To provide a consistent comparison by socioeconomic status, an SDI was estimated by state using equally weighted age-sex-state-year–specific geometric means of income per capita, educational attainment, and total fertility rate.
A, Percentage change in age-standardized DALY rate of CVD for men between 1990 and 2016. B, Percentage change in age-standardized DALY rate of CVD for women between 1990 and 2016. C, Age-standardized DALYs from CVD per 100 000 persons in 2016 for men. D, Age-standardized DALYs from CVD per 100 000 persons in 2016 for women.
Drivers explored in this analysis include population growth, population aging, trends in exposure to all risks included in the Global Burden of Disease 2016 Study, and all other unmeasured factors combined. Results are shown for all CVD DALYs by state. The circle on the bar graph indicates the total percentage change.
eMethods 1. Global Burden of Disease overview.
eMethods 2. Outcomes estimations.
eMethods 3. Fatal cause-specific estimation process.
eMethods 4. Nonfatal modeling methods.
eMethods 5. Risk factor cause-specific estimation process.
eTable 1. Disability-adjusted life-years (DALYs) and percentage change of DALYs for all cardiovascular causes by US state, total number and age-standardized rate for 1990, 2006, and 2016 for both sexes.
eTable 2. Age-standardized heart failure prevalence per 100 000 persons for 2016.
eFigure 1. US State rankings for age-standardized cardiovascular disease disability-adjusted life-year rates per 100 000 persons for both sexes combined in 2016.
eFigure 2. Proportion of cardiovascular disease disability-adjusted life-years due to years lived with disability in 2016.
eFigure 3. Leading level 2 cardiovascular risk factors for both sexes for Minnesota and Mississippi.
eFigure 4. US State drivers of change in cardiovascular disease from 1990 to 2016.
eFigure 5. Age-standardized percentage change in disability-adjusted life-year rate between 2010 and 2016 for all cardiovascular diseases.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Global Burden of Cardiovascular Diseases Collaboration. The Burden of Cardiovascular Diseases Among US States, 1990-2016. JAMA Cardiol. 2018;3(5):375–389. doi:10.1001/jamacardio.2018.0385
How does the total burden of cardiovascular diseases vary across US states?
In this study using the Global Burden of Disease methodology, large disparities in total burden of CVD were found between US states despite marked improvements in CVD burden.
These estimates can provide a benchmark for states working to focus on key risk factors, improve health care quality, and lower health care costs.
Cardiovascular disease (CVD) is the leading cause of death in the United States, but regional variation within the United States is large. Comparable and consistent state-level measures of total CVD burden and risk factors have not been produced previously.
To quantify and describe levels and trends of lost health due to CVD within the United States from 1990 to 2016 as well as risk factors driving these changes.
Design, Setting, and Participants
Using the Global Burden of Disease methodology, cardiovascular disease mortality, nonfatal health outcomes, and associated risk factors were analyzed by age group, sex, and year from 1990 to 2016 for all residents in the United States using standardized approaches for data processing and statistical modeling. Burden of disease was estimated for 10 groupings of CVD, and comparative risk analysis was performed. Data were analyzed from August 2016 to July 2017.
Residing in the United States.
Main Outcomes and Measures
Cardiovascular disease disability-adjusted life-years (DALYs).
Between 1990 and 2016, age-standardized CVD DALYs for all states decreased. Several states had large rises in their relative rank ordering for total CVD DALYs among states, including Arkansas, Oklahoma, Alabama, Kentucky, Missouri, Indiana, Kansas, Alaska, and Iowa. The rate of decline varied widely across states, and CVD burden increased for a small number of states in the most recent years. Cardiovascular disease DALYs remained twice as large among men compared with women. Ischemic heart disease was the leading cause of CVD DALYs in all states, but the second most common varied by state. Trends were driven by 12 groups of risk factors, with the largest attributable CVD burden due to dietary risk exposures followed by high systolic blood pressure, high body mass index, high total cholesterol level, high fasting plasma glucose level, tobacco smoking, and low levels of physical activity. Increases in risk-deleted CVD DALY rates between 2006 and 2016 in 16 states suggest additional unmeasured risks beyond these traditional factors.
Conclusions and Relevance
Large disparities in total burden of CVD persist between US states despite marked improvements in CVD burden. Differences in CVD burden are largely attributable to modifiable risk exposures.
Cardiovascular disease (CVD) was the leading cause of death in the United States in 2016, accounting for more than 900 000 deaths.1 Despite large declines in CVD mortality in the late 20th century attributed to advances in public health and health care, improvements in US life expectancy have slowed for some groups, and CVD mortality is no longer improving.2-5 The strongest signal for this alarming trend in US health is identified subnationally at the state and county level, where levels of risk exposure and health vary widely.6-8
Geographic variation in CVD has many determinants, but these are not usually evaluated in a consistent and comparable manner across all states. Rapid changes in average risk at the national level, such as large declines in plasma cholesterol levels over a relatively short period due to increased use of 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors, suggest that subnational evaluation of cardiovascular risk is needed to understand persistent health disparities.9,10 Geographic variation in the quality of primary, prehospital, acute, and long-term cardiovascular care also requires a comprehensive, subnational assessment.11-13
The Global Burden of Disease (GBD) Study 20161 was a study of global health across 332 causes of disease and injury and 84 risk factors in 195 countries and territories. In this article, we report the study’s US state-level results for CVD and its modifiable risk factors.
The methods of the GBD Study 2016 have been reported in detail previously.1,14-16 The study used data on incidence, prevalence, mortality, and risk exposure to produce comparable estimates of disease burden. All analyses were done separately by sex and aggregated by 5-year age categories. A detailed discussion of data sources and methods are provided in eMethods 1 through 5 in the Supplement, with a brief overview below. This study was reviewed and approved by the University of Washington institutional review board, and informed consent was waived because deidentified data were used.
Cardiovascular disease was estimated for the 10 most common global causes of CVD-related death and an additional category that combined all other CVD and circulatory conditions. These causes were ischemic heart disease (IHD), ischemic stroke, hemorrhagic and other stroke, atrial fibrillation, peripheral artery disease, aortic aneurysm, cardiomyopathy and myocarditis, hypertensive heart disease, endocarditis, and rheumatic heart disease. Death due to each underlying CVD cause was defined by categorization of International Classification of Diseases (ICD) codes.1 Disease incidence and prevalence were defined according to a set of standard case definitions mapped to these codes based on expert guidance.14 These included the third universal definition of myocardial infarction, the World Health Organization definition for stroke, electrocardiographic identification of atrial fibrillation, diagnosis of peripheral arterial disease by ankle-brachial index, the World Heart Federation criteria for definite rheumatic heart disease, and the Framingham Heart Study definition of congestive heart failure. Stroke deaths assigned to a non–subtype-specific code (ICD code I64) were reassigned to subtypes using the proportion of ischemic to hemorrhagic strokes.
Data sources and methods for estimation of CVD have been previously described.17 In brief, population counts were obtained from the US Census Bureau for each state.18 Death certificate data were obtained from the National Center for Health Statistics for each state. ICD-9 and ICD-10 codes were aggregated for each cause of CVD. Structured reviews of published literature were performed to identify published and unpublished data on incidence, prevalence, case fatality, and mortality related to CVD causes. State-level inpatient and outpatient claims data were obtained from a database of private and public insurance schemes for 2000, 2010, and 2012.19ICD-9 codes were aggregated for each CVD case definition and used to calculate the annual incidence (using inpatient data) or prevalence rate (using inpatient and outpatient data combined) for selected health conditions, stratified by age, sex, year, and state. A correction factor was applied to account for changes in coding of administrative claims data over time. Data on risk factor exposure were obtained from multiple sources, including the National Health and Nutrition Examination surveys, the Behavioral Risk Factor Surveillance surveys, satellite data and air sampling data for estimation of particulate matter less than 2.5 µm in diameter, and a systematic review of published scientific literature. Surveys with complex sampling design, including National Health and Nutrition Examination surveys and Behavioral Risk Factor Surveillance surveys, were analyzed using appropriate sample weights to accurately estimate variance. Risk factor definitions and attribution methods have been previously reported.15 Definitions of metabolic exposures included fasting plasma glucose level measured in millimoles per liter, total cholesterol level measured in millimoles per liter, systolic blood pressure measured in millimeters of mercury, and body mass index (calculated as weight in kilograms divided by height in meters squared).15
All-cause, all-cardiovascular, and cause-specific mortality were estimated using the Cause of Death Ensemble Model, which produces cause-specific smoothed trends over time by age, sex, and state. Atrial fibrillation mortality was estimated with a separate natural history model described below. DisMod-MR, a Bayesian meta-regression tool developed for the GBD Study,1 was used to estimate prevalence and incidence for each cause. This software produced estimates for 6 estimation years (1990, 1995, 2000, 2005, 2010, and 2016), including data from a selected number of years before and after each estimation year when estimating for these time points. Interpolation was performed to produce a continuous series of annual results. Analysis was performed at the level of specific disease sequelae (for example, IHD due to acute coronary syndrome, chronic stable angina, chronic ischemic heart disease, and ischemic cardiomyopathy) by age, sex, year, and state. Adjustments were made to data that did not follow the selected case definition (eg, electronic claims to clinical diagnosis) by a regression model that crosswalked values in the direction of case definition–based data.17 Heart failure prevalence was estimated and then attributed proportionally to its underlying causes, including IHD, nonischemic cardiomyopathy, and myocarditis. We include a separate analysis of total heart failure prevalence, given its importance to clinical care and public health. For atrial fibrillation, both prevalence and cause-specific mortality were estimated using DisMod-MR because mortality based on vital registration data alone provides an implausibly steep increase over time believed to represent changes in ascertainment rather than the disease’s epidemiology. Prevalence was estimated across a range of severities for each condition as well as an asymptomatic state. Severity levels for each disease were estimated using data from the Medical Expenditure Panel Survey except for stroke, which was estimated from a model of Rankin scores collected within stroke registries, as described previously.17 Disability weights were developed to represent functional capacity for each severity level and multiplied by prevalence to calculate years lived with disability (YLDs), a summary measure of health among those living with a condition. Years lived with disability for sequelae are summed for their parent cause. Disability weights for the GBD Study 2016,1 including data collection and methods, have been previously described.14,20,21 Adjustments were made for comorbidity using a microsimulation process in which persons had an independent probability of having each sequela, and the probability was derived from the prevalence estimates. Years of life lost (YLLs) prematurely due to a cause was calculated by multiplying observed deaths for a specific age in the year of interest by a global age-specific reference life expectancy estimated using life table methods.
The disability-adjusted life-year (DALY) is a summary measure of health that was calculated for each age-sex-year-state-cause strata by summing the fatal (YLL) and nonfatal (YLD) components.16 For example, age-sex-state-year–specific numbers of YLLs due to IHD were added to the matching YLDs due to IHD to produce DALYs due to IHD. By dividing by population for that same strata, a DALY rate per 100 000 individuals was calculated. In the absence of health examination data from states, we predicted mean systolic blood pressure and total cholesterol levels for each state with a regression model combining covariates from the Behavioral Risk Factor Surveillance surveys and National Health and Nutrition Examination surveys. For risk factors estimated as continuous variables, we developed an ensemble distribution for each risk modeled using a family of probability density functions, a fitting method, a model selection criteria, and the method of moments.15 Population-attributable fractions of disease by cause were modeled based on estimates of exposure level, relative risk, and theoretical (eg, counterfactual) minimum risk levels using methods previously described.15 We accounted for joint effects of combinations of risk factors when sufficient evidence existed for a causal relationship. We modeled mediation pathways using individual-level data from prospective cohort studies and estimated the proportion of cardiometabolic effect from each metabolic and behavioral risk factor.
We performed a decomposition analysis of the change in DALYs from 2006 to 2016, estimating the change in CVD DALYs that would be observed after removal of the effects of population aging, population growth, and GBD Study 2016 CVD–associated risks.1 The decomposition analysis was undertaken at the all-risk level, taking into account risk mediation at the most detailed cause level. This was repeated at the most detailed risk-outcome level. The contribution of risk exposures over longer periods, eg, 2006 to 2016, or at higher cause aggregates, eg, all-CVD mortality, were calculated as the linear aggregate of the effect of individual risks at the most detailed cause level and period.
To provide a consistent comparison by socioeconomic status, a sociodemographic index (SDI) was estimated by state using equally weighted age-sex-state-year–specific geometric means of income per capita, educational attainment, and total fertility rate. The metric of SDI was used for consistency across all global locations included in the GBD 2016 Study.
The 95% uncertainty intervals (UIs) reported for each estimate used 1000 samples from the posterior distribution from the respective step in the modeling process, reported as the 2.5th and 97.5th values of the distribution. Age standardization was calculated via the direct method, applying a global age structure. Differences in estimates were considered significant if 95% UIs did not overlap.
Several states had large rises in their relative rank ordering for total CVD DALYs among states, including Arkansas, Oklahoma, Alabama, Kentucky, Missouri, Indiana, Kansas, Alaska, and Iowa (Figure 1). A notable outlier was the (nonstate) District of Columbia, which achieved the highest SDI in the United States from 1990 to 2016 while decreasing its age-standardized CVD DALY rate from the highest in the United States in 1990 (7044 DALYs per 100 000; 95% UI, 6194-7482) to the 11th highest in 2016 (3821 DALYs per 100 000 persons; 95% UI, 3424-4209).
The age-standardized rate of CVD DALYs decreased significantly in all states between 1990 and 2016, but there was wide regional variation in the amount of this decline (Table; eTable 1 in the Supplement). The largest percentage change occurred in the District of Columbia, New Hampshire, and New York. The rate of decline varied by sex, with a slower decline for women than men in all states (Figure 2A and B). The slowest decline was observed for women in Oklahoma, Arkansas, and Alabama. Total CVD burden increased for both men and women from 2010 to 2016 in Indiana, Kentucky, Michigan, Mississippi, Missouri, New Mexico, and South Dakota.
There was wide geographic variation in the age-standardized CVD burden among US states in 2016, with the greatest burden concentrated in a band of states extending from the Gulf Coast to West Virginia. The highest rate of CVD DALYs was in Mississippi (4982 age-standardized DALYs per 100 000 persons; 95% UI, 4475-5487), followed by Arkansas, Oklahoma, Louisiana, Alabama, Tennessee, Kentucky, West Virginia, South Carolina, and Georgia (Table). Notably, several states outside this region had levels of CVD DALYs nearly as high, including Indiana, Missouri, Ohio, Michigan, North Carolina, Nevada, and Texas. The lowest rate of CVD DALYs was in Minnesota (2352 age-standardized DALYs per 100 000 persons; 95% UI, 2148-2552), followed by Colorado and areas of New England and the Pacific Northwest, including Massachusetts, New Hampshire, Washington, Connecticut, Vermont, and Oregon. Ischemic heart disease was the leading cause of age-standardized CVD DALYs in all states and the District of Columbia (eFigure 1 in the Supplement). The second-leading CVD cause was ischemic stroke in all states. The proportion of DALYs due to YLD (as opposed to YLL) ranged from 10% (in Mississippi) to 18% (in Connecticut) (eFigure 2 in the Supplement).
In 2016 in the United States, CVD as a proportion of all DALYs increased with age rapidly after age 40 years, rising to account for 20% of all DALY burden by age 65 years. The largest cause of CVD in the first year of life was cardiomyopathy (Figure 3). Hemorrhagic stroke accounted for an increasingly larger amount of CVD DALYs from age 1 to 14 years and then decreased slowly with increasing age, while ischemic stroke increased rapidly as a cause of CVD after age 60 years. Ischemic heart disease was the dominant source of CVD DALYs after age 40 years. Atrial fibrillation became an increasingly common cause of CVD burden for those 65 years and older. As noted above, estimates of heart failure have been disaggregated into their underlying cause in this analysis.
Large disparities between men and women existed for total CVD burden in 2016 (Figure 2C and D). Cardiovascular disease burden was generally twice as great for men compared with women in all states for ischemic heart disease, cardiomyopathy and myocarditis, and aortic aneurysm (Figure 4). While the patterns of states with higher and lower rates of age-standardized CVD DALYs are similar, the age-standardized rate for women is lower for every state. For example, the rate of CVD DALYs in Mississippi among women (3581 age-standardized DALYs per 100 000 persons; 95% UI, 3285-4399), the highest rate experienced by women in any state, is similar to the rate of CVD DALYs among men in Florida, Virginia, and South Dakota. The largest absolute difference in CVD DALY rates was observed between Minnesota and Mississippi for both men and women; however, this gap between the lowest and highest rates of CVD DALYs was much larger for men than women (3249 vs 2115 age-standardized DALYs per 100 000 persons). As noted above, estimates of heart failure have been disaggregated to their underlying cause in this analysis.
Age-standardized heart failure prevalence was greatest in many Midwestern and Eastern states and was least across the northern Great Plains and Western states (Figure 5; eTable 2 in the Supplement). New York had the greatest age-standardized prevalence rate for heart failure in 2016 (1319 cases per 100 000 persons; 95% UI, 1277-1350), followed by Indiana, Oklahoma, Kentucky, Michigan, West Virginia, and Ohio. Heart failure was least prevalent in Minnesota (760 cases per 100 000 persons; 95% UI, 702-827), with similarly low rates in Washington, Vermont, and Iowa.
For almost all states, the greatest proportion of age-standardized CVD DALYs was attributable to dietary risk factors, followed by high systolic blood pressure, high body mass index, high total cholesterol level, high fasting plasma glucose level, tobacco smoking, and low levels of physical activity (Figure 6). Notable risks that made up smaller proportions of CVD DALYs were ambient air particulate matter, impaired kidney function, and alcohol use. As an example, eFigure 3 in the Supplement shows the relative change in rank position for magnitude of the attributable age-standardized CVD DALY rate for risk factors in Mississippi and Minnesota. While dietary risks and elevated systolic blood pressure were leading risk factors for CVD in both Mississippi and Minnesota in both 1990 and 2016, high body mass index became a greater contributor and tobacco smoking became a lesser contributor to CVD burden over time.
Figure 7 shows the relative contributions of 4 mutually exclusive drivers of the observed change in CVD DALYs from 2006 to 2016 for each state: population growth, population aging, trends in exposure to all CVD risk factors measured in the GBD Study 2016,1 and all other unmeasured factors combined. The change from 1990 to 2016 is shown in eFigure 4 in the Supplement. Most states had an increase in CVD DALYs during this time despite all states experiencing a decrease in CVD-related risk exposures. Population aging and population growth accounted for most of this increase. Notably, the residual category of unmeasured factors, which would account for health care–related treatment and any other exposures not included in the GDB Study 2016 evaluation of traditional CVD risk factors, explained increases in many states, suggesting that unmeasured risk exposures are increasing the burden of CVD in many parts of the United States.
Large disparities in total burden of CVD persist between US states despite marked improvements in CVD burden. We found that it took 25 years for states with the largest burden of CVD to achieve levels observed among the healthiest states in 1990. States with the highest burden of CVD in 1990, such as Kentucky, West Virginia, Alabama, Arkansas, Louisiana, Tennessee, and Oklahoma, are only now achieving the 1990 levels of CVD burden in Massachusetts, Connecticut, and New Jersey. Mississippi continues to lag as the state with the largest CVD burden in the United States. These findings support the idea that tremendous gains in cardiovascular health are possible even in states with lower socioeconomic levels but that relative disparities between states have changed very little.22 These relative disparities may be of particular concern for Alabama, Mississippi, Oklahoma, and Tennessee, given their recent decision to not expand their respective Medicaid systems.23
We found increases in the total burden of CVD in 12 states from 2010 to 2016 (eFigure 5 in the Supplement). Several studies have noted increasing all-cause mortality for selected subgroups or regions of the United States. Life expectancy has been decreasing among women in some counties.6 It has been suggested that increasing body mass index will result in decreasing life expectancy in the United States.24 Our finding of increasing CVD burden is concerning and suggests that long-term decline in CVD may be ending. New clinical or public health interventions delivered earlier in the life course may be required to alter this alarming trajectory.
The District of Columbia, a small urban area tracked separately, is a notable outlier that demonstrates the potential for improvements in the burden of CVD for cities. This region experienced marked improvements in socioeconomic status since 1990, as reflected by our summary measure of SDI. The District of Columbia also experienced particularly rapid declines in CVD. The causal relationship between socioeconomic status and health has been well described.25 Also, wide variation in the rate of change for cardiovascular mortality has been shown for small geographic regions, such as counties.26 Migration of healthier individuals into the District of Columbia and states or migration of sicker individuals out of these locations may have also contributed to changes in CVD burden.
An intriguing finding of our study was that socioeconomic status did not fully explain a population’s level of CVD burden or risk factors. States with lower rates of CVD burden were found across the full range of SDI. Prior research has suggested a causal association between higher altitude and lower CVD mortality, which could explain lower CVD burden in some Mountain and Southwestern states.27 Variation in health care quality between states, another possible explanation, has been less well documented than variation between specific hospitals or health care referral regions but may be substantial. Finally, some aspects of socioeconomic status may not be well accounted for by our index, such as wealth (as opposed to income per capita) or attained level of maternal education.
Research and policy have focused extensively on race and ethnicity as independent risk factors for CVD in the United States.28 The GBD 2016 Study did not stratify health by these categories, and a full discussion of this important topic is beyond the scope of this article. For example, disparities attributed to race may in fact reflect differences in access to high-quality health care or genetic factors.29 The concentration of CVD burden in states with higher proportions of individuals that identify their race as black/African American or American Indian/Alaska Native is a well-known observation.30-32 The association of race/ethnicity and risk are complicated by the observation that self-reported race/ethnicity differs from genetic background. Furthermore, reported risk associations have differed for various regions of the country, suggesting effect modification by local factors.33,34 The addition of race-specific and ethnicity-specific state-level estimates is an important goal for future iterations of the GBD Study and will allow for further exploration of these issues.
Diseases caused wholly or in part by atherosclerotic vascular disease (IHD, stroke, peripheral artery disease, or aortic aneurysm) accounted for the largest portion of CVD in all states. Most of this burden was due to IHD. As noted above, estimates of heart failure were disaggregated to their underlying cause in this analysis, including IHD; burden due only to heart failure is not reported. More than 80% of CVD burden could be attributed to known modifiable risk factors. The prevention of CVD through the reduction of these well-known risk factors remains a major public health goal for the United States.35 Clinical trials have shown that medications should target lower levels of systolic blood pressure and plasma low-density lipoprotein cholesterol for many patients.36-39 Tobacco cessation also remains a major target for health systems and local governments.40 In addition to these clinical and public health goals, our analysis shows that a large proportion of CVD can also be attributed to dietary risks, high body mass index, and low physical activity. Notably, air pollution has continued to decrease in terms of its relative contribution to CVD in the United States.41 Both rheumatic heart disease and endocarditis account for a small but persistent proportion of CVD.
Our study has limitations. All estimates have been reported as a mean value with an estimate of uncertainty. Given the combination of diverse data sources used to produce these results, the 95% uncertainty range is an important feature of our analysis that should be considered whenever interpreting a particular point estimate. Our nonfatal modeling process has improved significantly over the lifetime of the GBD Study, yet several challenges remain, including incorporation of uncertainty because of using multiple nonreference case definitions (such as cohort and claims data), quantifying the generalizability of claims data, identifying additional data on disease severity, accounting for the interdependence of comorbidities, and moving from cross-sectional estimation to a method that accounts for birth cohort effects.
Our analysis has several specific limitations. First, estimates at the level of US states represent an aggregate across a range of substate geographies, such as counties and urban vs rural areas. State-level estimates remain important given that many policy decisions continue to be made uniformly at the state level; however, further analysis is needed to examine differences between other geographic categories, such as urban and rural regions. Bias in death certification related to CVD has been demonstrated.42 Our results correct for some of this bias by adjusting for the use of nonspecific and intermediate ICD codes. Second, for this analysis, we applied a method to account for the variable use between states of nonspecific or intermediate causes on death certificates. Other biases in death certification are more difficult to correct in a state-specific manner, such as the common coding of death to ICD code I64 (stroke, unspecified type), which were reassigned to stroke subtypes using the same ratio of subtypes for each state.43 Third, our source for administrative health care data was limited to a database of mostly commercial payers, which may underestimate incidence and prevalence of disease if those with employer-based health plans are healthier than the general population. Incidence data for CVD is particularly limited in the United States. These claims data are available at the individual level for inpatient and ambulatory care and for all ages in all states across multiple years but do not include care covered solely by Medicare. Prevalence of CVD may be lower among those with private, employee-based insurance than those using Medicare or without health insurance. Administrative data were used only for conditions where it was felt to be reliable based on validation studies and expert opinion; however, some degree of misclassification bias, with both overestimation and underestimation, is likely from these sources. Expanding the GBD Study to include Medicare data is an important goal for future versions of the study. Fourth, our attribution of CVD burden to modifiable risk factors assumes a theoretical minimum level of exposure rather than the treatment goal of any set of existing intervention. For example, we rely on data from prospective cohort studies and clinical trials to set the minimum level of systolic blood pressure at 110 to 115 mm Hg. This level was selected because it is at this level that the lowest association with CVD outcomes is observed.44 We do not make any assumptions (or provide any estimates of health gains) regarding a specific level to be targeted by antihypertensive medication. Furthermore, while there is good evidence for causality for each risk and paired CVD outcome, the precise causes in any given population cannot be known with certainty. Fifth, estimates of prevalent heart failure are based on administrative claims data and carry the assumption that all cases presented to hospitals. Subclinical heart failure is therefore not included in these estimates. Sixth, our study is, by design, cross-sectional and therefore descriptive and hypothesis-generating rather than the basis for causal claims. Finally, this analysis does not report CVD due to 2 cardiovascular conditions, Chagas cardiomyopathy and congenital heart disease. Several hundred thousand cases of Chagas cardiomyopathy may be present in the United States.45 Estimates for disease burden due to congenital heart disease in the United States have been reported previously.46
Cardiovascular disease is a major cause of lost health in the United States but varies widely in level among states. Most CVD burden in the United States is from atherosclerotic vascular disease, and 80% can be attributed to known causal risk factors. We found that CVD burden has improved for all states, but the rate of decline varies widely and is strongly associated with an index of socioeconomic level. For 12 states, CVD burden has increased since 2010. These estimates can provide a benchmark for states working to focus on key risk factors, improve health care quality, and lower health care costs.
Accepted for Publication: February 22, 2018.
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2018 Global Burden of Cardiovascular Diseases Collaboration. JAMA Cardiology.
Corresponding Author: Gregory A. Roth, MD, MPH, Division of Cardiology, Department of Medicine, Institute for Health Metrics and Evaluation, University of Washington, 2301 Fifth Ave, Ste 600, Seattle, WA 98121 (firstname.lastname@example.org).
Published Online: April 11, 2018. doi:10.1001/jamacardio.2018.0385
Global Burden of Cardiovascular Diseases Collaboration Authors: The following investigators take authorship responsibility for the study results: Gregory A. Roth, MD, MPH; Catherine O. Johnson, PhD; Kalkidan Hassen Abate, MS; Foad Abd-Allah, MD; Muktar Ahmed, BS, MPH, BPharm, MBA; Khurshid Alam, MS, MPH, PhD; Tahiya Alam, MPH; Nelson Alvis-Guzman, PhD; Hossein Ansari, BS, MS, PhD; Johan Ärnlöv, PhD, MD; Tesfay Mehari Atey, MS; Ashish Awasthi, MSc, PhD; Tadesse Awoke, MS, MPH; Aleksandra Barac, MPH, PhD, DSc, MD; Till Bärnighausen, MS, DSc, MD; Neeraj Bedi, MD, MPH, MBA; Derrick Bennett, PhD; Isabela Bensenor, MD, PhD; Sibhatu Biadgilign, BS, MSc, MPH; Carlos Castañeda-Orjuela, MSc, MD; Ferrán Catalá-López, PhD, MPH, MHEcon, PharmD; Kairat Davletov, MPH, PhD, MD; Samath Dharmaratne, MBBS, MSc, MD; Eric L. Ding, ScD; Manisha Dubey, MS, MPhil; Emerito Jose Aquino Faraon, BSPH, MD, MBA; Talha Farid, MD; Maryam S. Farvid, PhD; Valery Feigin, MD, PhD; João Fernandes, BS, MS, PhD; Joseph Frostad, MPH; Alemseged Gebru, MPH; Johanna M. Geleijnse, PhD; Philimon Nyakauru Gona, MA, MS, MPH, PhD; Max Griswold, MA; Gessessew Bugssa Hailu, MSc; Graeme J. Hankey, MD, FRACP, FRCP, FRCPE, FAHA, MBBS; Hamid Yimam Hassen, MPH; Rasmus Havmoeller, MD, PhD; Simon Hay, DSc, FMedSci; Susan R. Heckbert, PhD, MD; Caleb Mackay Salpeter Irvine, BA; Spencer Lewis James, BS, MPH, MD; Dube Jara, MPH; Amir Kasaeian, PhD; Abdur Rahman Khan, MD; Sahil Khera, MD, MPH; Abdullah T. Khoja, MPH, MD, MBA, DrPHc, ABFM; Jagdish Khubchandani, MPH, PhD, MD; Daniel Kim, MD, DrPH; Dhaval Kolte, PhD, MD; Dharmesh Lal, MD, DNB, DHA, MBBS; Anders Larsson, MD, PhD; Shai Linn, MPH, PhD, MD; Paulo A. Lotufo, MD, DrPH; Hassan Magdy Abd El Razek, MBBCH, BA; Mohsen Mazidi, BS, MS, PhD; Toni Meier, PhD; Walter Mendoza, MD; George A. Mensah, MD; Atte Meretoja, MD, PhD; Haftay Berhane Mezgebe, MS; Erkin Mirrakhimov, PhD, MD; Shafiu Mohammed, BS, MS, MPH, PhD, DSc; Andrew Edward Moran, MD, MPH; Grant Nguyen, MPH; Minh Nguyen, BS; Kanyin Liane Ong, PhD; Mayowa Owolabi, MD, MBBS, MSc; Martin Pletcher, BS; Farshad Pourmalek, MPH, PhD, MD; Caroline A. Purcell, BS; Mostafa Qorbani, MS, PhD; Mahfuzar Rahman, PhD, MD; Rajesh Kumar Rai, BA, MA, MPH, MPhil; Usha Ram, PhD; Marissa Bettay Reitsma, BS; Andre M. N. Renzaho, MPH, PhD, BSc Nutri; Maria Jesus Rios-Blancas, MPH; Saeid Safiri, BS, MS, PhD; Joshua A. Salomon, PhD; Benn Sartorius, BS, MS, PhD; Sadaf Ghajarieh Sepanlou, MPH, PhD, MD; Masood Ali Shaikh, MBBS, MPH, MGIS, MS; Diego Silva, PhD; Saverio Stranges, PhD, MD; Rafael Tabarés-Seisdedos, PhD, MD; Niguse Tadele Atnafu, BS, MS; J. S. Thakur, MD; Roman Topor-Madry, PhD, MD; Thomas Truelsen, PhD, DMSc, MD; E. Murat Tuzcu, MD; Stefanos Tyrovolas, PhD; Kingsley Nnanna Ukwaja, MD; Tommi Vasankari, MD, PhD; Vasiliy Vlassov, MD; Stein Emil Vollset, MD, DrPH; Tolassa Wakayo, BS, MS; Robert Weintraub, MB, BS (Hons); Charles Wolfe, MB BS, MRCOG, MD, MFPH, FFPH, FRCOG; Abdulhalik Workicho, MPH, BSc; Gelin Xu, MD, PhD; Simon Yadgir, BA; Yuichiro Yano, PhD, MD; Paul Yip, PhD; Naohiro Yonemoto, MPH; Mustafa Younis, DrPH, MA, MBA; Chuanhua Yu, PhD; Zoubida Zaidi, MPH, PhD, DSc, MD; Maysaa El Sayed Zaki, MD, PhD; Ben Zipkin, BA; Ashkan Afshin, MS, MPH, ScD, MD; Emmanuela Gakidou, PhD; Stephen S. Lim, BA, BSc (Hons), PhD; Ali H. Mokdad, PhD; Mohsen Naghavi, MD, PhD; Theo Vos, PhD, MSc; Christopher J. L. Murray, MD, DPhil.
Affiliations of Global Burden of Cardiovascular Diseases Collaboration Authors: Institute for Health Metrics and Evaluation, University of Washington, Seattle (Roth, Johnson, T. Alam, Frostad, Griswold, Hay, Heckbert, Irvine, James, G. Nguyen, M. Nguyen, Ong, Pletcher, Purcell, Reitsma, Yadgir, Zipkin, Afshin, Gakidou, Lim, Mokdad, Naghavi, Vos, Murray); Division of Cardiology, Department of Medicine, University of Washington, Seattle (Roth); Jimma University, Jimma, Ethiopia (Abate, Ahmed, Wakayo, Workicho); Cairo University, Giza, Egypt (Abd-Allah); The University of Western Australia, Perth, Western Australia, Australia (K. Alam); Universidad de Cartagena, Cartagena, Bolívar, Columbia (Alvis-Guzman); Zahedan University of Medical Sciences, Zahedan, Iran (Ansari); Karolinska Institutet, Stockholm, Sweden (Ärnlöv, Havmoeller); Mekelle University, Mekelle, Ethiopia (Atey, Gebru, Hailu, Mezgebe); Indian Institute of Public Health Gandhinagar, Public Health Foundation of India, Gandhinagar, Gujarat, India (Awasthi); University of Gondar, Gondar, Ethiopia (Awoke); University of Belgrade, Belgrade, Serbia (Barac); Harvard University, Cambridge, Massachusetts (Bärnighausen, Salomon); Jazan University, Jizan, Saudi Arabia (Bedi); University of Oxford, Oxford, England (Bennett); University of São Paulo, São Paulo, São Paulo, Brazil (Bensenor); Independent Public Health Consultant, Addis Abbaba, Ethiopia (Biadgilign); Observatorio Nacional de Salud, Instituto Nacional de Salud, Bogotá, DC, Colombia (Castañeda-Orjuela); INCLIVA Health Research Institute, Centro de Investigación Biomédica en Red Salud Mental, University of Valencia, Valencia, Spain (Catalá-López, Tabarés-Seisdedos); Asfendiyarov Kazakh National Medical University, Almaty, Kazakhstan (Davletov); University of Peradeniya, Peradeniya, Sri Lanka (Dharmaratne); Harvard T. H. Chan School of Public Health, Boston, Massachusetts (Ding, Farvid); International Institute for Population Sciences, Mumbai, India (Dubey, Ram); University of the Philippines Manila, Manila, Philippines (Faraon); University of Louisville, Louisville, Kentucky (Farid, Khan); Auckland University of Technology, Auckland, New Zealand (Feigin); Catholic University of Portugal, Lisbon, Portugal (Fernandes); Wageningen University, Wageningen, the Netherlands (Geleijnse); University of Massachusetts Boston, Boston (Gona); School of Medicine and Pharmacology, University of Western Australia, Perth, Western Australia, Australia (Hankey); Mizan Tepi University, Tepi, Ethiopia (Hassen); Debre Markos University, Debre Markos, Ethiopia (Jara); Hematology, Oncology, and Stem Cell Transplantation Research Center, Tehran University of Medical Sciences, Tehran, Iran (Kasaeian); New York Medical College, Valhalla (Khera); Al Imam Mohammad Ibn Saud Islamic University, Riyadh, Saudi Arabia (Khoja); Ball State University, Muncie, Indiana (Khubchandani); Northeastern University, Boston, Massachusetts (Kim); Brown University, Providence, Rhode Island (Kolte); Public Health Foundation of India, New Delhi, India (Lal); Department of Medical Sciences, Uppsala University, Uppsala, Sweden (Larsson); University of Haifa, Haifa, Israel (Linn); Clinical Research Center, University Hospital, University of São Paulo, São Paulo, São Paulo, Brazil (Lotufo); Faculty of Medicine, Mansoura University, Mansoura, Egypt (Magdy Abd El Razek, Zaki); State Key Laboratory of Molecular Developmental Biology, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Chaoyang, Beijing (Mazidi); Martin Luther University of Halle-Wittenberg, Halle, Germany (Meier); United Nations Population Fund, New York, New York (Mendoza); National Institutes of Health, Bethesda, Maryland (Mensah); University of Melbourne, Melbourne, Victoria, Australia (Meretoja); Kyrgyz State Medical Academy, Bishkek, Kyrgyzstan (Mirrakhimov); Ahmadu Bello University, Zaria, Kaduna State, Nigeria (Mohammed); Columbia University, New York, New York (Moran); Department of Medicine, University of Ibadan, Ibadan, Oyo State, Nigeria (Owolabi); University of British Columbia, Vancouver, British Columbia, Canada (Pourmalek); Noncommunicable Diseases Research Center, Alborz University of Medical Sciences, Hassan Abad, Karaj, Iran (Qorbani); BRAC, Dhaka, Bangladesh (Rahman); Society for Health and Demographic Surveillance, West Bengal, India (Rai); Western Sydney University, Penrith, New South Wales, Australia (Renzaho); National Institute of Public Health, Wako, Japan (Rios-Blancas); Maragheh University of Medical Sciences, East Azerbaijan Province, Iran (Safiri); University of KwaZulu-Natal, Durban, South Africa (Sartorius); Tehran University of Medical Sciences, Tehran, Iran (Sepanlou); Independent Consultant, Karachi, Pakistan (Shaikh); Federal University of Santa Catarina, Florianópolis, Santa Catarina, Brazil (Silva); Department of Epidemiology and Biostatistics, Schulich School of Medicine & Dentistry, Western University, London, Ontario, Canada (Stranges); Addis Ababa University, Addis Ababa, Ethiopia (Tadele Atnafu); Postgraduate Institute of Medical Education and Research, Chandigarh, India (Thakur); Jagiellonian University Medical College, Krakow, Poland (Topor-Madry); University of Copenhagen, Copenhagen, Denmark (Truelsen); Cleveland Clinic, Cleveland, Ohio (Tuzcu); Hospital Sant Joan de Déu Barcelona, Sant Joan de Déu Research Foundation, Centro de Investigación Biomédica en Red Salud Mental, Universitat de Barcelona, Barcelona, Spain (Tyrovolas); Department of Internal Medicine, Federal Teaching Hospital, Abakaliki, Ebonyi State, Nigeria (Ukwaja); UKK Institute for Health Promotion Research, Tampere, Finland (Vasankari); National Research University Higher School of Economics, Moscow, Russia (Vlassov); Norwegian Institute of Public Health, Oslo, Norway (Vollset); Royal Children’s Hospital, Melbourne, Victoria, Australia (Weintraub); King’s College London, London, England (Wolfe); Nanjing University School of Medicine, Nanjing, China (Xu); The University of Mississippi Medical Center, Jackson (Yano); University of Hong Kong, Pokfulam, Hong Kong (Yip); Kyoto University, Kyoto, Japan (Yonemoto); Jackson State University, Jackson, Mississippi (Younis); Wuhan University, Wuhan, China (Yu); University Hospital of Setif, Setif, Algeria (Zaidi).
Author Contributions: Drs Roth and Murray had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Roth, Johnson, Abate, Abd-Allah, Dharmaratne, Hay, Moran, Rios-Blancas, Yip, Younis, Afshin, Mokdad, Naghavi, Vos, Murray.
Acquisition, analysis, or interpretation of data: Roth, Johnson, Ahmed, K Alam, T Alam, Alvis-Guzman, Ärnlöv, Mehari Atey, Awasthi, Awoke, Barac, Bedi, Bennett, Bensenor, Biadgilign, Castañeda-Orjuela, Catalá-López, Davletov, Dharmaratne, Ding, Dubey, Faraon, Farid, Farvid, Feigin, Fernandes, Frostad, Gebru, Geleijnse, Gona, Griswold, Hailu, Hankey, Hassen, Havmoeller, Hay, Heckbert, Irvine, James, Jara, Kasaeian, Khan, Khera, Khoja, Khubchandani, Kim, Kolte, Lal, Larsson, Linn, Lotufo, Magdy Abd El Razek, Mazidi, Meier, Mendoza, Mensah, Meretoja, Mezgebe, Mirrakhimov, Mohammed, G Nguyen, M Nguyen, Ong, Owolabi, Pourmalek, Purcell, Qorbani, Rahman, Rai, Ram, Reitsma, Renzaho, Safiri, Salomon, Sartorius, Sepanlou, Shaikh, Silva, Stranges, Tabarés-Seisdedos, Tadele Atnafu, Thakur, Topor-Madry, Truelsen, Tuzcu, Tyrovolas, Ukwaja, Vasankari, Vlassov, Vollset, Wakayo, Weintraub, Wolfe, Workicho, Xu, Yadgir, Yano, Yip, Yonemoto, Yu, Zaidi, Zaki, Zipkin, Afshin, Gakidou, Mokdad, Naghavi, Vos, Murray.
Drafting of the manuscript: Roth, T Alam, Barac, Irvine, Jara, Lotufo, Magdy Abd El Razek, Mazidi, Silva, Zaki.
Critical revision of the manuscript for important intellectual content: Roth, Johnson, Abate, Abd-Allah, Ahmed, K Alam, Alvis-Guzman, Ärnlöv, Mehari Atey. Awasthi, Awoke, Barac, Bedi, Bennett, Bensenor, Biadgilign, Castañeda-Orjuela, Catalá-López, Davletov, Dharmaratne, Ding, Dubey, Faraon, Farid, Farvid, Feigin, Fernandes, Frostad, Gebru, Geleijnse, Gona, Griswold, Hailu, Hankey, Hassen, Havmoeller, Hay, Heckbert, James, Jara, Kasaeian, Khan, Khera, Khoja, Khubchandani, Kim, Kolte, Lal, Larsson, Linn, Magdy Abd El Razek, Mazidi, Meier, Mendoza, Mensah, Meretoja, Mezgebe, Mirrakhimov, Mohammed, Moran, G Nguyen, M Nguyen, Ong, Owolabi, Pourmalek, Purcell, Qorbani, Rahman, Rai, Ram, Reitsma, Renzaho, Rios-Blancas, Safiri, Salomon, Sartorius, Sepanlou, Shaikh, Silva, Stranges, Tabarés-Seisdedos, Tadele Atnafu, Thakur, Topor-Madry, Truelsen, Tuzcu, Tyrovolas, Ukwaja, Vasankari, Vlassov, Vollset, Wakayo, Weintraub, Wolfe, Workicho, Xu, Yadgir, Yano, Yip, Yonemoto, Younis, Yu, Zaidi, Zaki, Zipkin, Afshin, Gakidou, Mokdad, Naghavi, Vos, Murray.
Statistical analysis: Roth, Johnson, Abate, K Alam, Awoke, Barac, Biadgilign, Fernandes, Frostad, Gebru, Griswold, Hailu, Hassen, Hay, Irvine, Jara, Kasaeian, Khubchandani, Larsson, Linn, Lotufo, Mazidi, Meier, Mohammed, G Nguyen, M Nguyen, Purcell, Qorbani, Reitsma, Salomon, Shaikh, Silva, Ukwaja, Yadgir, Younis, Yu, Zaidi, Zipkin, Afshin, Gakidou, Mokdad, Naghavi.
Obtained funding: Mokdad, Murray.
Administrative, technical, or material support: T Alam, Awasthi, Bensenor, Davletov, Dubey, Fernandes, Hay, James, Khoja, Magdy Abd El Razek, Mazidi, Mirrakhimov, Mohammed, Tabarés-Seisdedos, Tyrovolas, Ukwaja, Vasankari, Zaki, Mokdad.
Study supervision: Roth, Gakidou, Mokdad, Vos, Murray.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Hankey has received personal fees from Bayer and Medscape outside the submitted work. Dr Lotufo has received personal fees from AbbVie Brazil and Amgen Brazil as well as grants from Fundação Vale, Brazil. Dr Mendoza currently works for the Peru Country Office of the United Nations Population Fund, which does not necessarily endorse this study. No other disclosures were reported.
Funding/Support: The Institute for Health Metrics and Evaluation received funding from the Bill and Melinda Gates Foundation.
Role of the Funder/Sponsor: The funder 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.
Additional Contributions: We express our appreciation to Paul Briant, BS, Adrienne Chew, ND, and Kris Krohn, MPH (Institute for Health Metrics and Evaluation, University of Washington, Seattle), for their assistance in manuscript editing and production. They are salaried employees of the Institute for Health Metrics and Evaluation but otherwise received no compensation for their contributions.