Cardiovascular Disease Mortality by County, Race, and/or Ethnicity, United States, 2000–2019

Abstract

Background:

Cardiovascular disease (CVD) remains the leading cause of death and an important contributor to US health disparities.

Objectives:

To assess county-level CVD mortality and trends by race and/or ethnicity, and by detailed CVD cause of death.

Methods:

We estimated age-standardized mortality rates (ASMR) for 5 racial and/or ethnic groups (American Indian or Alaska Native [AIAN], Asian or Pacific Islander [Asian], Black or African American [Black], Latino or Hispanic [Latino], and White), 3110 counties, 21 causes of CVD death, and 20 years (2000–2019) using small-area estimation methods and death certificate data from the US National Vital Statistics System. Estimated mortality rates were adjusted for misreporting of race and/or ethnicity on death certificates and age-standardized to the 2010 census. Counties were classified into persistently high-mortality, persistently low-mortality, and intermediate strata based on age-adjusted CVD mortality rates, stratified by race and ethnicity.

Results:

ASMR for total CVD was 237.8 (95% uncertainty interval: 237.2–238.4) deaths per 100,000 in 2019. Mortality was highest for the Black population for total CVD and 12 of 17 detailed CVD causes, and lowest for the Asian population for total CVD and 13 of 17 detailed causes. From 2000 to 2019, mortality declined for total CVD and 12 of 17 detailed causes; however, mortality increased for three detailed causes. Mortality also varied widely among counties—much more than among states—both within and among racial and/or ethnic populations. Spatial mortality patterns varied considerably by CVD cause and by race and/or ethnicity.

Conclusions:

While declining CVD mortality reflects meaningful clinical and public health progress, persistent and pervasive disparities demand solutions-oriented, place-informed research to drive further progress. Critically, these pre-pandemic data establish a baseline essential for isolating pandemic-attributable shifts from pre-existing trends. These granular, county-level data stratified by CVD cause, race and/or ethnicity offer a scalable, repeatable framework for agencies to tailor health interventions and rigorously track disparities over time..

Condensed Abstract

This study examined cardiovascular disease (CVD) mortality across US counties by race and/or ethnicity from 2000 to 2019, employing small-area estimation and National Vital Statistics System mortality data. We calculated age-standardized mortality rates for five racial and/or ethnic groups across 3,110 counties and 21 CVD causes. Findings showed a general decline in CVD mortality, with the highest rates in Black populations and lowest in Asians. Mortality rates varied widely among counties, more so than among states. Recent rate increases in certain CVD causes were seen. Despite overall progress, these findings call for focused interventions addressing racial, ethnic, and geographic disparities.

Introduction

Cardiovascular disease (CVD) remains the leading cause of death worldwide¹ and the major contributor to overall mortality and health disparities in the United States (US).² National public health campaigns focused on prevention and treatment of CVD have reduced death rates, but the decline in mortality has slowed since 2011.^3,4 Demographic changes in the US, including an aging population, are projected to expand the CVD health impact, disproportionally affecting Black, Latino, and AIAN individuals.^5–7

Nationally, and within states, CVD mortality varies markedly among racial and/or ethnic groups. Black adults are much more likely to die of CVD than their White counterparts, with much of the mortality risk difference explained by socioeconomic status and CVD risk factors such as hypertension and smoking.^2,8 Similarly, CVD mortality is higher for American Indian populations in comparison to White populations in the US.^7,9 However, the extent to which these disparities differ by location, and how their patterns vary by specific cause of CVD death, remain incompletely understood. Where consistent and comparable measures of mortality by detailed CVD cause are available, they rarely go beyond comparisons of the three largest populations—Black, Latino, and White.^2,10,11 Detailed knowledge of county-level variation is important for all racial and/or ethnic populations and can inform appropriate local-level clinical and public health practices, policies, and strategies needed to have a sustained effect on reducing or eliminating racial, ethnic, and place-based CVD disparities.^12,13

In this study, we examine patterns and trends in cardiovascular disease (CVD) mortality across 21 causes of death — total CVD, 17 detailed causes, and 3 sub-totals — at the county level and by race and/or ethnicity over a 20-year period (2000–2019)—just before the onset of the COVID-19 pandemic in the U.S. We further characterize county- and race and/or ethnicity-level disparities in CVD mortality and describe how these disparities vary across detailed causes of death.

This pre-pandemic baseline is especially important given that the COVID-19 pandemic substantially disrupted cardiovascular care and accelerated CVD mortality across much of the United States. Emerging evidence shows that CVD mortality increased in numerous states between 2019 and 2023,^14–17 yet interpreting the magnitude and disparities implications of these pandemic-era increases requires a rigorous reference point — one that is cause-specific, geographically granular, and stratified by race and/or ethnicity. The present study provides exactly that foundation, establishing county-by-cause-by-population estimates of CVD mortality trends from 2000 to 2019 against which pandemic and post-pandemic trajectories can be measured and evaluated.

Methods

For this analysis, we leveraged methods previously developed for estimating cause-specific mortality by county and race and/or ethnicity for a comprehensive set of causes of death from the Global Burden of Diseases cause list.¹⁸ A detailed description of the analytic approach is available elsewhere (see the appendix for full details).¹⁸ Here, we summarize these methods as they were applied to estimate CVD mortality. This study complies with the Guidelines for Accurate and Transparent Health Estimates Reporting¹⁹ (Supplemental Appendix 1) and received institutional review board approval from the University of Washington. No primary data were collected for this study, and we had no contact with human subjects.

Unit of Analysis

We estimated CVD mortality by county, combined race and/or Latino ethnicity, CVD cause of death, and year (2000–2019). We merged counties as needed to create geographic units of analysis that remained the same over time, which reduced the number of areas considered from 3143 to 3110.²⁰ We estimated for five mutually exclusive racial and/or ethnic populations: American Indian or Alaska Native (AIAN), Asian or Pacific Islander (Asian), Black (Black), Latino or Hispanic (Latino), and White (White).²¹ We were unable to consider separate Asian and Native Hawaiian or Pacific Islander (NHPI) populations in this analysis, as individuals with less common Asian or NHPI ethnicities were combined in an “Other Asian or Pacific Islander” category in the deaths data prior to 2011, preventing disaggregation. The estimates for the combined Asian and NHPI population primarily represent the experience of the much larger (>30 times)²² Asian population, so we refer to this combined population as “Asian”.

This study focuses on 21 causes of death including total CVDs, 17 detailed causes, and 3 sub-totals. The 17 detailed causes are: rheumatic heart disease, ischemic heart disease, ischemic stroke, intracerebral hemorrhage, subarachnoid hemorrhage, hypertensive heart disease, non-rheumatic calcific aortic valve disease, non-rheumatic degenerative mitral valve disease, myocarditis, alcoholic cardiomyopathy, other cardiomyopathy, pulmonary arterial hypertension, atrial fibrillation and flutter, aortic aneurysm, lower extremity peripheral arterial disease, endocarditis, and the remaining cardiovascular and circulatory diseases (as other). The three subtotals are stroke, non-rheumatic valvular heart disease, and cardiomyopathy and myocarditis. We did not produce separate estimates for other non-rheumatic valve diseases, as the total number of deaths over the study period was small; however, deaths due to this cause are included in the total non-rheumatic valvular disease estimates and in the total CVDs estimates. The ICD-10 codes corresponding to each cause are listed in Supplemental Appendix 1. Heart failure is not included as a primary cause of death in GBD because it is generally the result of other CVDs, and the GBD study assigns mortality to those diseases to ensure accurate attribution of death and avoid duplication.

Data

We tabulated death records from the US National Vital Statistics System by county, racial and/or ethnic population, age group (0, 1–4, 5-year age bands from 5–9 to 80–84, and 85+), sex, year, and cause of death. We used the supplied “bridged” race for individuals with multiple racial identities.²³ We applied algorithms from the GBD to proportionately reallocate deaths initially assigned an intermediate, insufficiently specific, or otherwise implausible cause of death.^24,25 We merged the tabulated death data with population estimates from the National Center for Health Statistics as previously described.

Statistical Analysis

We fitted separate small-area estimation models for each cause and by sex to estimate mortality by county, racial and/or ethnic population, age, and year (purple boxes in the Statistical Methods Diagram, Supplemental Appendix 1). These small-area models were generalized linear mixed effects models and were designed to “borrow strength” by smoothing across locations, racial and/or ethnic populations, age groups, and time, and by leveraging additional information in the form of covariates. As demonstrated by our previous validation analyses.¹⁸ this approach enabled us to generate stable estimates of cause-specific mortality rates despite the typically small number of deaths observed in any individual county, racial and/or ethnic population, and year. We then multiplied the estimated mortality rates from the small-area estimation model by the existing estimates of misclassification ratios²⁶ for each racial and/or ethnic population to adjust for racial and/or ethnic misclassification on death certificates (blue boxes in the Statistical Methods Diagram). Next, we performed a calibration procedure to ensure consistency among causes of death (e.g., that estimates for all detailed CVDs add up to the estimates for total CVD) and to guarantee that adjustment for misclassification did not change the overall mortality rate for all racial and/or ethnic populations combined (green boxes in the Statistical Methods Diagram). To generate state- and national-level estimates, we calculated the population-weighted mean of the county-level estimates; similarly, to generate estimates for males and females combined, we calculated the population-weighted mean of the sex-specific estimates (yellow boxes in the Statistical Methods Diagram). All estimates were age-standardized using the population distribution from the 2010 Census as the standard; we selected this standard as 2010 is approximately half-way through the study period. To estimate uncertainty, we generated 1,000 draws from the approximated posterior distribution of the mortality rates from the small area model and carried out all subsequent steps described above separately for each of these 1,000 draws. We then generated 95% uncertainty intervals (UIs) from the 2.5^th and 97.5^th percentile of these draws and point estimates from the mean of these draws. We described changes over time as “statistically significant” when the posterior probability that the change was greater than 0 was less than 2.5% or greater than 97.5%.

Data for all counties and racial and/or ethnic populations were included in the statistical models and contributed to the national and state total estimates, however, we suppressed (i.e., did not report) estimates for individual counties and racial and/or ethnic populations where the mean annual population size was less than 1,000 (orange boxes in the Statistical Methods Diagram). These masked counties include <0.1% of the total population, 17.8% of the AIAN population, 2.6% of the Asian population, 1.0% of the Black population, 1.2% of the Latino population, and <0.1% of the White population.

Ethics Approval

Institutional review board approval was obtained from the University of Washington. No primary data were collected for this study, and we had no contact with human participants.

Results

County-level and other geographic disparities

In 2019, total CVD mortality varied substantially and showed large differences among counties within the same state (Figure 1). There was also a high degree of variation among counties within each race and/or ethnic group.

Figure 1: State and county age-standardized mortality rates for total CVD, 2019.

Counties and racial and/or ethnic populations with a mean annual population <1000 are not shown.

For the total population, total CVD mortality was elevated throughout much of a large swath of the US, stretching from Texas in the west to Northern Florida through Southern Virginia in the east, and to Michigan and Ohio in the north (Figure 2). Within this broader region, particularly high mortality rates were observed in certain counties in Northern and Eastern Texas, Eastern Oklahoma, the lower Mississippi Valley, Central Appalachia, and parts of Alabama, Northern Florida, Georgia, South Carolina, and Virginia. Elevated mortality rates were also observed in some counties outside of this region. The spatial patterns of total CVD mortality for the Black and White populations were like that observed for the total US population, although the absolute overall level of mortality was noticeably higher for the Black population throughout most of the US. In contrast, overall CVD mortality in 2019 among the Latino population was 25% lower (178.3/100,000 vs. 240.9/100,000) than for the White population, but relatively high in selected counties in Texas and Hawaii. Other areas in the Southwest, Wyoming, Central Florida, and parts of the Great Lakes and New England states had CVD mortality in the Latino population well above the national average. The pattern for the AIAN population was more dispersed: counties with relatively high or relatively low mortality could be found in many parts of the US, independent of the proportion of AIAN population. Some of the highest mortality rates among all racial and/or ethnic groups were observed for the AIAN population, including in parts of Mississippi, New Jersey, Kansas, Oklahoma, Maine, and North Carolina. Mortality for the Asian population was lower than for the total population in nearly all counties, but within this population the mortality varied, such that counties with relatively high and relatively low mortality were located throughout the country.

Figure 2: County-level age-standardized total CVD mortality rates by racial and/or ethnic population, 2019.

Counties and racial and/or ethnic populations with a mean annual population <1000 are not shown.

AIAN = American Indian or Alaska Native

To move beyond describing geographic and racial and/or ethnic variation in CVD mortality, we added a diagnostic layer to county-level estimates by classifying counties into three performance strata based on age-adjusted CVD mortality rates (Figure 3). For the total population and for each racial and ethnic group, counties in the highest decile in 2019 and across all study years were designated persistently high-mortality, whereas those in the lowest decile in 2019 and across all study years were designated persistently low-mortality. All remaining counties were classified as intermediate. This framework operationalizes county-level estimates as a targeting tool, identifying counties where CVD mortality warrants priority attention, distinguishing them from counties with consistently favorable outcomes, and from the heterogeneous middle stratum. In the sections below, we provide additional detailed county-level findings based on analyses by race and/or ethnicity within each population.

Figure 3: Counties in the highest and lowest decile of age-standardized total CVD mortality by racial and/or ethnic population.

Counties and racial and/or ethnic populations with a mean annual population <1000 are not shown. Deciles are calculated separately for each racial and/or ethnic population and year based on the age-standardized mortality rate.

AIAN = American Indian or Alaska Native

American Indian/Alaska Native Population

Among the AIAN population, the largest number of counties in the highest-mortality decile were found in Oklahoma (16), North Carolina (6), Montana (4), and Kansas (4) in 2019; and in North Carolina (6) and Maine (3) when considering all years (Figure 3). Four counties had mortality rates that exceeded 600 per 100,000 in 2019: Neshoba County, Mississippi; Cumberland County, New Jersey; and Sedgwick and Shawnee Counties, Kansas (Supplemental Appendix 2). Counties in the lowest-mortality decile were most commonly found in Texas (11), Ohio (6), South Carolina (4), Indiana (4), Louisiana (4), and New York (4) in 2019 and in Texas (11), Ohio (6), Indiana (4), and Louisiana (4) across all years. The highest 2019 estimated rate per 100,000 was Neshoba County, Mississippi (861.3 [95% UI: 698.9–1056.4]) and the lowest was Madison County, Alabama (47.9 [34.7–65.0]).

Asian Population

Within the Asian population, the largest number of counties in the highest-mortality decile in 2019 were found mostly in California (10), Wisconsin (9), Texas (5), and Oklahoma (5); across all years, California (9) and Wisconsin (6) had the highest number (Figure 3). The lowest decile mortality counties in 2019 were most common in New York (9), Michigan (6), Indiana (4), and Florida (4); Florida also had four counties in the lowest decile in all years. The three highest estimated mortality rates per 100,000 in 2019 were in Washington County, Arkansas (328.8 [262.7–397.9]); Sheboygan County, Wisconsin (274.0 [220.4–332.1]); and Brown County, Wisconsin (257.9 [208.0–313.6]); while the lowest rates were seen in Houghton County, Michigan (73.5 [49.4–103.9]); Los Alamos County, New Mexico (58.5 [39.0–84.5]); and combined Aleutians East Borough and Aleutians West Census Area, Alaska (41.4 [22.3–74.1]; Supplemental Appendix 2).

Black Population

Among the Black population, the highest mortality decile counties in 2019 were most concentrated in southern states, especially Texas (25), Louisiana (23), Mississippi (18), and Tennessee (15; Figure 3). Martinsville City, Virginia had the highest 2019 estimated mortality rate (665.6 [552.7–797.2]), followed by Webster County, Mississippi (532.6 [454.5–615.3]; Supplemental Appendix 2). Across all years, the lowest mortality decile counties were spread across the Northeast, Midwest, and the West with a few in Hawaii (3), southwest Florida (2), Georgia (2), southern Texas (2), and Oklahoma (1). Chippewa County, Michigan (64.9 [41.5–97.5]) had the lowest estimated rate in 2019.

Latino Population

Within the Latino population, counties in the highest mortality decile were concentrated in Texas, both in 2019 (96) and across all years (73; Figure 3). Other states with a concentration of counties in the highest mortality rate decile in 2019 and across all years included New Mexico (11, 5), Michigan (9, 6), and Colorado (8, 5). States with large numbers of lowest-decile counties in 2019 included Minnesota (17), Mississippi (16), Alabama (12), Ohio (12), Wisconsin (9), Illinois (9), and Iowa (9). Six counties (four in Texas, two in Hawaii) had estimated mortality rates exceeding 350 per 100,000 in 2019 (Supplemental Appendix 2). Honolulu County, Hawaii had the highest mortality rate (466.0 [417.9–517.4). Counties in the lowest decile in all years, were located primarily in Mississippi (15), Alabama (10), North Carolina (8), Georgia (6), and Wisconsin (5) and all had estimated mortality rates lower than 94 per 100,000 in 2019.

White Population

The mortality distribution within the White population shows a distinct pattern of the highest-mortality counties concentrated in the South and Appalachia in 2019 and across all years while the lowest mortality counties were most heavily concentrated in the West, Upper Midwest, and Northeast in 2019 (Figure 3). In 2019, the three highest mortality counties were all in Virginia (Martinsville City, 643.1 [561.5–731.8]; Petersburg City, 541.3 [473.1–609.8]; Covington City, 483.2 [414.5–560.4]) while the three lowest-mortality counties were in North Slope Borough, Alaska (77.8 [42.9–131.9]); and Summit County (75.1 [65.0–86.3]) and Pitkin County (72.3 [63.4–82.9]), Colorado (Supplemental Appendix 2).

Total Population

Among the total US population, the mortality distribution mostly mirrors what is described above for the White and Black populations with the highest mortality counties concentrated in the South and Appalachia. The lowest mortality counties were similarly most heavily concentrated in the West, Upper Midwest, and Northeast (Figure 3). Four of the five lowest mortality counties in 2019 were in Colorado with estimated mortality rates lower than 107 per 100,000 while the fifth was in combined Aleutians East Borough and Aleutians West Census Area, Alaska with a 2019 mortality rate of 115.3 (91.9–142.0) per 100,000 (Supplemental Appendix 2).

Findings for Detailed CVD Causes

Large differences in mortality among counties were noted for every detailed CVD cause (Figure 4). For some detailed causes (e.g., ischemic heart disease, intracerebral hemorrhage, and subarachnoid hemorrhage), the spatial pattern was like that observed for total CVD, while others had notably different patterns. For example, for non-rheumatic valvular heart disease (total and both sub-types), mortality was elevated in counties throughout much of the northern US apart from North and South Dakota. Rheumatic heart disease and aortic aneurysm both had elevated mortality in most counties in the Northern and Central part of the US, while areas with relatively low mortality were primarily found in states along the southern border, parts of the East Coast, and some urban areas in the interior. Alcoholic cardiomyopathy had yet another pattern, with relatively high mortality rates concentrated primarily in states along the West Coast, Alaska, and east of the Mississippi river. As was the case for total CVD, large differences in mortality among counties were also observed within each racial and/or ethnic populations for every detailed CVD cause of death (Supplemental Appendix 3).

Figure 4: County-level mortality rate ratio by CVD cause of death, 2019.

The figure displays the ratio of the county-level mortality rate to the national mortality rate for the same CVD cause of death. Areas shown in orange have higher-than-national-average mortality, while those shown in purple have lower-than-national-average mortality.

Counties and racial and/or ethnic populations with a mean annual population <1000 are not shown.

National-level disparities and trends by cause of CVD death

In 2019, the age-standardized mortality rate (ASMR) for total CVD was 237.8 (95% UI 237.2–238.4) deaths per 100,000 (Table 1). Among 17 detailed causes, mortality was highest for ischemic heart disease (126.9 [126.4–127.5]), ischemic stroke (30.6 [30.4–30.8]), hypertensive heart disease (20.1 [19.9–20.2]), intracerebral hemorrhage (15.0 [14.9–15.2]), and atrial fibrillation and flutter (9.3 [9.2–9.5]). The top five causes were in the same order for Latino and White populations; patterns were similar for AIAN, Asian, and Black populations with key exceptions: in the AIAN population, hypertensive heart disease ranked second and ischemic stroke third; in the Asian population, intracerebral hemorrhage ranked third and hypertensive heart disease fourth; and in the Black population, other cardiomyopathy ranked fifth while atrial fibrillation and flutter ranked eighth.

Table 1:

National age-standardized mortality rates by racial and/or ethnic group in 2019

Cause of Death	Total			AIAN			Asian			Black			Latino			White
	Rate	Cause rank	Rate	Cause rank	Race rank	Rate	Cause rank	Race rank	Rate	Cause rank	Race rank	Rate	Cause rank	Race rank	Rate	Cause rank	Race rank
Cardiovascular diseases (Total)	237.8 (237.2–238.4)	-	256.7 (228.9–288.2)	-	2	140.3 (136.2–144.2)	-	5	304.8 (302.7–306.9)	-	1	178.3 (174.5–182.0)	-	4	240.9 (240.2–241.7)	-	3
Rheumatic heart disease	1.4 (1.4–1.5)	14	1.6 (1.3–2.0)	14	1	1.0 (0.9–1.1)	13	4	1.4 (1.3–1.5)	14	3	0.9 (0.8–1.0)	14	5	1.5 (1.4–1.5)	13	2
Ischemic heart disease	126.9 (126.4–127.5)	1	139.4 (123.7–157.1)	1	2	71.3 (68.9–73.7)	1	5	152.7 (151.2–154.2)	1	1	95.8 (93.6–98.0)	1	4	130.4 (129.8–131.1)	1	3
Stroke (Total)	50.5 (50.3–50.7)	-	49.6 (43.8–55.7)	-	2	38.9 (37.6–40.1)	-	5	70.9 (70.1–71.7)	-	1	42.9 (41.9–43.9)	-	4	49.1 (48.8–49.4)	-	3
Ischemic stroke	30.6 (30.4–30.8)	2	26.0 (22.8–29.3)	3	4	21.5 (20.7–22.3)	2	5	39.6 (39.1–40.2)	2	1	26.1 (25.4–26.8)	2	3	30.5 (30.3–30.7)	2	2
Intracerebral hemorrhage	15.0 (14.9–15.2)	4	17.3 (15.2–19.4)	4	2	13.2 (12.7–13.7)	3	4	23.6 (23.3–24.0)	4	1	12.5 (12.2–12.9)	4	5	14.1 (14.0–14.2)	4	3
Subarachnoid hemorrhage	4.9 (4.8–4.9)	9	6.3 (5.6–7.1)	7	2	4.2 (4.0–4.4)	6	5	7.6 (7.4–7.8)	7	1	4.3 (4.1–4.4)	6	4	4.5 (4.5–4.6)	9	3
Hypertensive heart disease	20.1 (19.9–20.2)	3	27.1 (23.6–31.0)	2	2	10.0 (9.6–10.5)	4	5	34.6 (34.0–35.1)	3	1	14.6 (14.1–15.0)	3	4	19.3 (19.1–19.5)	3	3
Non-rheumatic valvular heart disease (Total)	7.5 (7.4–7.6)	-	6.0 (5.1–7.1)	-	2	3.5 (3.3–3.7)	-	5	5.6 (5.4–5.8)	-	3	4.1 (4.0–4.3)	-	4	8.3 (8.2–8.4)	-	1
Non-rheumatic calcific aortic valve disease	6.4 (6.3–6.4)	6	5.1 (4.3–5.9)	9	2	2.8 (2.7–3.0)	8	5	4.5 (4.4–4.7)	9	3	3.5 (3.4–3.7)	8	4	7.1 (7.0–7.1)	6	1
Non-rheumatic degenerative mitral valve disease	1.0 (1.0–1.1)	15	0.9 (0.7–1.1)	15	3	0.6 (0.6–0.7)	15	4	1.0 (0.9–1.0)	15	2	0.6 (0.5–0.6)	15	5	1.1 (1.1–1.1)	15	1
Cardiomyopathy and myocarditis (Total)	7.8 (7.7–7.9)	-	9.9 (8.8–11.2)	-	2	4.1 (3.9–4.3)	-	5	13.1 (12.8–13.3)	-	1	5.4 (5.2–5.6)	-	4	7.5 (7.4–7.6)	-	3
Myocarditis	0.3 (0.2–0.3)	17	0.4 (0.3–0.5)	17	2	0.2 (0.1–0.2)	17	5	0.4 (0.4–0.5)	17	1	0.2 (0.2–0.2)	17	4	0.2 (0.2–0.2)	17	3
Alcoholic cardiomyopathy	1.5 (1.4–1.5)	13	2.8 (2.5–3.3)	13	2	0.7 (0.7–0.8)	14	5	2.9 (2.8–3.0)	12	1	1.0 (0.9–1.1)	13	4	1.4 (1.3–1.4)	14	3
Other cardiomyopathy	6.1 (6.0–6.2)	7	6.7 (5.8–7.7)	6	2	3.2 (3.0–3.4)	7	5	9.8 (9.5–10.0)	5	1	4.2 (4.0–4.4)	7	4	5.9 (5.8–6.0)	7	3
Pulmonary arterial hypertension	0.5 (0.4–0.5)	16	0.5 (0.4–0.6)	16	2	0.2 (0.2–0.2)	16	5	0.6 (0.6–0.7)	16	1	0.3 (0.3–0.3)	16	4	0.5 (0.4–0.5)	16	3
Atrial fibrillation and flutter	9.3 (9.2–9.5)	5	7.6 (6.4–8.8)	5	2	4.6 (4.4–4.9)	5	5	6.8 (6.5–7.0)	8	3	5.6 (5.4–5.8)	5	4	10.3 (10.1–10.4)	5	1
Aortic aneurysm	3.1 (3.1–3.2)	10	2.9 (2.4–3.5)	12	3	2.2 (2.1–2.4)	10	4	3.7 (3.6–3.9)	11	1	1.9 (1.8–2.0)	11	5	3.2 (3.2–3.3)	10	2
Lower extremity peripheral arterial disease	2.8 (2.8–2.9)	11	3.4 (2.8–4.0)	10	2	1.1 (1.0–1.2)	11	5	4.3 (4.2–4.5)	10	1	2.0 (1.9–2.2)	10	4	2.8 (2.8–2.9)	12	3
Endocarditis	2.6 (2.6–2.7)	12	3.1 (2.6–3.6)	11	1	1.1 (1.0–1.2)	12	5	2.7 (2.6–2.8)	13	3	1.6 (1.5–1.7)	12	4	2.8 (2.8–2.9)	11	2
Other cardiovascular and circulatory diseases	5.2 (5.1–5.3)	8	5.5 (4.8–6.4)	8	2	2.3 (2.1–2.5)	9	5	8.4 (8.2–8.7)	6	1	3.1 (3.0–3.3)	9	4	5.1 (5.1–5.2)	8	3

Age-standardized mortality rates are presented as estimate (95% Uncertainty Interval) with units deaths per 100,000 population. The cause rank is the rank among the 17 most detailed causes within a given racial and/or ethnic population. The race rank is the rank among racial and/or ethnic groups within a given cause.

National-level disparities by race and/or ethnicity.

In 2019, total CVD ASMR was highest in the Black population (304.8 [302.7–306.9]), followed by AIAN (256.7 [228.9–288.2]), White (240.9 [240.2–241.7]), and Latino (178.3 [174.5–182.0]) populations, and lowest in the Asian population (140.3 [136.2–144.2]); this ordering persisted over the study period for total CVD, though it varied for selected detailed causes (Central Illustration). In 2019, mortality was highest in the Black population for 12 of 17 detailed causes; highest in the White population for three causes (non-rheumatic calcific aortic valve disease, non-rheumatic degenerative mitral valve disease, atrial fibrillation and flutter); and highest in the AIAN population for two causes (rheumatic heart disease, endocarditis). Mortality was lowest for the Asian population for 13 causes and for the Latino population for four causes.

Central Illustration: National age-standardized mortality rates by CVD cause of death, racial and/or ethnic population, and year, 2000–2019.

AIAN = American Indian or Alaska Native

Evidence of slowing progress in CVD mortality reduction

From 2000 to 2019, total CVD age-standardized mortality (ASMR) declined 34.0% (95% UI 33.8–34.2), an absolute reduction of 122.6 (121.6–123.7) deaths per 100,000. Mortality declined in 12 of 17 detailed causes; the largest relative declines were for aortic aneurysm (53.1% [52.0–54.2]), ischemic heart disease (44.8% [44.5–45.1]), and other cardiomyopathy (39.8% [38.6–41.0]), while the largest absolute declines were for ischemic heart disease (103.1 [102.0–104.1]), ischemic stroke (13.9 [13.6–14.2]), and intracerebral hemorrhage (5.2 [5.0–5.3]). Statistically significant increases occurred for atrial fibrillation and flutter (20.3% [17.9–23.1]; 1.6 [1.4–1.8]), other cardiovascular and circulatory diseases (53.6% [49.4–58.0]; 1.8 [1.7–1.9]), and hypertensive heart disease (60.1% [58.0–62.2]; 7.5 [7.3–7.7]). Declines in total CVD were steeper in 2000–2011 (3.2% [3.2–3.2] annualized) than 2011–2019 (0.8% [0.8–0.9]). Among the 12 causes with overall declines, seven slowed after 2011 and three showed small increases in 2011–2019 (rheumatic heart disease 0.5% [−0.1–1.1], non-rheumatic degenerative mitral valve disease 0.9% [0.3–1.5], ischemic stroke 0.9% [0.8–1.0]).

Discussion

Most previous studies on CVD mortality disparities have emphasized national- or state-level trends, single conditions, and aggregated racial and/or ethnic groups or Black-White comparisons. In contrast, our study’s core contribution is a single, coherent analytic framework that produces internally-consistent estimates at the county-by-cause-by-population level using 21 CVD causes, five racial and/or ethnic populations, including adjustment for racial and ethnic misclassification and calibration across causes. This lens reveals divergences and vulnerabilities that state or national summaries obscure—substantial within-state and within-group heterogeneity, population-specific high-risk geographies, and cause-specific spatial patterns that do not uniformly track “total CVD.” By showing disparities at this geographic granularity, the county-level findings clarify where burden concentrates, where outcomes are comparatively favorable, and which causes shape local disparities, providing a more actionable map for interpretation and response than aggregated reporting can offer.

County-Level Findings

Our findings for the period of 2000–2019 establish a pre-pandemic baseline in which county-level CVD mortality was highest among Black adults, elevated among AIAN populations, and concentrated in socioeconomically-deprived Southern and reservation-adjacent counties with large Black, Latino, and AIAN populations. Wadhera, et al.²⁷ have previously shown that during the pandemic, Black, Hispanic, and Asian populations saw the greatest increases in cardiovascular mortality. Kobo, et al. showed that CVD mortality remained elevated through 2023, especially for stroke and hypertension-related mortality.²⁸ Although total CVD mortality saw a pandemic surge with partial post-pandemic recovery, comparably detailed county-level analyses by race and/or ethnicity as a pre-pandemic baseline remain limited.

National-Level Findings

Nationally, the CVD mortality profile was dominated by IHD and stroke, but the distribution of leading causes was not uniform across race and/or ethnicity. Differences in cause rankings—particularly the relative prominence of hypertensive heart disease in AIAN populations and the distinct composition of leading causes in Black populations—suggest that “total CVD” obscures meaningful heterogeneity in the pathways contributing to mortality across groups. Across the entire study period, large and persistent racial and/or ethnic disparities were evident. Black populations consistently bore the highest overall CVD mortality burden and had the highest mortality for most detailed CVD causes, indicating that disparities were not confined to a single diagnostic category. Conversely, Asian populations had the lowest mortality for most causes, underscoring substantial between-group differences in risk and/or protection that remained stable at the national level even as overall rates changed.

Although total CVD mortality declined substantially from 2000 to 2019, the pace of improvement clearly slowed after 2011. This deceleration was mirrored in several detailed causes and was accompanied by adverse trends in specific conditions: hypertensive heart disease and atrial fibrillation and flutter increased overall, and ischemic stroke showed a post-2011 increase despite longer-term declines. Together, these patterns indicate that recent national progress has been uneven—continued improvement in many causes co-occurring with worsening trends in conditions closely linked to blood pressure–related disease and arrhythmia.

Although our analysis includes more racial and/or ethnic populations than are typically included in county-level research, there is nonetheless variation within these populations that is not captured by the current study. For example, one study found that age-standardized CVD mortality among adults aged 35 and older was 1.5 times higher among NHPI compared to Asian individuals.²⁹ This finding emphasizes the importance of differentiating small populations that have traditionally been included with a larger group. Among Latino subpopulations, another study from 2003–2012 mortality data found that Puerto Rican populations had total CVD mortality like White populations compared to much lower CVD mortality for Mexican heritage Latinos.³⁰ Stroke mortality, however, was higher among Mexican heritage populations compared to Puerto Rican and Cuban populations.³⁰ Analysis of subpopulation differences within a race and/or ethnic group will continue to be a research challenge.

Worsening CVD Risk Factor Prevention and Control

These adverse mortality findings in total CVD, hypertensive heart disease, stroke (especially ischemic stroke), and atrial fibrillation are consistent with reports showing lack of progress in blood pressure control in the US from 2009 to 2020.³¹ In 2017 to 2020, only 48.2% of persons with hypertension were controlled (defined as systolic BP <140 mm Hg and diastolic BP <90 mm Hg) compared to 52.8% in 2009 to 2012 (p=0.034 for trend), and even among those on antihypertensive medications, no progress had been made with about two-thirds of persons controlled.³¹ The importance of these findings is also mirrored in a population-based cohort study conducted from January 2015 to December 2019 using Medicare administrative data for AIAN adults aged 65 years and older, where hypertension prevalence was 72.2%, half of all participants had at least one severe cardiovascular condition, and the incidence of acute myocardial infarction and heart failure increased during the study period.³²

Hypothesized Drivers of Adverse CVD Mortality Disparities and Trends

While our study was not designed to identify the specific drivers of adverse cardiovascular mortality, the consistent presence of racial and/or ethnic disparities across states and counties suggests an important role for shared upstream root causes of CVD. Importantly, counties with the lowest mortality rates within the AIAN and Black populations had total CVD mortality rates that were significantly lower than the national average for the White population, reiterating that high CVD mortality rates are not an intrinsic racial and/or ethnic phenomenon but rather are in large part driven by both social and environmental factors as well as individual behavioral, clinical, and related cardiometabolic risk factors.

There is robust published evidence from pooled cohort analyses and large observational studies showing that social, behavioral, clinical, and cardiometabolic risk factors are independently associated with higher CVD mortality and explain much of the observed differences across racial and/or ethnic groups.^33,34 Recent studies in AIAN populations similarly report high and rising burdens of diabetes and hypertension, with authors warning that, without targeted prevention, widening disparities in these risk factors are likely to worsen disparities in CVD outcomes.³² Additionally, national analyses of CVD mortality through 2022 show that excess CVD deaths during the COVID-19 pandemic were greatest in demographic groups with higher prevalences of hypertension, diabetes, obesity, and other cardiometabolic risks, underscoring the role of underlying risk factor burden.^14,35

Study Period and Pandemic-Era Relevance

Our analysis extends through 2019, the most recent data available at the time of statistical analysis. Although this predates the COVID-19 pandemic, our pre-pandemic estimates are analytically indispensable for contextualizing the heterogenous post-pandemic CVD mortality landscape. Several previous analyses have documented increases in age-adjusted total and cause-specific CVD mortality rates across many US states during the pandemic and in some states between 2019 and 2023.^14–16,28 Understanding whether current mortality levels represent a temporary disruption or a permanent upward shift requires knowing the trajectory and rate of improvement that existed beforehand. Importantly, our data show that progress in reducing total CVD mortality had already slowed markedly after 2011, and that mortality for hypertensive heart disease, atrial fibrillation and flutter, and ischemic stroke was already increasing or plateauing before the pandemic began. The pandemic therefore did not strike a system on a strong improving trajectory; rather, it accelerated trends that were already concerning in specific causes and populations. Distinguishing pandemic-attributable excess mortality from pre-existing vulnerability is only possible with the kind of rigorous, cause-specific, county-by-race and/or ethnicity baseline this study provides, and doing so is essential for setting realistic recovery benchmarks and for identifying communities where post-pandemic mortality has returned to, or exceeded, pre-pandemic high-risk levels.

Public Health Implications

These findings are directly actionable for clinical practices and public health agencies in setting policies, designing programs, and planning interventions. CVD mortality is driven by well-characterized and modifiable risk factors as well as by socioeconomic deprivation and disparities in access to and quality of care. For many of these risk factors, effective clinical tools already exist, but their population-level impact depends on ensuring patients are identified, treated, and sustained in care. Our county-by-cause-by-race and/or ethnicity data identify where burden is highest and which causes dominate locally, enabling resources to be directed where they are most needed rather than where average national or state conditions suggest.

Critically, because our analytic framework is designed for repeated application as new data become available, these estimates serve not only as a targeting tool but as an evaluation platform, allowing assessment of whether specific interventions, programs, and policies are narrowing geographic and racial and/or ethnic gaps over time. At the county level, health departments, health systems, and community organizations can use these data to prioritize outreach, identify communities for partnership, and tailor interventions to local burden. At the state level, health agencies can allocate resources to counties performing poorly relative to state and national peers. At the federal level, these estimates provide a mechanism for monitoring whether national CVD prevention programs are reaching the highest-burden communities and whether disparities are narrowing.

Limitations and Strengths

This study has several limitations. First, each of the data inputs (deaths, population, and covariates) is subject to error. For example, atrial fibrillation deaths have often been miscoded as stroke deaths. Increased detection and recognition of atrial fibrillation as an underlying cause has resulted in a shift in coding. While our data processing approach is designed to address these issues, some bias likely remains. This may contribute to the large apparent increase in atrial fibrillation mortality estimated for some racial and/or ethnic populations. Second, our estimates are associated with often substantial uncertainty. This is especially the case for small populations and for racial and/or ethnic populations that require greater adjustment for misclassification on death certificates (particularly the AIAN population).

However, this study also has several important strengths. Our methodological approach robustly addresses three key challenges with mortality data, namely the use of intermediate, insufficiently specific, or otherwise implausible underlying cause of death codes; racial and ethnic misclassification on death certificates; and statistical noise arising from the small population size in many counties and racial and/or ethnic groups. Our analysis is also uniquely comprehensive, covering 3,110 counties, five racial and/or ethnic groups, and 21 causes of CVD death over a 20-year period. These estimates were generated using a consistent methodology, enabling comparison across causes of death.

Conclusions

Sustained declines in CVD mortality signal meaningful progress, yet profound racial and/or ethnic and place-based disparities persist. Importantly, national or state averages conceal where excess mortality concentrates. Our county-by-cause-by-population perspective sharpens this reality: mortality varies far more across counties than across states, and variation within racial and/or ethnic populations remains substantial. Moreover, specific CVD causes show large county differences, with some causes tracking total CVD geography and others exhibiting distinct spatial signatures—an observation that elevates cause-specific epidemiology as central to interpretation and action in clinical and public health settings.

For clinical practice, these findings point to the value of integrating local epidemiology into risk assessment, outreach, and care delivery. Health systems and clinicians need to align prevention strategies with the dominant local causes and the populations experiencing the greatest burden. In a clinical context, this means addressing improved blood pressure control, management of other risk factors, and promoting healthier behaviors. For public health practice, the magnitude of within-state heterogeneity supports targeting resources to high-burden counties and evaluating progress using county-level metrics. For biomedical research, the observed heterogeneity clarifies two complementary opportunities: deeper etiologic study of upstream, place-based drivers and community-engaged implementation research efforts that are locally tailored and sustained, with evaluation designs capable of detecting whether interventions narrow geographic and racial and/or ethnic gaps. Our findings provide an actionable foundation for solution-oriented health disparities research aligned with the new unified strategy for advancing the NIH mission.³⁶

Clinical Perspectives.

The findings from this study have clinical implications for at least four areas: (1) guideline-directed intensification of prevention, treatment, and control of hypertension and other cardiovascular risk factors, informed by local, county-level data on disease and risk-factor burden in specific populations; (2) improved detection and guideline-directed management of atrial fibrillation and flutter; (3) targeted deployment of clinical resources using county-level data, prioritizing high-burden counties and historically underserved racial and/or ethnic populations; and (4) routine stratification of quality measures and outcomes by race and/or ethnicity and geography to identify gaps, guide interventions, and track progress over time

Abbreviations List

AIAN

American Indian or Alaska Native

ASMR

Age-standardized mortality rate

BMI

Body mass index

CVD

Cardiovascular disease

GBD

Global Burden of Diseases, Injuries, and Risk Factors Study

ICD

International Classification of Diseases

NHPI

Native Hawaiian or Pacific Islander