A, Age-standardized all cardiovascular disease (CVD) mortality rates, 2012 to 2015. B, Average annual percentage change of all CVD mortality rates, 2000 to 2015.
eFigure 1. Average annual percent change of CVD mortality rates by states, stratified by race/ethnicity, sex, and age groups, 2000-15
eFigure 2. Average annual percent change of mortality rates due to hypertensive heart disease by states, 2000-15
eFigure 3. County-level (A) age-standardized hypertensive heart disease mortality rates, 2012-2015; (B) average annual percent change of hypertensive heart disease mortality rates, 2000-2015, and (C) multivariate quasi-Poisson regression by county-level risk factors 2012-15
eTable 1. Trends and age-standardized premature mortality (age 25-64 years) rates due to specific cardiovascular disease in the US, 2000-03 compared to 2012-15, rates per 100,000
eTable 2. Relative risk and 95% confidence intervals for county-level CVD premature mortality by country-level risk factors, adjusting for age and all five county-level risk factors, 2012-15
eTable 3. Relative risk and 95% confidence intervals (CI) for county-level premature mortality due to hypertensive heart disease by county-level risk factors, adjusting for age and all five county-level risk factors, 2012-15
Customize your JAMA Network experience by selecting one or more topics from the list below.
Chen Y, Freedman ND, Albert PS, et al. Association of Cardiovascular Disease With Premature Mortality in the United States. JAMA Cardiol. 2019;4(12):1230–1238. doi:10.1001/jamacardio.2019.3891
How have the trends in premature mortality from cardiovascular disease (CVD) changed in the United States during 2000 to 2015 by demographics and county-level factors, including education, rurality, and the prevalence of smoking, obesity, and diabetes?
In this study, CVD mortality rates increased significantly among American Indian/Alaska Native individuals age 25 to 49 years and plateaued among white women in the same age group. County-level diabetes prevalence showed the strongest association with mortality from CVD.
Identifying groups/characteristics of counties where rates and trends are substantially high is important for targeted public health interventions.
Cardiovascular disease (CVD) is a leading cause of morbidity and mortality in the United States. Despite substantial declines in CVD mortality rates during past decades, progress against cardiovascular deaths in midlife has stagnated, with rates increased in some US racial/ethnic groups.
To examine the trends in premature (ages 25-64 years) mortality from CVD from 2000 to 2015 by demographics and county-level factors, including education, rurality, and the prevalence of smoking, obesity, and diabetes.
Design, Setting, and Participants
This descriptive study used US national mortality data from the Surveillance, Epidemiology, and End Results data set and included all CVD deaths among individuals ages 25 to 64 years from January 2000 to December 2015. The data analysis began in February 2018.
Age, sex, race/ethnicity, and county-level factors.
Main Outcomes and Measures
Age-standardized mortality rates and average annual percent change (AAPC) in rates by age, sex, race/ethnicity, and county-level factors (in quintiles) and relative risks of CVD mortality across quintiles of each county-level factor.
In 2000 to 2015, 2.3 million CVD deaths occurred among individuals age 25 to 64 years in the United States. There were significant declines in CVD mortality for black, Latinx, and Asian and Pacific Islander individuals (AAPC: range, −1.7 to −3.2%), although black people continued to have the highest CVD mortality rates. Mortality rates were second highest for American Indian/Alaskan Native individuals and increased significantly among those aged 25 to 49 years (AAPC: women, 2.1%; men, 1.3%). For white individuals, mortality rates plateaued among women age 25 to 49 years (AAPC, 0.05%). Declines in mortality rates were observed for most major CVD subtypes except for ischemic heart disease, which was stable in white women and increased in American Indian/Alaska Native women, hypertensive heart disease, for which significant increases in rates were observed in most racial/ethnic groups, and endocarditis, for which rates increased in white individuals and American Indian/Alaska Native men. Counties with the highest prevalence of diabetes (quintile 5 vs quintile 1: relative risk range 1.6-1.8 for white individuals and 1.4-1.6 for black individuals) had the most risk of CVD mortality.
Conclusions and Relevance
There have been substantial declines in premature CVD mortality in much of the US population. However, increases in CVD mortality before age 50 years among American Indian/Alaska Native individuals, flattening rates in white people, and overall increases in deaths from hypertensive disease suggest that targeted public health interventions are needed to prevent these premature deaths.
Cardiovascular diseases (CVDs) are the leading cause of mortality and morbidity in the United States, accounting for more than 900 000 deaths and 14 million disability-adjusted life-years in 2016 alone.1 Substantial progress has been made in combating CVD over the past 40 years, with mortality rates falling from 507 in 100 000 in 1980 to 253 in 100 000 in 2014.2 However, considerable disparities by race/ethnicity and geographic region persist.2-4 The American Heart Association has set a strategic goal of reducing deaths from CVD by 20% during 2010 to 2020.5 Similar goals have been proposed as part of the Million Hearts Initiative, which aims to prevent a million cardiovascular events during 2017 to 2022.6
Nevertheless, recent evidence suggests that progress made against premature mortality from heart disease has stagnated and rates have even increased in some US subgroups over the past 15 years.7 However, it remains unclear how trends in the specific type of CVD vary across population subgroups as well as the associated contributions of specific risk factors to CVD mortality. To better characterize 2000 to 2015 trends in rates of CVD premature mortality in the United States, we describe trends and patterns in cardiovascular deaths among people age 25 to 64 years by type of CVD, age group, sex, race/ethnicity, state, and county-level characteristics, including education, rurality, and the prevalence of smoking, obesity, and diabetes. The selection of this group is because (1) CVD death rates continue to decline among individuals 65 years and older in the United States8 and (2) the reduction of premature mortality is one of the United Nations Sustainable Development Goals9 and CVD mortality in the United States is divergent. A more detailed characterization of trends over time may provide insight into factors associated with increases and decreases in CVD deaths, which ultimately may inform the prioritization of prevention strategies at local and national levels.
Causes of death and demographics were ascertained from national death certificate data from the Surveillance, Epidemiology, and End Results data set, which contains national mortality data from 2000 to 2015. Institutional review board approval and informed consent were waived because the study used publicly available deidentified data. This analysis focused on premature deaths due to CVD, defined based on the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes: all CVD (I00-02, I05-09, I10-15, I20-I25, I26-I28, I60-69, I70-79, and I30-51), including ischemic heart disease (I20-25), cerebrovascular disease (I60-69), rheumatic heart disease (I00-02 and I05-09), hypertensive heart disease (I10-15), peripheral arterial disease (I70-79), heart failure (I42, I43, I50), cardiac arrest (I46), arrhythmia (I44-49), and endocarditis (I33 and I38). Age-, sex-, and race/ethnicity–specific data were ascertained from the US Census intercensal populations.
We assessed the contribution of 5 county-level risk factors to CVD premature mortality: population with a bachelor degree (%), rurality, prevalence of smoking, diabetes, and obesity (defined as a body mass index [calculated as weight in kilograms divided by height in meters squared] of ≥30). We selected these 5 county-level factors based on their association with CVD mortality. Cardiovascular deaths vary by socioeconomic status. Therefore, we included county-level bachelor degree and rurality as proxies for socioeconomic status. We also included smoking, obesity, and diabetes prevalence because they are common risk factors for CVD and mortality.
Because county-level socioeconomic status (SES) characteristics from 2011 to 2015 have been reported to be most relevant to current mortality rates,10 we ascertained the percentage of individuals with a bachelor degree (based on a population 25 years or older) from the 2011 to 2015 Census American Community Survey.11 The rurality of counties was defined based on the 2013 Rural-Urban Continuum codes that were developed by the US Department of Agriculture.12 Data on county-level smoking prevalence during 2008 to 2010 for 18 years or longer were ascertained from the Behavioral Risk Factor Surveillance System and the National Health Interview Survey.13,14 County-level obesity and diabetes prevalence for a population 20 years or older were estimated based on the Institute for Health Metrics and Evaluation.15
We estimated age-standardized mortality rates (ASRs) overall and stratified by calendar period (2000–2003 and 2012–2015), age group (25-49 years and 50-64 years), sex, and race/ethnicity (non-Hispanic white [ie, white], non-Hispanic black [ie, black], Asian and Pacific Islander, American Indian and Alaska Native, and Latinx) using the standard 2000 US population in 5-year age groups as the reference. The selection of these 2 age groups was based on published data examining premature all-cause mortality in the United States by county using the US death certificate data.10 We restricted estimates for American Indian/Alaska Native individuals to contract health services delivery areas to reduce racial/ethnic misclassification.16 We also estimated the average annual percentage change (AAPC) in rates during 2000 to 2015 by age group, sex, state, and race/ethnicity stratified by CVD types. Analyses were conducted using SEER*Stat software (National Cancer Institute).
We then estimated age-standardized CVD mortality rates during 2012 to 2015 among white, black, and Latinx individuals by each county-level factor. We categorized counties by the following county-level factors: (1) percentage with a bachelor degree (1%-20%, 21%-26%, 27%-30%, 31%-37%, and 38%-79%), (2) rurality (metropolitan areas with ≥1 million people, metropolitan areas with 250 000 to <1 million people, metropolitan areas of <250 000 people, urban areas with ≥20 000 people, urban areas with 2500 to <20 000 people, and completely rural populations with <2500 people), (3) smoking prevalence (6%-20%, 21%-23%, 24%-26%, 27%-29%, and 30%-43%), (4) obesity prevalence (18%-34%, 35%-36%, 37%-38%, 39%-40%, and 41%-53%), and (5) diabetes prevalence (5%-8.5%, 8.6%-9.7%, 9.8%-10.6%, 10.7%-11%, and 12%-21%). Missing values were categorized to a separate group. Because we aggregated counties, we included all counties regardless of population size and number of deaths. We also estimated the annual percentage change in mortality rates during 2000 to 2015 by the level of each county-level factor.
We then conducted a multivariate quasi-Poisson regression, adjusting for age (5-year age group) and 5 county-level factors to examine the associations of these risk factors with recent CVD mortality rates (rates in 2012-2015), including all CVD, ischemic heart disease, and hypertensive heart disease. Because the total number of events was too sparse for certain races/ethnicities after stratifying by county-level risk factors, we restricted the regression analyses to white and black individuals. Regression analyses were conducted using the R program, version 3.4.2 (R Foundation) and statistical significance was set at P < .05.
During 2000 to 2015, more than 2.3 million CVD deaths occurred among individuals aged 25 to 64 years in the United States. Among women, the ASR declined by 20% from 60 in 100 000 during 2000 to 2003 to 48 in 100 000 during 2012 to 2015 (AAPC, −1.9%). Similarly, among men, the ASR declined by 20% from 130 in 100 000 during 2000 to 2003 to 104 in 100 000 during 2012 to 2015 (AAPC, −1.8%).
There was substantial variation in total CVD mortality rates and trends by racial/ethnic and age groups. Among women, black individuals had the highest ASR during 2012 to 2015 (101/100 000), followed by American Indian/Alaska Native (74/100 000), white (44/100 000), and Latinx individuals (28/100 000). Asian Pacific Islander individuals had the lowest ASR (19/100 000). Similarly, in men, ASR during 2012 to 2015 were highest among black individuals (190/100 000), followed by American Indian/Alaska Native (161/100 000), white (100/100 000), Latinx (69/100 000), and Asian Pacific Islander individuals (55/100 000) (Table).
Black individuals showed statistically significant declines in ASR during 2000 to 2015 for women and men (AAPC, −2.9% in women; −2.4% in men), along with Latinx (−3.0% in women; −2.3% in men), Asian Pacific Islander (−3.2% in women; −1.7% in men), and white individuals (−1.3% in women; -1.9% in men). In contrast, ASR increased significantly among American Indian/Alaska Native individuals age 25 to 49 years (AAPC: women, 2.1%; men, 1.3%), and ASR among white women aged 25 to 49 years plateaued (AAPC, 0.05%) (Table).
The analysis of state-specific trends was limited to black and white individuals because of sparse data in other racial/ethnic groups. During 2000 to 2015, most groups of black and white individuals showed significant declines in CVD mortality, particularly among men aged 50 to 64 years; whereas white men had significant declines in all states, black men had more pronounced declines in most states, except Utah (AAPC = 2.8%) (Figure 1). However, compared with other groups, white women aged 25 to 49 years showed considerable variations in trends, with AAPCs ranging from −3.3% in Oregon to 2.3% in Alabama (eFigure 1 in the Supplement).
Ischemic heart disease (54% of all CVD premature deaths), cerebrovascular disease (13%), and hypertensive heart disease (10%) contributed to the largest fraction of CVD premature deaths from 2000 to 2015. Declines in ASR due to ischemic heart disease were the primary contributor to the decline in overall CVD mortality, with decreases in most age and racial/ethnic groups (AAPC: range, −4.7 to −0.5%). Mortality declines in ASR were also observed for cerebrovascular disease (−4.5% to −0.4%), peripheral arterial disease (−5.5% to −0.1%), and rheumatic heart disease (−6.6% to −0.3%) (eTable 1 in the Supplement). In contrast, ASRs of ischemic heart disease increased among American Indian/Alaska Native women aged 25 to 49 years (AAPC: 1.7%) and were stagnant for white women aged 25 to 49 years (eTable 1 in the Supplement).
During 2000 to 2015, ASR from hypertensive heart disease, which was the third most common cause of CVD deaths in 2012 to 2015, increased in most racial/ethnic groups (Figure 2). Compared with other groups, white women aged 25 to 49 years had the highest increase in hypertensive heart disease mortality. An additional analysis of trends showed increases in ASR of hypertensive heart disease in most states (AAPC, from 0.4% in South Carolina to 19.7% in South Dakota), except Washington, DC (−2.5%) and Vermont (−1.6%) (eFigure 2 in the Supplement). The ASR of endocarditis increased significantly in white individuals aged 25 to 49 years (AAPC: women, 3.9%; men, 3.7%) and American Indian/Alaska Native men (5.6%) (Figure 2).
Recent CVD premature mortality rates (2012–2015) varied substantially by county-level risk factors. The ASRs were consistently higher in counties with lower socioeconomic factors (ie, lower education levels and rural areas). The ASRs were also generally higher in counties with a high prevalence of diabetes, obesity, and smoking. These increases followed a gradient across quintiles of county-level factors and were more pronounced in white than black and Latinx individuals (Figure 3A). During 2000 to 2015, ASR among black and Latinx individuals declined over time across counties regardless of county-level factors. In contrast, ASR among white women aged 25 to 49 years predominantly increased among counties with a lower percentage of bachelor degrees, and a higher prevalence of diabetes, obesity, and smoking and among more rural counties (Figure 3B). A multivariate quasi-Poisson regression confirmed associations between county-level factors and the risk of CVD mortality, in which county-level diabetes prevalence showed the strongest effect (Figure 4). The associations were more pronounced in white individuals aged 25 to 49 years than black (Figure 4). For example, for women aged 25 to 49 years, compared with counties with a diabetes prevalence of 5% to 8.5%, the risk of CVD mortality was 1.4 times higher among white individuals and 1.3 times higher among black individuals who lived in counties with a diabetes prevalence of 9.8% to 10.6%; the risk of CVD mortality was 1.8 times higher among white individuals and 1.4 times higher among black individuals in counties with a diabetes prevalence of 12% to 21% (Figure 4 and eTable 2 in the Supplement).
As mortality rates of hypertensive heart disease increased significantly across most age, sex, and racial/ethnic groups, we further examined rates and trends by county-level factors. Age-standardized mortality rates of hypertensive heart disease were generally higher in counties with a high prevalence of diabetes and obesity, but the associations between county-level smoking prevalence and education levels and mortality rates were less pronounced (eFigure 3 in the Supplement). Increases occurred across many types of counties but generally followed a gradient across quintiles of county-level factors, although these patterns were less clear among black and Latinx individuals (eFigure 3 in the Supplement). The multivariate quasi-Poisson regression showed a predominant association between county-level diabetes prevalence and hypertensive heart disease mortality (eFigure 3 and eTable 3 in the Supplement).
During 2000 to 2015, we observed significant declines in CVD premature mortality rates for black, Latinx, and Asian and Pacific Islander individuals but significant increases in rates among American Indians/Alaska Native individuals age 25 to 49 years; rates plateaued among white women aged 25 to 49 years. These increases were driven by a lack of progress against mortality from ischemic heart disease and increases in mortality from hypertensive heart disease and endocarditis.
The American Heart Association has set a strategic goal for reducing deaths from CVD by 20% from 2010 to 2020.5 While there was an overall 28% decrease in CVD mortality rates in the United States during 2003 to 2015,17 reflecting the progress made in prevention and advances in cardiovascular disease treatment, equivalent declines have not been observed across age groups.18 Improvements in mortality also vary by race/ethnicity. A recent US Centers for Disease Control and Prevention report examining mortality trends found that heart disease death rates increased for white people aged 45 to 64 years during 2009 to 2017 and for black individuals aged 45 to 64 years during 2011 to 2017 despite an overall of 22% of decline in rates from 1999 to 2011.19 In this study, we compared premature mortality due to all CVD during 2012 to 2015 with rates from 2000 to 2003. While we observed declines in CVD mortality, we found stagnation in ischemic heart disease mortality among young white women and significant increases in mortality from hypertensive heart disease across most races/ethnicities. This highlights the importance of understanding the diversity and disparities in CVD mortality by risk factors among different groups.
Disparities in CVD mortality by SES have been well established in the United States.20 Important risk factors, including smoking, obesity, and diabetes, are more prevalent among lower-SES counties compared with higher-SES counties.21 As observed previously,2,7 we found substantial differences in mortality rates by age, sex, and racial/ethnic groups across different types of counties. For example, CVD mortality rates were generally lower in high-SES counties and more urban areas. These findings highlight potential opportunities among states to target public health resources for primary and secondary CVD prevention toward counties based on SES status and rurality.22
In contrast to the trends for overall CVD mortality, we found marked increases in premature mortality associated with hypertensive heart disease across most age and racial/ethnic groups. This increase occurred across the country, although gradients by county characteristics were noted. Previous data suggest that only half of US individuals with hypertension had their condition effectively controlled.23 Consistent with our data, the proportion of US mortality rates attributable to hypertension increased by 10.5% from 2005 to 2015 and the absolute number of deaths attributable to hypertension rose by 37.5%.24 Hypertension is common in patients with diabetes25 and patients with diabetes and hypertension experience more severe cardiovascular outcomes than patients with either condition alone.25 We found that compared with other risk factors, diabetes prevalence had the strongest association with premature mortality from hypertensive heart disease.
Without widespread intervention, continued increases in diabetes prevalence in the United States will likely exacerbate the adverse trends observed here. Coupled with this, increases in premature mortality associated with hypertensive heart disease are most concerning given the change in the hypertension definition in the 2017 clinical guidelines.26 The new guidelines markedly reset the diagnosis and treatment of hypertension, which potentially results in more individuals receiving a hypertension diagnosis.27 Public health policies to mitigate these trends in CVD mortality will need to focus on diagnosis, treatment, and control of hypertension in younger populations, particularly in lower-resourced, underserved communities. In addition, continuously monitoring mortality from hypertensive heart disease is important to evaluate the effect of changes in the clinical guideline and how this may shape the CVD mortality rates overall.
The significant increase in premature mortality that is associated with endocarditis among young white individuals and young American Indian and Alaska Native men is also worrisome, despite relatively few deaths. Coincident with the marked rise in deaths from opioid use, increase in endocarditis among certain groups may be associated with the opioid epidemic. Although this study could not directly address this hypothesis, a disproportionate increase in deaths from endocarditis has been found among young white individuals who inject drugs.28 In 2016, approximately 1 million people used heroin in the United States, which was 2-fold higher than the total number of heroin users in the period over 2002 to 2013.29 Consequently, many of the population were at risk of bacteremia from contaminated needles and subsequent endocarditis. Future studies investigating the association of injection drug use with trends in endocarditis are needed.
A strength of this study was the stratification of the analysis by age groups, allowing us to identify trends that would be missed across the full age range by age standardization. Another strength of this study was the estimation and comparison of county-level CVD risk factors. County-level estimates enable the identification of local trends and patterns of CVD mortality that are associated with important risk factors. However, as macro-level estimates, county-level risk factors do not represent the factors of specific individuals who died of CVD nor age- and sex-specific variations within each county. Additionally, county-level risk factor estimates were restricted to white, black, and Latinx individuals as there were fewer deaths occurring among other groups, which precluded systematic examination by county.
Another limitation of this study is the use of death certificate data. Racial/ethnic misclassification is a main concern, as race/ethnicity data on death certificates are typically recorded by funeral directors, whereas self-reported information on the census is used to calculate the denominators in this study. Nevertheless, race/ethnicity recording on death certificates has been reported as highly accurate for white and black individuals, and misclassification in Latinx and Asian and Pacific Islander individuals is minor.30 To reduce misclassification in American Indian and Alaska Native individuals, we restricted the analysis to contract health services delivery areas.16 There are also concerns about the possible misclassification of CVD and changes in coding over time. Use of these data for CVD in aggregate has been validated and reduces the potential for misclassification.31,32 However, we should be cautious regarding examining trends in hypertensive heart disease and other types. For example, with increasing emphasis on blood pressure control, deaths from heart failure, coronary artery disease, and stroke may be misclassified as hypertensive heart disease. This misclassification may overestimate the effect of hypertensive heart disease, but it may also underestimate the effect of other heart diseases. Despite this, the use of death certificate data is the most feasible current method to conduct this type of analysis on the US population, as opposed to community-based cohorts in which the findings would have limited generalizability. Additionally, as CVD in aggregate includes several diseases with distinct etiologies, it is important to explore the contributions of individual types of CVD to the overall picture over time.
Identifying groups with increasing CVD mortality rates is important to guide targeted prevention efforts. Recent work identifies important areas of research to better understand the high rates of CVD mortality among American Indian and Alaska Native populations, including work to elucidate the association of adverse childhood experiences and other psychosocial factors with CVD in these communities.33 Future interventions, policies, and systems approaches to promote healthy behaviors and reduce CVD risk are needed among American Indian and Alaska Native youth. For younger women at risk for CVD, including young white women, screening for CVD risk factors must be considered across medical specialties that regularly care for women.34 Tailored interventions to promote cardiovascular health may most easily reach women outside of the clinical setting by leveraging social media and mobile health technology.35
Despite the marked progress that has been made in preventing cardiovascular deaths in the United States, substantial increases in CVD mortality rates have occurred among American Indians and Alaska Native individuals aged 25 to 49 years and rates have plateaued among young white women, reflecting a lack of progress against mortality from ischemic heart disease and increases in mortality rates from hypertensive heart disease and endocarditis in these groups. Mortality from hypertensive heart disease is a growing concern that affects most groups. Counties with a high prevalence of obesity, diabetes, and smoking have continued to experience higher rates of CVD mortality and should be targeted for intervention. Although CVD risk factors are modifiable, sustained efforts at such interventions are needed. Without rapid and sustained progress against cardiovascular risk factors, public health goals to further reduce the burden of cardiovascular disease mortality in the United States are unlikely to be met.
Accepted for Publication: August 8, 2019.
Corresponding Author: Yingxi Chen, MD, PhD, National Cancer Institute, National Institutes of Health, 9609 Medical Center Dr, Room 7E-504, Bethesda, MD 20850 (email@example.com).
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Chen Y et al. JAMA Cardiology.
Published Online: October 16, 2019. doi:10.1001/jamacardio.2019.3891
Correction: This article was corrected on December 18, 2019, to fix a numerical error in the Results and errors in the x-axis labels in Figures 2 and 3.
Author Contributions: Dr Chen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Chen, Freedman, Withrow, Spillane, Berrington.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Chen, Powell-Wiley, Berrington.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Chen, Albert, Shiels, Spillane, Berrington.
Administrative, technical, or material support: Freedman, Powell-Wiley.
Supervision: Freedman, Huxley.
Conflict of Interest Disclosures: Dr Powell-Wiley reported grants from National Institutes of Health during the conduct of the study. No other disclosures were reported.
Funding/Support: This study was funded by the Intramural Research Program of the National Cancer Institute and National Institute on Minority Health and Health Disparities.
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Create a personal account or sign in to: