproteins; LV: left ventricular.

During a median follow-up of 27.4 (27.1 for AAs, 27.5 for Whites, 25^th and 75^th percentile 20.0, 28.5) years, 547 participants experienced SCD (215 AAs, 332 Whites). The cumulative incidence was highest in AA men, followed by White men, AA women and White women (Figure 1). In AA men, the cumulative incidence increased from 1.1% (95% CI 0.2, 2.0) at age 55 to 9.6% (7.7, 11.5) at age 85. The corresponding numbers for AA women, White men, and White women was 0.7 (0.2, 1.2), 0.9 (0.4, 1.4), and 0.2 (0.0, 0.3) at age 55, and 6.6 (5.3, 7.9), 6.5 (5.6, 7.4), and 2.3 (1.8, 2.9) at age 85, respectively (Table 2).

Figure 1.

Cumulative incidence of SCD by race and sex groups.

Curves were estimated using competing risk analysis.

Table 2.

Cumulative incidence (%) and 95% confidence intervals of sudden cardiac death by race and sex groups.

Age	Women		Men		Overall
	AAs	Whites	AAs	Whites
55	0.7 (0.2, 1.2)	0.2 (0.0, 0.3)	1.1 (0.2, 2.0)	0.9 (0.4, 1.4)	0.6 (0.4, 0.8)
60	1.0 (0.4, 1.6)	0.3 (0.1, 0.4)	2.3 (1.2, 3.5)	1.8 (1.2, 2.3)	1.1 (0.8, 1.3)
65	2.0 (1.3, 2.7)	0.5 (0.3, 0.8)	3.3 (2.1, 4.5)	2.6 (1.9, 3.2)	1.7 (1.5, 2.0)
70	3.0 (2.2, 3.8)	0.8 (0.6, 1.1)	5.2 (3.9, 6.6)	3.6 (2.9, 4.3)	2.6 (2.3, 2.9)
75	4.2 (3.2, 5.1)	1.1 (0.8, 1.4)	7.0 (5.5, 8.6)	4.7 (3.9, 5.4)	3.4 (3.1, 3.8)
80	5.9 (4.7, 7.0)	1.8 (1.4, 2.2)	8.7 (7.0, 10.5)	5.5 (4.7, 6.3)	4.4 (4.0, 4.8)
85	6.6 (5.3, 7.9)	2.3 (1.8, 2.9)	9.6 (7.7, 11.5)	6.5 (5.6, 7.4)	5.2 (4.7, 5.7)

Data in the Table are percentages obtained from cumulative incidence functions taking into account competing risks by other causes of death.

AA: African American.

The sex-adjusted hazard ratio for SCD comparing AAs with Whites was 2.12 (95% CI 1.79, 2.51; Table 3). Adjustment for multiple socioeconomic factors, cardiovascular risk factors, and electrocardiographic variables attenuated the association, but the association was still statistically significant in fully adjusted models (HR 1.38, 95% CI 1.11, 1.71). Additional analyses adjusting only for baseline values of the covariates and taking competing risks into account showed similar results (not shown).

Table 3.

Hazard ratios (95% CI) for sudden cardiac death comparing African Americans with Whites.^*

	Overall		Women		Men
	AAs	Whites	AAs	Whites	AAs	Whites
Events/p-y	215 / 76,344	332 / 240,514	111 / 48,983	89 / 131,436	104 / 27,361	243 / 109,078
IR per 1,000 p-y	2.8 (2.5, 3.2)	1.4 (1.2, 1.5)	2.3 (1.9, 2.7)	0.7 (0.6, 0.8)	3.8 (3.1, 4.6)	2.2 (2.0, 2.5)
Model 1†	2.12 (1.79, 2.51)	Ref (1)	3.25 (2.46, 4.29)	Ref (1)	1.60 (1.27, 2.02)	Ref (1)
Model 2‡	1.29 (1.06, 1.57)	Ref (1)	2.06 (1.53, 2.77)	Ref (1)	0.95 (0.74, 1.23)	Ref (1)
Model 3§	1.34 (1.08, 1.65)	Ref (1)	1.91 (1.40, 2.61)	Ref (1)	1.06 (0.81, 1.38)	Ref (1)
Model 4‖	1.38 (1.11, 1.71)	Ref (1)	1.91 (1.39, 2.62)	Ref (1)	1.10 (0.84, 1.45)	Ref (1)

^*Results derived from competing risk analysis using time-varying covariates.

^†Model 1: Adjusted for sex (women or men).

^‡Model 2: Additionally adjusted for education (<high School, high School, or >high school), insurance (no or yes), annual household income (<16000, 16000–< 25000, 25000–<35000, 35000–<50000, or >50000), smoking status (never, former, or current), alcohol intake (non-current or current) and physical activity (continuous).

^§Model 3: Additionally adjusted for body mass index (continuous), systolic blood pressure (continuous), anti-hypertensive medication use (no or yes), diabetes, total cholesterol (continuous), HDL-cholesterol (continuous), use of lipid lowering medications (no or yes), estimated GFR (continuous), and heart rhythm medication use (no or yes).

^‖Model 4: Additionally adjusted for heart rate (continuous), QT interval (continuous), QRS duration (continuous), LV hypertrophy (no or yes), and prevalent CHD (no or yes).

AA: African American; p-y: person-years; IR: incidence rate.

In mediation analysis, socioeconomic factors, cardiovascular risk factors, and electrocardiographic variables explained 65.3% (37.9, 92.8%) of the excess risk of SCD in AAs compared to Whites (Table 4). The single most important factor explaining this difference was income (50.5%), followed by education (19.1%), hypertension (22.1%), diabetes (19.6%), LV hypertrophy (15.0%), alcohol intake (12.9%), physical activity (12.7%), BMI (6.4%), and smoking (4.4%). For traditional CVD risk factors, the associations with SCD were stronger in Whites compared with AAs for BMI and prevalent CHD, and weaker for hypertension (p-value for interaction 0.01, 0.03, and 0.03; Supplemental Table 1).

Table 4.

Mediation analysis of the association between race and sudden cardiac death.

	% Change (95% CI) in the HR coefficient for race
	Overall*	Women	Men	p-value†
Smoking Status	4.4 (0.8, 8.0)	−0.4 (−3.5, 2.6)	9.2 (−11.5, 30.0)	0.002
Current alcohol consumption	12.9 (3.2, 22.5)	7.8 (−5.5, 21.1)	18.1 (0.3, 35.9)	0.20
Physical activity	12.7 (6.3, 19.1)	6.6 (1.9, 11.2)	23.2 (3.7, 42.7)	0.02
Education	19.1 (7.4, 30.8)	18.3 (5.5, 31.2)	19.9 (−3.0, 42.9)	0.91
Income	50.5 (31.5, 69.4)	42.8 (23.7, 62.0)	65.2 (21.1, 109.2)	0.35
BMI	6.4 (1.3, 11.5)	7.7 (−0.8, 16.2)	0.5 (−3.4, 4.4)	0.08
Diabetes	19.6 (12.8, 26.3)	19.1 (7.7, 30.5)	22.1 (−0.2, 44.3)	0.77
Hypertension	22.1 (14.9, 29.3)	19.9 (10.4, 29.4)	28.3 (7.6, 49.0)	0.69
Heart rhythm medication	−5.4 (−10.1, −0.8)	−0.1 (−2.1, 1.9)	−14.7 (−33.3, 3.9)	0.001
eGFR	−12.6 (−18.9, −6.3)	−6.3 (−12.6, 0.0)	−22.1 (−31.7, −12.4)	0.06
QRS duration	−2.6 (−5.4, 0.2)	2.9 (0.0, 5.7)	−8.9 (−15.3, −2.5)	0.001
Left ventricular hypertrophy	15.0 (8.6, 21.3)	7.5 (−0.3, 15.2)	28.7 (3.9, 53.6)	0.02
Prevalent CHD	−6.8 (−12.0, −1.6)	1.0 (−2.5, 4.4)	−17.8 (−47.6, 12.0)	0.02
Joint effect of all the above covariates	65.3 (37.9, 92.8)	65.5 (40.1, 90.9)	66.4 (8.8, 124.1)	0.99

^*Adjusted for sex.

^†P-value comparing % change in men vs. women.

In stratified analysis by pre-specified subgroups, associations between race and SCD were present across all groups. Interestingly, the association was stronger in women compared with men, and in non-obese compared with obese participants (p-values for interaction of 0.004, and 0.06, respectively; Figure 2). The interaction of race and age was not statistically significant (p-values of 0.48, 0.28, and 0.40, overall, in men, and in women, respectively).

Figure 2.

Hazard ratios (95% CI) for sudden cardiac death comparing African Americans with Whites in pre-specified subgroups.

Models were adjusted for sex, education, insurance, income, smoking status, alcohol intake, physical activity, body mass index, systolic blood pressure, anti-hypertensive medication use, diabetes, LDL-, HDL-cholesterol, triglycerides, use of lipid lowering medications, estimated GFR, heart rhythm medication, heart rate, QT interval, QRS duration, left ventricular hypertrophy, and prevalent CHD at baseline.

Discussion

In this large community-based cohort, the lifetime cumulative incidence of SCD at age 85 was 9.6, 6.6, 6.5, and 2.3% for AA men, AA women, white men, and white women, respectively. AAs had a higher risk for SCD compared to Whites. Socioeconomic status, and traditional risk factors explained approximately 65% of the race difference in SCD. Even after including multiple time-varying risk factors in the models, AAs showed an increased risk of SCD compared to Whites that was not explained by available factors. The increased risk associated with race was more pronounced in women than in men, although the proportion of the risk difference associated with race explained by risk factors was similar in men and women. The high burden of SCD and the racial disparities observed in our study represent a major public health problem with important clinical implications.

Our findings of racial disparity with SCD, using physician-adjudicated outcome data, is consistent with prior studies using different SCD surveillance methodologies. AAs were at increased risk of out-of-hospital sudden cardiac arrest in studies based on emergency medical service systems in Seattle,⁷ Chicago,⁵ New York City,⁶ Illinois,⁸ and Portland.⁴ AA patients were also less likely to survive in-hospital sudden cardiac arrests based on cardiac and resuscitation registry data, although the differences in survival have decreased during the past 5 years.^{9, 21} These studies were focused on describing incidence and survival of sudden cardiac arrest, but could not evaluate the reasons for the race differences.

One cohort of patients with pre-existing comorbidities have also identified a higher risk of SCD in AAs compared to Whites. The LIFE study followed 533 AA and 8,660 non-AA patients with hypertension over 4.8 years of follow-up (17 SCD events in AAs and 161 in non-AAs), and found that AA race was associated with almost double the risk of SCD compared to non-AAs even after multivariate adjustment (HR 1.98, 95% CI 1.12, 3.59).¹² The study was limited by the use of selected study populations, the small number of AAs, and the short follow-up time. Our study findings can be directly applied to community settings, where the majority of SCD cases occur, and indicate that even taking into account multiple measures of numerous cardiovascular risk factors, AAs are still at increased risk of SCD compared to Whites.

The mechanisms underlying the race differences in our study are unclear. In sex-adjusted analysis, AAs had over 2 times the risk of SCD compared to Whites. With the exception of hypertension, the strength of the association of individual risk factors with SCD risk was either similar in AA as in Whites or stronger in Whites, implying that the main contribution of known risk factors on racial differences in SCD was through a higher prevalence of the risk factors in AAs, and not through stronger effects. Our analysis thus implies that intervention and control of known risk factors should reduce the gap in SCD between AAs and Whites. Control of hypertension in AAs, in particular, should have a major role in SCD control as hypertension is both highly prevalent among AAs and has a stronger association with SCD in AAs compared to Whites. While improvement in the awareness, treatment and control of hypertension has been made in the US during the past 30 years, additional effort could be targeted in more effective treatment for controlled hypertension among AAs, which has lower proportion than Whites.²²

In our analysis, the income and education difference between AAs and Whites were the main factors explaining racial differences in SCD. Indeed, the incidence of cardiac arrest was 30–80% higher in US areas of lowest vs. highest socioeconomic status.²³ The associations of income and education with SCD were similarly strong for AAs and Whites, but AAs were 4 times more likely to be in the lowest income category. Income and education are upstream factors with wide implications on cardiovascular risk, including associations with multiple cardiovascular risk factors.²⁴ Low income and education are also associated with unhealthy behaviors, low disease awareness, and limited access to care, which could all contribute to poor SCD outcomes.^25–27

In a recent analysis that combined ARIC, the Cardiovascular Health Study, and the Reasons for Geographic and Racial Differences in Stroke Study (REGARDS), AAs had a higher age-adjusted risk for fatal CHD compared with Whites, but after accounting for social, lifestyle, and health determinants, the risk difference became insignificant.²⁸ In our analysis, restricted to SCD, adjustment for risk factors does not completely explain the racial differences. Although around 80% of SCD is caused by coronary artery abnormalities,² some of the excess risk in AAs might be explained by hypertrophic cardiomyopathy and other forms of non-ischemic cardiomyopathy (i.e. hypertensive heart disease) that may be more prevalent in AAs.^{29, 30} In our analysis, the race differences in SCD persisted even after including the development of coronary heart disease as a time-varying covariate, further pointing to a non-ischemic etiology for these differences.

In ARIC, we did not have information on the success of resuscitation efforts, and we could not evaluate if the race differences were due to differences in the rate of life-threatening arrhythmias or to differences in resuscitation rates. Compared to Whites, the initial cardiac rhythm in sudden cardiac arrest in AA is more likely to be asystole or electromechanical dissociation, which is associated with a lower probability of survival than ventricular fibrillation.^{6, 7, 21} Disparities in emergency management and resuscitation procedures between AAs and Whites may also contribute to SCD mortality differences.^{5, 8, 9, 31, 32} For out-of-hospital cardiac arrests occurring in Chicago, AAs were more likely to have an unwitnessed cardiac arrest and longer response time of the EMS ambulance compared with Whites.⁵ Neighborhood characteristics may further contribute to mortality differences. In multi-center studies conducted in the US, out-of-hospital cardiac arrests occurring in low-income AA neighborhoods were less likely to receive bystander cardiopulmonary resuscitation and less likely to survive to hospital discharge than those occurring in high-income White neighborhoods.^{31, 32} Survival of in-hospital cardiac arrest was also lower in low-income neighborhoods, driven by poorer quality of treatment delivered by medical institutions in those areas.^{9, 33} Socioeconomic factors and neighborhood infrastructure likely play a major role in racial discrepancies in cardiac arrest survival. Interventions aiming at improving awareness, cardiopulmonary resuscitation outreach, response to cardiac arrest events, medical service access, and quality of medical institutions in predominantly AA and disadvantaged neighborhoods are critical to reduce this gap.

The race differences in SCD rates comparing AAs and Whites were particularly striking in women. In our study, the prevalence of CHD was higher in AAs among women, but lower among men. The sex difference, however, was still evident after adjusting for cardiovascular risk factors and for CHD. In AA women, non-ischemic heart disease, including dilated cardiomyopathy and valvular heart disease, may be a more common precursor of SCD.^34–36 The cumulative incidence of SCD in AA women was similar to that in White men, indicating that AA women are a particularly vulnerable subgroup for SCD that should be targeted in current preventive strategies. Additional research is needed for developing effective means of prevention of SCD in AA women.

In addition to socio-economic and traditional cardiovascular risk factors, other mechanisms could explain racial differences in SCD. For instance, AAs have a higher prevalence of mutations in the cardiac sodium channel variants SCN5A, which could mediate an increased risk of ventricular arrhythmias.^37–40 Our understanding of the mechanisms for racial differences in SCD is still incomplete, and additional research is research into these mechanisms is granted.

Our study has several limitations. We did not have information on the structural substrate underlying cardiac arrest or on the causative arrhythmia at the time of SCD. Also, we did not have data on access to care or resuscitation procedures, and we could not determine if these factors contributed the racial disparities. Furthermore, our case-definition for SCD excluded resuscitated cardiac arrests that survived to hospital discharge. While we could not determine the number of resuscitated cases of cardiac arrest, this number is likely small as sudden cardiac arrest is nearly always fatal without rapid intervention by emergency medical response providers. Additionally, race was self-reported. This reporting is subject to bias and that may further contribute to observed differences in outcomes. Finally, race is a complex construct that involves genetic, socioeconomic, environmental, cultural, and lifestyle factors.⁴¹ These factors, including education and income, are distributed very differently by race. While our study collected and rigorously adjusted for numerous variables in the analysis, it is unlikely that we captured all factors that differ by race, and we could not tease out all their effects on the association between race and SCD.

The strengths in our study included the community-based setting, the large sample size and the long duration of follow-up, the extensive quality control and quality assurance procedures in ARIC, and the rigorous adjudication process for SCD events using multiple sources of information. This allows for capturing 30–40% more SCD cases that would otherwise be missed in studies using emergency medical service systems as the single source of events.⁴ We were also able to adjust for time-varying risk factors repeatedly collected throughout follow-up, as well as to adjust for electrocardiographic measurements that predict SCD.

This large prospective study of participants in community demonstrated that AAs had a significantly higher cumulative lifetime risk of SCD compared with whites. The higher burden of SCD in AA was more evident in women, people who were non-obese and without prevalent CHD, and was in part attributed to, but independent of multiple risk factors, including income, hypertension, and other cardiovascular risk factors. Low socioeconomic status is the major contributor of excess SCD risk in AA. Further studies are needed to elucidate the underlying cause of increased SCD risk in AAs, and determine whether targeted clinical practice that takes race into account can provide a better approach for risk-stratified prevention and therapy for SCD.

What is new?

• African Americans showed an increased risk of sudden cardiac death (SCD) compared with Whites that was not fully explained by available factors.

• The increased risk associated with race was more pronounced in women than in men.

• The cumulative incidence (95% confidence intervals) of SCD for African American men, African American women, White men, and White women at age 85 was 9.6% (7.7, 11.5), 6.6 (5.3, 7.9), 6.5 (5.6, 7.4), and 2.3 (1.8, 2.9), respectively.

• The income and education difference between African Americans and Whites were the main factors explaining racial differences in SCD.

What are the clinical implications?

• The high burden of SCD in the general population, particularly in African Americans, requires better approaches to improve preventive measures.

• Efforts to reduce the SCD risk in African Americans should focus on improving cardiopulmonary resuscitation outreach, medical care engagement in response to cardiac arrest events, and quality of treatment in medical institutions in predominantly African neighborhoods.