Dotted lines: a leading cause has decreased in rank between 1990 and 2016; solid lines, a cause has maintained or ascended to a higher ranking. Causes in white boxes were not among the top 25 in either 1990 or in 2016. COPD, indicates chronic obstructive pulmonary disease.
aIncludes tracheal, bronchus, and lung cancer.
bIndicates HIV/AIDS resulting in other diseases.
cIndicates chronic kidney disease (CKD) due to diabetes.
Negative values (where bars extend left of zero) indicate a protective effect.
States are listed in descending order according to probability of death in 2016. Data for Washington, DC, were not included in this analysis.
Boxes are colored green if significantly less than the US rate (P <.05), red if significantly more than the US rate (P <.05), and yellow if not significantly different from the US rate (P ≥.05).
Ratio details: Alabama’s, stroke ratio (eg, 1.0 [observed and expected rates were similar]; diabetes [2.85 × above expected]). See Appendix Table 2 in Supplement 2 for explanation of terms.
See Figure 7 caption for details. See Appendix Table 2 in Supplement 2 for explanation of terms.
Section 1. GBD Overview
Section 2. GBD 2016 causes of death database
Section 3. Causes of death modeling methods
Section 4. Central computation
Section 5. Nonfatal outcome estimation
Section 6. Risk factor estimation
Section 7. Methods on the attribution of deaths to causes
Section 8. Modeling specific methods on causes of death estimation
Section 9. Modeling specific methods on nonfatal estimation
Section 10. Modeling specific methods on risk factor estimation
Appendix Figure 1. Analytical flowchart for the development of the GBD 2016 COD database
Appendix Figure 2. GBD 2016 Causes of Death estimation flowchart by modeling group
Appendix Figure 3. GBD 2016 DisMod MR 2.1 analytical cascade
Appendix Figure 4. Causal Web of the causes of health outcomes with the categories of causes
Appendix Figure 5. Analytical flowchart of the comparative risk assessment for the estimation of PAFs
Appendix Figure 6. ST-GPR flowchart
Appendix Table 1. GBD 2016 location hierarchy with levels
Appendix Table 2. Causes included in the GBD 2016 Global Burden of Disease study
Appendix Table 3. GBD 2016 CodCorrect cause hierarchy with levels
Appendix Table 4. List of ICD codes mapped to the GBD cause list for causes of death
Appendix Table 5. GBD 2016 cause and sequela hierarchy
Appendix Table 6. GBD 2016 sequelae, health states, health state lay descriptions, and disability weights
Appendix Table 7. GBD 2016 methods of estimating YLDs for 35 residual categories
Appendix Table 8. GBD 2016 risk factor hierarchy with levels, modeling strategies, and the main type of data sources used to estimate exposure levels
Appendix Table 9. Types of Comparative Risk Assessments (CRA) based on the time perspective and the nature of the counterfactual level or distribution of exposure
Appendix Table 10. Percent change in age-standardised summary exposure values (SEVs) for leading ten risk factors for the United States 1990-2016, both sexes
Appendix Table 11. Socio-Demographic Index values for all estimated GBD 2016 locations, 1980–2016
Appendix Table 12. The probability of death between 0 and 20 years in the United States, 50 states, and DC from 1990 to 2016 for Level 2 GBD causes
Appendix Table 13. The probability of death between 20 and 55 years in the United States, 50 states, and DC from 1990 to 2016 for Level 2 GBD causes
Appendix Table 14. The probability of death between 55 and 90 years in the United States, 50 states, and DC from 1990 to 2016 for Level 2 GBD causes
Appendix Table 15. 35q20 comparison with US, 50 states, and DC and GBD countries
Figures on leading most detailed causes of DALYs for the US states and DC from 1990 to 2016: states and DC are presented in alphabetical order
Customize your JAMA Network experience by selecting one or more topics from the list below.
The US Burden of Disease Collaborators. The State of US Health, 1990-2016: Burden of Diseases, Injuries, and Risk Factors Among US States. JAMA. 2018;319(14):1444–1472. doi:https://doi.org/10.1001/jama.2018.0158
How have the levels and trends of burden of diseases, injuries, and risk factors in the United States changed from 1990 to 2016 by state?
This study, involving examination of 333 causes and 84 risk factors, demonstrated that health in the United States improved from 1990 to 2016, although the drivers of mortality and morbidity have changed in some states, with specific risk factors such as drug use disorders, high body mass index (BMI), and alcohol use disorders being associated with adverse outcomes. In 5 states, the probability of death between ages 20 and 55 years has increased more than 10% between 1990 and 2016.
Differences in health outcomes and drivers of morbidity and mortality at the state level indicate the need for greater investment in preventive and medical care across the life course. The intersection of risk, mortality, and morbidity in particular geographic areas needs to be further explored at the state level.
Several studies have measured health outcomes in the United States, but none have provided a comprehensive assessment of patterns of health by state.
To use the results of the Global Burden of Disease Study (GBD) to report trends in the burden of diseases, injuries, and risk factors at the state level from 1990 to 2016.
Design and Setting
A systematic analysis of published studies and available data sources estimates the burden of disease by age, sex, geography, and year.
Main Outcomes and Measures
Prevalence, incidence, mortality, life expectancy, healthy life expectancy (HALE), years of life lost (YLLs) due to premature mortality, years lived with disability (YLDs), and disability-adjusted life-years (DALYs) for 333 causes and 84 risk factors with 95% uncertainty intervals (UIs) were computed.
Between 1990 and 2016, overall death rates in the United States declined from 745.2 (95% UI, 740.6 to 749.8) per 100 000 persons to 578.0 (95% UI, 569.4 to 587.1) per 100 000 persons. The probability of death among adults aged 20 to 55 years declined in 31 states and Washington, DC from 1990 to 2016. In 2016, Hawaii had the highest life expectancy at birth (81.3 years) and Mississippi had the lowest (74.7 years), a 6.6-year difference. Minnesota had the highest HALE at birth (70.3 years), and West Virginia had the lowest (63.8 years), a 6.5-year difference. The leading causes of DALYs in the United States for 1990 and 2016 were ischemic heart disease and lung cancer, while the third leading cause in 1990 was low back pain, and the third leading cause in 2016 was chronic obstructive pulmonary disease. Opioid use disorders moved from the 11th leading cause of DALYs in 1990 to the 7th leading cause in 2016, representing a 74.5% (95% UI, 42.8% to 93.9%) change. In 2016, each of the following 6 risks individually accounted for more than 5% of risk-attributable DALYs: tobacco consumption, high body mass index (BMI), poor diet, alcohol and drug use, high fasting plasma glucose, and high blood pressure. Across all US states, the top risk factors in terms of attributable DALYs were due to 1 of the 3 following causes: tobacco consumption (32 states), high BMI (10 states), or alcohol and drug use (8 states).
Conclusions and Relevance
There are wide differences in the burden of disease at the state level. Specific diseases and risk factors, such as drug use disorders, high BMI, poor diet, high fasting plasma glucose level, and alcohol use disorders are increasing and warrant increased attention. These data can be used to inform national health priorities for research, clinical care, and policy.
Quiz Ref IDPrevious studies have reported on health disparities in US states and counties.1,2 These studies showed that health disparities have increased with time. Recent attention has focused on increased mortality in some age groups and a decline in life expectancy.3 In addition, the performance of the US health care system does not match its level of spending on health and lags behind countries with similar financial resources.4 For example, in 2014, US life expectancy ranked 43rd in the world, although the United States spent the most ($3.0 trillion) on health care, exceeding the median amount spent by Organisation for Economic Co-operation and Development countries by 35%.5
Several studies have shown large variations in risk factors by state and county, and these variations have contributed to differences in health outcomes.6-9 In the Global Burden of Disease Study 2010 (GBD 2010) US Burden of Disease report, the following risk factors were reported as the main causes associated with US morbidity and mortality (percent contributed to total disability-adjusted life-years [DALYs] in 2010): poor diet (14%), smoking (11%), high blood pressure (8%), and obesity (11%).10-12 None of the previous studies of US health have been as comprehensive as the GBD study.5,13-18 The GBD systematically accounts for differences in data sources and biases and analyzes levels and trends for causes and risk factors within the same computational framework, which maximizes comparability across states, years, and different age groups by sex. GBD is now conducted on an annual cycle, with GBD 2016 providing updated estimates of mortality, morbidity, and risk factors in 195 locations, including the United States, from 1990 to 2016.
The findings of GBD 2016 indicate that while the United States overall is experiencing improvements in health outcomes, the patterns of health burden at the state level vary across geography. Routinely monitoring the trend of burden of disease at the state level is essential given the vital role of states in many aspects of health and social policy19—from the Medicaid program to regulation of private insurers20 and considering that individual states also experience different economic circumstances. The current study uses GBD 2016 to report the change in burden of disease, including injuries and risk factors at the state level, from 1990 to 2016.
The GBD study is estimated annually and each round of results is internally consistent (cause-specific mortality estimates are summed to match all-cause mortality estimates) and collectively exhaustive (residual categories [“other”] are captured to enable quantifying total burden) (Sections 1-5 in Supplement 1). The numbers reported in the previous round of GBD are not identical to those of the current round (GBD 2016) for 2 main reasons. First, since the GBD 2010 Special Communication regarding US risk factors, there has been further refinement of the “garbage coding” (ie, ill-defined causes of death) redistribution methods (Supplement 1). Second, the new analysis at the state level changes some of the estimation slightly when aggregated to the national level. GBD 2016 provides a new time series.
Quiz Ref IDThe GBD 2016 methodology has been published previously.5,13-18 GBD uses several metrics to report results for health loss related to specific diseases, injuries, and risk factors: deaths and death rates, years of life lost (YLLs) due to premature mortality, prevalence and prevalence rates for sequelae, years lived with disability (YLDs), and DALYs (Box; Sections 2, 3, and 5 in Supplement 1 and Appendix Table 2 in Supplement 2). GBD provides a comprehensive assessment of all-cause mortality and estimates for death due to 264 causes in 195 countries and territories from 1990 to 2016, as well as 333 causes of DALYs (Appendix Table 2 in Supplement 2). GBD 2016 has 4 levels of causes that are mutually exclusive (Appendix Table 3 in Supplement 2). Level 1 has 3 causes: communicable, maternal, neonatal, and nutritional disorders; noncommunicable diseases; and injuries. Level 2 has 21 causes. Levels 3 and 4 consist of more disaggregated causes. GBD 2016 documented each step of the estimation processes, as well as data sources, in accordance with the Guidelines for Accurate and Transparent Health Estimates Reporting.21 A more detailed methodology is available in the appendix to this article (Sections 8 and 9 in Supplement 1).
Disability-adjusted life-years: a summary metric of population health. DALYs represent a health gap and, as such, measure the state of a population’s health compared to a normative goal. The goal is for individuals to live the standard life expectancy in full health. DALYs are the sum of 2 components: years of life lost (YLLs) and years lived with disability (YLDs).
Healthy life expectancy: the number of years that a person at a given age can expect to live in good health, taking into account mortality and disability.
Summary Exposure Value: the relative risk–weighted prevalence of exposure (developed for Global Burden of Diseases Study 2015).
Years lived with disability: computed as the prevalence of different disease sequelae and injury sequelae multiplied by disability weights for that sequela. Disability weights are selected on the basis of surveys of the general population about the health loss associated with the health state related to the disease sequela.
Years of life lost due to premature mortality: computed by multiplying the number of deaths at each age by a standard life expectancy at that age. The standard selected represents the normative goal for survival and has been computed based on the lowest recorded death rates across countries in 2010.
To estimate the US burden of disease prevalence, computation for each sequela began with a systematic analysis of published studies and available data sources providing information on prevalence, incidence, remission, and excess mortality, such as the National Health and Nutrition Examination Surveys,22 state inpatient databases,23 the National Ambulatory Medical Care Survey,24 National Hospital Ambulatory Medical Care Survey,25 Medical Expenditure Panel Survey,26 National Comorbidity Survey,27 National Epidemiological Survey on Alcohol and Related Conditions,28 National Survey on Drug Use and Health,29 US Department of Agriculture Continuing Survey of Food Intakes,30 Marketscan,31 National Health Interview Survey,32 Behavioral Risk Factor Surveillance System,33 and the Centers for Disease Control and Prevention Disease Surveillance Reports.34
Hospital inpatient data were extracted and used for this analysis. Moreover, outpatient encounter data were available for the United States through aggregate data derived from a database of claims information for US private and public insurance schemes for the years 2000, 2010, and 2012. GBD methodology applied several correction factors to account for bias in health service encounter data from these claims that were available as aggregated by International Classification of Diseases (ICD) code and by primary diagnosis only. First, for chronic disorders, the study estimated the ratio between prevalence from primary diagnoses and prevalence from all diagnoses associated with a claim. Second, the claims data were used to generate the mean number of outpatient visits per disorder. Similarly, the study generated per-person discharge rates from hospital inpatient data in the United States.
All-cause mortality was estimated by age, sex, geography, and year using 6 modeling approaches to assess cause-specific mortality; the Cause of Death Ensemble Model was used to generate estimates for the vast majority of causes. This analysis used deidentified death records from the National Center for Health Statistics (NCHS)35 and population counts from the US Census Bureau,36 NCHS, and the Human Mortality Database.37 Deaths and population were tabulated by county, age group, sex, year, and (in the case of death data) cause. The cause list developed for the GBD13 is arranged hierarchically in 4 levels. Within each level, the cause list is designed such that all deaths are assigned exactly 1 cause. As part of the GBD study, a map has been developed that allows ICD-9 and ICD-10 codes to be translated to GBD causes.
Previous studies have documented the existence of insufficiently specific or implausible causes of death used in death registration data that may lead to misleading geographic and temporal patterns.38 Algorithms developed for the GBD were used to reallocate deaths assigned one of these garbage codes to plausible alternatives.39 First, plausible target causes were assigned to each garbage code or group of garbage codes. Second, deaths were reassigned to specified target codes according to proportions derived in 1 of 4 ways: (1) published literature or expert opinion; (2) regression models; (3) according to the proportions initially observed among targets; and (4) for HIV/AIDS specifically, by comparison to years before HIV/AIDS became widespread. More detail on each of these methods is provided in Section 2 of Supplement 1.
Based on standard GBD methods, YLLs were computed by multiplying the number of deaths from each cause in each age group by the reference life expectancy at the mean of age of death among those who died in the age group. The YLLs computation is based on the precedent set by GBD and uses the same life table standard for calculating YLLs in all locations and years (essential for comparing estimates of YLLs across locations and years). The standard is meant to represent the mortality experience of a population with minimal excess mortality using the lowest observed age-specific mortality rates in 2016 among all countries with a population greater than 5 million. This standard does not vary with time because for most populations, the number of YLLs (once normalized for population size) is larger in earlier years than in later years due to improving survival rather than an artifact of the standard used.
In this study, incidence and prevalence of diseases by age, sex, cause, year, and geography were estimated using a wide range of updated and standardized analytical procedures. GBD used DisMod-MR, a Bayesian meta-regression tool, to determine prevalence and incidence by cause and sequelae.40
Data sources used for quantifying nonfatal outcomes are available online in the GBD results tool41 and in Section 3 of Supplement 1. Prevalence of each sequela was multiplied by the disability weight for the corresponding health state to calculate YLDs for the sequela. The sum of all YLDs for relevant sequelae equated to overall YLDs for each disease. Details on disability weights for GBD 2016, including data collection and disability weight construction, are described elsewhere.14
Following GBD 2016 methods, national and state-level DALYs were computed by summing YLLs and YLDs for each cause, age, and sex in 1990, 1995, 2000, 2005, 2010, and 2016 (Section 4 in Supplement 1). DALYs were computed for 333 causes, with 95% uncertainty intervals (UIs) capturing the uncertainty for both YLL and YLD rates. HALE was calculated using the Sullivan method and generated 95% UIs that indicated uncertainty for age-specific death rates and YLDs per capita for each geography, age group, sex, and year. HALE was calculated for the United States and for each individual state using multiple-decrement life tables and estimated YLDs per capita; additional details on HALE methodology are provided in Section 4 in Supplement 1.
GBD 2016 used the comparative risk assessment framework to estimate attributable deaths, DALYs, and trends in exposure by age group, sex, year, and geography for risks from 1990 to 2016. GBD has 84 behavioral, environmental and occupational, and metabolic risks or clusters of risks (Section 5 in Supplement 1). Risk-outcome pairs were included in the GBD 2016 study if they met World Cancer Research Fund criteria for convincing or probable evidence. Relative risk (RR) estimates were extracted from published and unpublished randomized clinical trials, cohorts, and pooled cohorts. Risk exposures were estimated based on published studies, household surveys, US Census data, satellite data, and other sources. Two modeling approaches, a Bayesian meta-regression model and a spatiotemporal Gaussian process regression model, were developed for the GBD study and used to pool data from different sources, adjust for bias in the data, and incorporate potential covariates. GBD used the counterfactual scenario of theoretical minimum risk exposure level (ie, the level for a given risk exposure that could minimize risk at the population level) to attribute burden. A summary exposure value was developed for GBD 2015 as the RR-weighted prevalence of exposure (range, 0 [no excess risk exists in a population] to 1 [population is at the highest risk]).16
Where Pri is prevalence of category i exposure; RRi is the RR of the category i; and RRmax is the maximum RR observed (between categories). This quantity is estimated for each age group, sex, geography, and year. In the case of dichotomous exposure, summary exposure value is equal to prevalence. For continuous risks, summary exposure value is defined as follows:
Where (χ) is the density of exposure at level χ of exposure; (χ) is the RR of the level χ; and RRmax is the highest RR that is supported by data and reflects a level in which more than 1% of the population is exposed globally. In this study, summary exposure value is reported on a scale from 0% to 100% to emphasize that it is risk-weighted prevalence.
To calculate risk-attributable fractions of disease burden by cause, the effects of risk exposure levels were modeled, RRs associated with risk exposure and specific health outcomes were documented, and counterfactual levels of risk exposure on estimates of national and state-level deaths, YLLs, YLDs, and DALYs were computed. Detailed descriptions of the GBD 2016 methods for risk factor assessment and attribution are published elsewhere (Section 5 in Supplement 1).5,13-18
The probability of death was calculated for 3 summary age intervals and the cause-specific contributions to each of these summary indicators for ages 0 to 20, 20 to 55, and 55 to 90 years. These age groups were chosen to reflect variations in trends and burden for adolescents, young adults, and older people. For each probability of death, the multiple decrement life-table method was used to compute the probability of death from each cause and the overall contribution of each cause of death to the summary probability of death. Although discrete age categories from life table calculations were used, the age categories slightly overlap for calculations of probability of death (ages 20 years and 55 years; see Section 6 in Supplement 1). To decompose the key drivers of life loss, the probability of death was determined and examined in parallel to the cause fractions for that same age group. Additional information on the decomposition of changes in the probability of death, including the formulas used, is available in the online methods section (Supplement 1).
GBD 2015 created a summary indicator that combines measures of income per capita, educational attainment for age 15 years or older, and total fertility rates.39,42-46 This indicator is updated for each GBD round. The current sociodemographic index (SDI) was used to compare observed patterns of health loss to expected patterns for countries or locations with similar SDI scores. The SDI was computed similarly to the computation of the human development index to improve interpretability. Each component of the SDI was weighted equally and rescaled (range, 0 [lowest observed value during 1980-2016] to 1 [highest observed value during 1980-2016]). In the United States in 2016, the SDI ranged from 0.874 in Mississippi to 0.978 in Washington, DC (global SDI values in 2016 ranged from 0.268 in Somalia to 0.978 in Washington, DC).
Table 1 lists the 25 leading causes of death and premature mortality from 1990 to 2016. Ischemic heart disease (IHD); cancer of the trachea, bronchus, and lung; chronic obstructive pulmonary disease; Alzheimer disease and other dementias; and cancer of the colon and rectum were the 5 leading causes of death. Despite a 50.7% decline in age-standardized mortality and a 50.4% decline in age-standardized YLLs, IHD remained the leading cause of death and premature mortality. There was an increase in age-standardized mortality and in age-standardized YLLs from 1990 for chronic obstructive pulmonary disease (13.8% for deaths and 4.6% for YLLs) and for Alzheimer disease and other dementias (11.6% for deaths and 5.5% for YLLs). There was a decrease in age-standardized mortality and in age-standardized YLLs for colon and rectal cancer (29.6% for deaths and 27.9% for YLLs) and for breast cancer (32.6% for deaths and 36.0% for YLLs). Deaths from endocrine, metabolic, blood, and immune disorders increased by 89.1%, and YLLs increased by 60.3% from 1990 to 2016 (an increase in rank from 37 in 2010 to 22 in 2016). Other notable findings seen in Table 1 are declines in deaths from self-harm by firearm (13.2%) and physical violence by firearm (28.5%) but an increase in self-harm by other means (16.9%).
Table 2 provides the 25 leading diseases and injuries contributing to YLDs. Low back pain and major depressive disorders remained the first and second causes of YLDs in 2016. Age-standardized rates of low back pain declined by 12.4%, while age-standardized rates of major depressive disorder did not change from 1990. Diabetes, which was the third leading cause of YLDs, had a 29.6% increase in age-standardized rates from 1990, and increased in rank from 8 in 1990 to 3 in 2016. From 1990 to 2016, falls had an increase of 22.2% in YLDs, opioid use disorders had an increase of 10.9% in YLDs, and asthma had a decline of 12.4% in age-standardized YLD rates. Other notable findings include an increase of 9.7% in age-standardized YLD rates of hearing loss due to aging and other causes.
Figure 1 shows the 25 leading causes of DALYs in 1990 and 2016 with their mean percentage change during the period. IHD and lung cancer were the leading causes of DALYs in both years, but the age-standardized rate declined between 1990 and 2016 by 49.7% for IHD and by 32.5% for lung cancer. The age-standardized DALY rate for chronic obstructive pulmonary disease (the third leading cause in 2016) increased by 5.5% between 1990 and 2016, and for diabetes (the fourth leading cause in 2016), it increased by 11%. Diabetes increased from the sixth leading cause in 1990 to the fourth in 2016, while low back pain declined from the third leading cause to the fifth. Three leading causes of DALYS had declines in age-standardized rates from 1990 to 2016: motor vehicle road injuries (by 35.0%), breast cancer (by 34.3%), and colorectal cancer (by 27.4%). Four leading causes of DALYS had increases in age-standardized rates from 1990 to 2016: opioid use disorders (by 47.9%), chronic kidney disease (by 44.3%), self-harm by other means (by 20.3%), and falls (by 19.0%).
Figure 2 shows the number of deaths and the percentage of DALYs from 17 risk factors in 2016. Diet, tobacco use, and high systolic blood pressure were the leading causes of deaths while tobacco use, high body mass index, and diet were the leading risk factors for DALYs. For example, dietary risks accounted for 529 299 deaths in 2016, with 83.9% of these deaths due to cardiovascular diseases, and the remainder due to a combination of neoplasms and diabetes, and to urogenital, blood, and endocrine diseases. Alcohol and drug use were the eighth leading cause of death and the fourth leading cause of DALYs. In 2016, each of the 6 following risks accounted for more than 5% of DALYs: tobacco consumption, high body mass index, diet, alcohol and drug use, high fasting plasma glucose levels, and high blood pressure.
In 2016, 44.9% of total DALYs in the United States were attributable to risk factors. Behavioral risk factors accounted for the largest percentage of the attributable fraction of DALYs due to all causes (43.5%), followed by metabolic (22.7%), and environmental and occupational risks (3.7%) (Supplement 2).
GBD 2016 showed substantial variations in the burden of diseases, injuries, and risk factors by state. There was also a variation in trends by age, sex, and state (key findings and results of burden by state in Supplement 3).
Life expectancy and HALE at birth for both sexes combined for the United States, all 50 states, and for Washington, DC are shown in Table 3. Hawaii had the highest life expectancy at birth in 2016 (81.3 years [95% UI, 80.6 to 81.9]), while Mississippi had the lowest (74.7 years [95% UI, 73.5 to 76.1]; a 6.6-year difference). Other states with high life expectancy were California (80.9 years [95% UI, 79.9 to 81.9]), Connecticut (80.8 years [95% UI, 79.7 to 81.8]), Minnesota (80.8 years [95% UI, 80.0 to 81.6]), New York (80.5 years [95% UI, 79.4 to 81.6]), Massachusetts (80.4 years [95% UI, 79.6 to 81.1]), Colorado (80.2 years [95% UI, 79.4 to 80.9]), New Jersey (80.2 years [95% UI, 79.3 to 80.9]), and Washington (80.2 years [95% UI, 79.5 to 80.8]). Other states with low life expectancy were West Virigina (75.3 years [95% UI, 74.4 to 76.0]), Alabama (75.4 years [95% UI, 74.1 to 76.7]), Louisiana (75.6 years [95% UI, 74.9 to 76.4]), Oklahoma (75.7 years [95% UI, 75.0 to 76.4]), Arkansas (75.8 years [95% UI, 74.9 to 76.8]), and Kentucky (75.8 years [95% UI, 74.9 to 76.6]). In 2016, Minnesota had the highest HALE at birth with 70.3 years, while West Virginia had the lowest at 63.8 years, a 6.5-year difference. Only 2 states, Minnesota and Hawaii, had HALE values greater than 70.0 years at birth in 2016. In terms of life expectancy in 2016, only 9 states had life expectancy values greater than 80.0 years.
Male life expectancy and HALE at birth for the United States overall, for all states, and for Washington, DC, are shown in Table 4. Minnesota had the highest life expectancy in 2016 (78.7 years [95% UI, 77.5 to 79.8]) and HALE (69.1 years [95% UI, 66.3 to 71.9]), followed by California (life expectancy, 78.6 years [95% UI, 77.2 to 80.1]; and HALE, 68.6 years [95% UI, 65.5 to 71.6]). Mississippi had the lowest life expectancy for males in 2016 (71.8 years [95% UI, 70.1 to 73.8]) and ranked 49th for HALE (63.0 years [95% UI, 60.3 to 65.6]), while West Virginia had the lowest HALE (62.2 years [95% UI, 54.9 to 65.0]) and ranked 49th for life expectancy (72.7 years [95% UI, 71.5 to 73.9).
Female life expectancy and HALE at birth for the United States, for all states, and for Washington, DC, are shown in Table 5. Hawaii had the highest life expectancy in 2016 (84.1 years [95% UI, 83.2 to 85.0) and HALE (71.9 years [95% UI, 68.3 to 75.1), followed by life expectancy for California (83.1 years [95% UI, 81.8 to 84.3]) and life expectancy for Connecticut (83.1 years [95% UI, 81.7 to 84.4]); the second highest HALE was for Minnesota (71.4 years [95% UI, 68.3 to 74.5]). Mississippi had the lowest life expectancy for females (77.7 years [95% UI, 76.1 to 79.6]), while West Virginia had the lowest HALE (65.5 years [95% UI, 61.9 to 68.5]).
Table 6 presents the age-standardized death rates, age-standardized YLL rates, and age-standardized YLD rates in 1990 and 2016 and their ranks by state. The 3 measurements varied widely between the states in 2016, ranging from 767.6 deaths per 100 000 in Mississippi to 465.8 deaths per 100 000 in Hawaii, from 17 775.9 YLLs per 100 000 in Mississippi to 9901.8 YLLs per 100 000 in Minnesota, and from 13 090.6 YLDs per 100 000 in West Virginia to 10 582.8 YLDs per 100 000 in Minnesota. A notable improvement was observed in Washington, DC (decreases from 1042.7 deaths per 100 000 to 603.3 deaths per 100 000, from 29 536.9 YLLS per 100 000 to 13 635.9 YLLs per 100 000, and from 12 230.8 YLDs per 100 000 to 11 421.1 YLDs per 100 000) and in California (decreases from 719.1 deaths per 100 000 to 491.7 deaths per 100 000, from 15 903.4 YLLS per 100 000 to 9987.0 YLLs per 100 000, and from 11 170.5 YLDs per 100 000 to 10 990.4 YLDs per 100 000). Decreases in mortality and increases in morbidity were more apparent in Ohio (from 761.5 deaths per 100 000 to 644.1 deaths per 100 000, from 16 349.6 YLLs per 100 000 to 13 853.3 YLLs per 100 000, and from 12 009.0 YLDs per 100 000 to 12 334.7 YLDs per 100 000) and in Oklahoma (from 773.8 deaths per 100 000 to 725.3 deaths per 100 000, from 17 062.7 YLLs per 100 000 to 16 379.3 YLLs per 100 000, and from 12 036.5 YLDs per 100 000 to 12 549.7 YLDs per 100 000). The age-standardized death rates and age-standardized YLL rates declined in all states, but the level of decline for deaths ranged from 6.3% in Oklahoma to 42.1% in Washington, DC and the level of decline for YLLs ranged from 4.0% for Oklahoma to 53.8% for Washington, DC. Age-standardized YLD rates increased by 4.4% for West Virginia and declined by 6.6% for Washington, DC.
The decomposition of change in the probability of death from birth to age 20 years, ages 20 to 55 years, and ages 55 to 90 years are shown in Figure 3, Figure 4, and Figure 5. For the United States and each state, these figures show the change in the probability of death from 1990 to 2016. In addition, these figures show changes in the probability of death over the interval due to changes in causes of death (GBD cause hierarchy level 2). The change in the probability of death from birth to age 20 years declined in all states. The most pronounced declines were in South Carolina, Georgia, Alaska, and New York, at higher than a 0.85-point decrease in probability (Figure 3). In contrast, Maine had the lowest decline of 0.32-point probability. In the United States as a whole, there was a decline of 0.70, which was associated with neonatal disorders, other noncommunicable diseases (including congenital), and a large contribution from injuries, with slight increases from mental and substance use disorders (Figure 3).
The largest reductions in probability of death for ages 20 to 55 years were observed in New York (3.5) and California (2.5) and the highest increases were observed in West Virginia (2.6) and Oklahoma (2.0) (Figure 4). In 21 states, the probability of death has actually increased from 1990 to 2016, but of these, only 5 showed an increase of greater than 10% (Kentucky, Oklahoma, New Mexico, West Virginia, and Wyoming). Conversely, 31 states and Washington, DC have seen decreases in the probability of death among adults aged 20 to 55 years over the same period, but only in 15 states was that decrease more than 10% (New York, California, Illinois, New Jersey, Georgia, Maryland, Florida, Nevada, Texas, Virginia, Connecticut, North Carolina, Massachusetts, Washington, South Carolina). Decreases in the probability of death in US states were influenced by declines in HIV/AIDS across all state groups, as well as declines in road injuries and neoplasms, while increases in probability of death were influenced by increased burden of drug use disorders, alcohol use disorders, and chronic kidney disease, among others (Figure 4).
All states experienced a considerable reduction in probabilities of death for ages 55 to 90 years, largely associated with reductions in the probability of dying from cardiovascular diseases (Figure 5). The highest point decline was observed in California at 12.6 points, compared with lowest decline of 3.5 points for Mississippi. These declines were somewhat offset by increases in the death rates associated with cirrhosis and other liver disease, neurological disorders, and mental and substance use disorders in all states. Hawaii was the only state in which the probability of death was less than 65% for ages 55 to 90 years. Other notable findings include the variation in the decline of probability of death between the 3 age groups, with an 8.3-point decline of probability of death for ages 55 to 90 years, a 1-point decline for ages 20 to 55 years, and a 0.7-point decline for ages 0 to 20 years.
The age-standardized YLL rates for the United States, all states, and Washington, DC in 2016 for the top 20 causes are grouped by 3 levels of significance (Figure 6) indicating significantly below the mean, indistinguishable from the mean, and significantly higher than the mean. The heat map shows a clear pattern of performance ranging from Minnesota to Mississippi, with some clear patterns of exception for some causes. For example, Colorado had a YLL rate from self-harm significantly above the mean (760), while Washington, DC had a YLL rate significantly lower than the mean (306). Mississippi, West Virginia, Alabama, Oklahoma, Kentucky, Tennessee, South Carolina, Indiana, Missouri, and Ohio had YLLs significantly higher than the mean with a few exceptions of causes that were indistinguishable from the mean. Other notable findings were that Louisiana had YLLs for all causes higher than that of the US mean except for Alzheimer diseases and other dementias that was significantly higher than the mean, while Georgia had only 1 cause, drug use disorder, that was significantly above the US mean.
The ratio of observed YLLs to those expected, based on the SDI for the 10 leading causes of YLLs for the United States overall, each individual state, and Washington, DC, are shown in Figure 7. For example, in Alabama, the stroke ratio was 1, indicating that the observed rates are similar to what would be expected given the state’s SDI, whereas diabetes is observed at 2.85 times more than expected. For the United States overall, the ratios of observed YLLs to those expected were 0.75 for IHD, 1.11 for lung cancer, 1.88 for chronic obstructive pulmonary disease, 0.61 for stroke, 1.19 for road injury, 1.14 for Alzheimer disease, 0.98 for self-harm, 7.17 for drugs, 0.84 for colorectal cancer, and 1.27 for lower respiratory tract infection. In general, most states performed better on IHD and stroke but worse for chronic obstructive pulmonary disease and drug use disorders. Colorado had the best performance for IHD at 0.58 observed to expected ratio of YLLs, while West Virginia had the highest observed to expected ratio of YLLs for drug use disorders at 14.38. Other notable findings are the high rates in Washington, DC for interpersonal violence, drug use disorders, HIV/AIDS, and chronic kidney disease, given that Washington, DC has the highest SDI rank in the US and in the world.
The ratio of the YLDs observed to those expected based on SDI for the 10 leading causes of YLDs for the United States overall, the 50 states, and Washington, DC are shown in Figure 8. Minnesota had lower observed YLDs from low back and neck pain (0.63 ratio) and migraine (0.79), but higher YLD rates from drug use disorders (2.32). Most states had lower than expected YLDs from low back and neck pain but higher rates for drug use disorders and other musculoskeletal disorders. A notable finding is the higher than expected rates of YLDs from depression, anxiety, and skin diseases, and lower than expected rates from falls in most states.
The ratio of DALYs observed to those expected based on SDI for the 10 leading causes of DALYs for the United States overall, the 50 states, and Washington, DC are shown in Figure 9. All states had lower than expected DALY rates from IHD except for Alabama, Arkansas, Kentucky, Mississippi, Oklahoma, Tennessee, and West Virginia. The highest observed rates of DALYs compared with expected from drug use disorders were in West Virginia at 7.77 and in Kentucky at 7.31. Other notable findings are the higher than expected rates of drug use disorders and interpersonal violence in Washington, DC, and the lower than expected rates of lung cancer in California.
The age-standardized percent changes in summary exposure values for the top 10 risk factors from 1990 to 2016 are shown in Table 7. High fasting plasma glucose increased by 76% and high body mass index by 53.2%, while smoking declined by 42.8% during the same time period in the United States. There were clear patterns in these variations by states. High fasting plasma glucose increased in all states; the increase ranged from 127.2% in Mississippi to 1.7% in Pennsylvania. Drug use increased in all states except Arkansas, Maryland, and Oregon. Other notable findings include reductions in high systolic blood pressure, high total cholesterol levels, and diet low in whole grains in all states.
The rank of risk factors by state in 2016 are shown in Figure 10. Tobacco smoking was the leading risk factor for DALYs in the United States and in 33 states. Alcohol and drug use were the leading risk for DALYs for 7 states and Washington, DC, while high BMI was the leading cause for 10 states (California, Connecticut, Hawaii, Illinois, Maryland, New Jersey, New York, North Dakota, Texas, and Virginia). Another notable finding is that diet was the third leading cause of DALYs in the United States overall but the second in 20 states.
Quiz Ref IDThis study provides a comprehensive report on the burden of disease and its patterns in the United States and the individual 50 states from 1990 to 2016 and reveals wide disparities in burden of disease at the state level. Moreover, these findings show distinct trends in different age bands and demonstrate that improvement in some health outcomes, such as IHD, lung cancer, and neonatal preterm complications, are balanced by rising death rates from drug use disorders, chronic obstructive pulmonary disease, self-harm, chronic kidney disease, cirrhosis, and hypertensive heart disease. Summary measures, such as life expectancy, that do not differentiate the trends in different age groups mask the heterogeneous directions for US health status by age and state. Above and beyond the drivers of divergent trends, the study reveals that there has been far greater progress in reducing the burden of some major causes of YLLs, such as IHD and lung cancer, but no progress in addressing some of the leading causes of YLDs such as mental health disorders and musculoskeletal disorders. These findings should be used to examine the causes of health variations and to plan, develop, and implement programs and policies to improve health overall and eliminate disparities in the United States.
Mortality reversals in 21 states for adults ages 20 to 55 years are strongly linked to the burden of substance use disorders, cirrhosis, and self-harm, and this study shows that the trends for some of these conditions differ considerably across different states. Case and Deaton have called some of these conditions “deaths of despair” and argued that they are linked to the social and economic status of white US adults.3 A wide range of interventions have been proposed to address substance abuse, cirrhosis, and self-harm. For substance abuse,28,47-49 prevention programs should account for the root causes of substance use, the socioeconomic factors involved, and causes of relapses during treatment.50,51 Physicians have a major role to play in addiction control by counseling their patients who are on medication for pain control.52-54 For cirrhosis, intervention strategies to treat hepatitis C and decrease excessive alcohol consumption are important. For self-harm, the most promising approaches relate to decreasing the case-fatality rate from suicide attempts by restricting access to lethal means; in the United States, a large share of suicides are due to firearms.55,56 While multiple strategies are available for dealing with these problems, they have not until very recently garnered attention. The increases in these causes of death have been underway for 20 years, but the US health policy community has been slow to recognize this rising set of problems.
Quiz Ref IDThis study shows that high BMI, smoking, and high fasting plasma glucose are the 3 most important risk factors in the United States, and that although smoking is decreasing, BMI and fasting plasma glucose levels are steadily increasing. These 2 risk factors pose unique challenges in the United States given that unabated, they have the potential to change the health trajectory for individuals in many states. Levels of overweight and obesity increased during the study period. US residents need to do more to maintain their weight or reduce it, when needed, as well as access systems to support these intentions.57 Although physical activity increased during the study period, the levels of increase were not enough to control weight gain.8 Physical inactivity is a risk factor for many diseases, but increasing activity is not enough on its own to reduce weight or prevent weight gain.58 Obesity is associated with increased diabetes, cardiovascular diseases, some neoplasms, and poor health-related quality of life. This study calls for renewed efforts to control weight gain at the community level.
Several studies have reported on the effect of taxes on sugary drinks or subsidies to encourage consumption of healthy foods, although only a small fraction of obesity can be linked to sugar-sweetened beverages.59 A comprehensive plan is needed to address obesity because it adversely affects health and drives use of health care resources.4,60-62 Rising BMI is driving up fasting plasma glucose levels and diabetes, and diabetes increased as a cause of burden in almost all states during the study period. Diabetes is a costly disease that consumes approximately 4.82% of the US health care budget.4 A recent study estimated that the cost of diabetes care increased by 6.1% from 1996 to 2013.4 Diabetes is associated with many conditions and disabilities.63 This rise in burden and its cost is noteworthy given the projected increase in diabetes as obesity increases in the United States.
Quiz Ref IDThis study shows that there were reductions in the death rates from cardiovascular diseases for all age groups. This progress has been, in part, influenced by reductions in systolic blood pressure and cholesterol, but the role of increased access to effective treatment has also been considerable.64 In fact, age-standardized CVD death rates have decreased in the United States by 32.8% over the last 20 years. The important role of treatment in reducing death rates highlights the ongoing importance of ensuring financial and physical access to care and the importance of quality of care. As declines in the rate of CVD may be slowing down, adverse trends due to the diseases of despair and adverse risk trends may mean that historical progress in improved life expectancy may not continue in the future.
The strategies for dealing with the remaining inequalities and new threats are 3-fold: (1) address some of the key modifiable risks, including diet; tobacco, alcohol, and drug use; insufficient physical activity; and obesity; (2) improve access to and, more importantly, quality of care in key areas, such as chronic kidney disease and ongoing care for substance use disorders; and (3) address the social determinants of health. We have previously shown an association between socioeconomic and race/ethnicity factors and a 60% county-level variation in life expectancy, behavioral and metabolic risk factors and a 74% variation, and health care factors and a 27% variation.65 Combined, these factors are associated with 74% of overall variation. We also reported that most of the association between socioeconomic and race/ethnicity factors and life expectancy was mediated through behavioral and metabolic risk factors. Research has shown that some environmental factors have an effect on risk factors such as obesity and low physical activity. Low socioeconomic settings often have an imbalance of few grocery stores and numerous fast food options, and access to safe outdoor places spaces for recreation is limited.66,67
To date, the strategies for addressing the social determinants of health in the United States have been elusive. Lack of progress and rising social inequalities should not engender complacency. Therefore, addressing risk factors may prove to be an important opportunity to reduce disparities and deal with some of the challenges for improving US health. Opportunities to decrease the burden of disease through reducing tobacco, alcohol, drug use, blood pressure, and cholesterol; increasing physical activity; and improving diet emphasize that the United States should invest more in prevention that targets these risks. To increase the likelihood of prevention to succeed, it has to be a priority for all stakeholders—physicians, nurses, hospital systems, policy makers, health insurance companies, patients and their families, and advocacy groups.
This study showed a wide range of challenges encountered by different states and by some counties within states. Given the diversity of risks, communities, and workplaces there is no simple menu of effective programs for risk reduction. Indeed, local experimentation to determine what works in a given community is likely to be necessary. There is a need to change strategies of funding and evaluation of innovative interventions and policies, and independent evaluation of whether these efforts work or not should be documented. To succeed, these innovative programs should forge a connection between health care provision and progress for individuals and communities in health outcomes. The notion of accountability should be broadened beyond providing high-quality care to encompass achieving risk reduction in partnership with patients and communities.
The interpersonal violence burden in the United States has to be properly addressed, although declines have been observed: the age-standardized death rate decreased 32.43% from 1990 to 2016 in the United States. However, self-harm by firearm accounted for 6.39 deaths per 100 000 persons in the United States in 2016, and physical violence by firearm accounted for 3.98 deaths per 100 000 persons. There is evidence that gun control achieved through background checks reduces homicide and suicide.68-72 Previous studies reported success in reducing the burden of gun deaths through policy changes in Brazil and other countries.71 Indeed, enforcing gun control policies has proven effective in reducing mortality in a variety of contexts. There is a need for comprehensive studies of the epidemiology of gun violence in the United States to inform the ongoing gun control debate.
Age-standardized death rates due to alcohol increased by 17.50% from 1990 to 2016 in the United States, and alcohol use disorders accounted for 2.89 deaths per 100 000 persons in 2016. Previous studies have shown that alcohol consumption and binge drinking have increased in the United States, especially among females.6,73 Alcohol is a major risk factor for burden in the United States10 and is associated with adverse outcomes including sexually transmitted diseases, violence, and accidents.16 Traffic injuries have received a considerable amount of attention, but the true burden of alcohol is much bigger and goes beyond driving while intoxicated. Programs to educate US residents about the true harms of excessive drinking are needed.
Many of the risk factors that contribute to the disparities in burden are amenable to medical treatment within the context of supportive behavioral and lifestyle changes. For example, many cardiovascular risk factors, such as high blood pressure and high cholesterol levels, can now be treated more effectively with early detection and proper follow-up. Safe, effective, and affordable antihypertensive medications are now widely available, often as generic preparations, especially at discount pharmacies, and in many cases, without the need for health insurance for medications. The Affordable Care Act (ACA) allows for the expansion of insurance coverage, which ultimately increases access to care. Indeed, the ACA expansion of Medicaid coverage in participating states to all nonelderly adults with incomes below 133% of the federal poverty level provides an opportunity for early detection and follow-up of some of the main health risk factors. However, many US residents do not have health insurance, even after the ACA was introduced, and hence have little access to medical diagnosis and treatment. Therefore, expanding health coverage for certain conditions and medications should be considered and adopted to reduce burden.
This study showed that the United States overall and many individual states have made progress in reducing mortality but have had limited success in reducing disability. For instance, the burden of drug use disorders in total DALYs increased in the United States during the study period by 61.4% and accounted for about 3.81 million DALYs in 2016, depressive disorders increased 17.32% and accounted for 2.72 million DALYs, and anxiety disorders increased by 16.7% and accounted for approximately 1.76 million DALYs in 2016. These findings point to an urgent need to address mental health and drug use disorders in the United States. There is a need for improved access to quality mental health care and screening to improve outcomes, as well as programs to prevent mental disorders and promote mental health.
As the US population ages, the burden of musculoskeletal disorders will increase. More US residents have neck and back pain, and the incidence of falls is increasing. Musculoskeletal disorders are associated with a high medical cost.4 Preventive measures to reduce the burden of these risk factors in all stages of life are urgently needed. Programs for avoiding harm and injuries at work among both younger and older ages are needed. The programs should include prevention of falls in the older population through examining the risk factors that lead to falls among adults. Screening tools and interventions to address this burden should be implemented.
The results of GBD 2016 have shown that occupational risks and air pollution were the 9th and 10th leading causes for DALYs. Although the findings show reductions in attributable burden from 1990 to 2016, occupational risks still account for 948.75 DALYS per 100 000 persons, and air pollution accounts for 584.97 DALYs per 100 000 persons—large numbers in the United States. Indeed, renewed efforts to reduce the burden of environmental and occupational risks are needed to ensure continued progress in reducing their effect on health in the United States.
Several studies have shown that poor diet is a major challenge in the United States, and little improvement has occurred over the past decades. US residents are not consuming a healthy diet; they tend to consume more calories than needed, and composition is not ideal.49,62 Some recent studies have shown modest improvement in certain aspects of US diet, especially decreases in consumption of sugary drinks. The United States needs a comprehensive program to improve dietary intake at national and local levels. This program should offer financial incentives and disincentives for more vs less healthful food products by agriculture producers, food manufacturers, and retailers, as well as for choices by consumers. This effort should also implement comprehensive wellness programs in schools, workplaces, and government offices, and inform the public of the importance of a healthful diet.
Given the scope of this analysis, this study has several important limitations. The overall limitations of the GBD methods, as noted in other publications, apply to the US analysis. First, the accuracy of the estimates depends on the availability of data by time period and state. Second, it is challenging to separate measurement error from variation in disease occurrence. GBD corrects for known bias from nonreference methods or case definitions, but often has to rely on sparse data at the state level to make those adjustments. Third, GBD applies garbage code redistribution for 13% of causes of deaths in the United States; this ranged from 8.4% in South Dakota to 21.3% in Alabama. Therefore, the causes of death may not match those in other publications but are more robust because they control for the between-states variation in the prevalence of garbage codes. Fourth, GBD methods adjust for hospital admissions using a large nonrepresentative source of medical claims data. The generalizability of claims data, the use of primary diagnosis only or all diagnostic fields, and the trends of claims data have been questioned.13 Also, there may be considerable interstate variation in how diseases are treated between inpatient and outpatient settings. GBD methods adjust for such potential biases by using a covariate on claims and hospital admissions data to correct for systematic error. Fifth, GBD includes risk-outcome pairs that meet the World Cancer Research Fund criteria of causality. However, some risk-outcome pairs might not meet criteria that develop as evidence from new studies is published. Sixth, there is limited information on dietary intake at the state level. The Behavioral Risk Factors Surveillance System has 6 dietary questions attempting to capture fruit and vegetable consumption. Therefore, GBD 2016 used commercial sales data to adjust estimates of dietary intake. Seventh, some of the data used in the analyses have a lower quality and consistency across states and age groups. GBD 2016 reports 95% UIs to show the effect of this limitation on the estimates. Eighth, the study reports disparities between states but does not examine the within-state variations of burden, which could be substantial, especially in large states. Ninth, claims data were only available through 2012 at the time of these analyses. Additionally, this study does not report the burden of the social determinants of health; it focuses only on behavioral, environmental, and metabolic risks.
There are wide differences in the burden of disease at the state level. Specific diseases and risk factors, such as drug use disorders, high BMI, poor diet, high fasting plasma glucose level, and alcohol use disorders are increasing and warrant increased attention. These data can be used to inform national health priorities for research, clinical care, and policy.
Accepted for Publication: February 13, 2018.
Corresponding Author: Christopher J. L. Murray, MD, DPhil, Institute for Health Metrics and Evaluation, University of Washington, 2301 5th Ave, Ste 600, Seattle, WA 98121 (email@example.com).
US Burden of Disease Collaborators: Ali H. Mokdad, PhD; Katherine Ballestros, PhD; Michelle Echko, BS; Scott Glenn, MSc; Helen E. Olsen, MA; Erin Mullany, BA; Alex Lee, BS; Abdur Rahman Khan, MD; Alireza Ahmadi, MD; Alize J. Ferrari, PhD; Amir Kasaeian, PhD; Andrea Werdecker, PhD; Austin Carter, BS; Ben Zipkin, BS; Benn Sartorius, PhD; Berrin Serdar, PhD; Bryan L. Sykes, PhD; Chris Troeger, MPH; Christina Fitzmaurice, MD; Colin D. Rehm, PhD; Damian Santomauro, PhD; Daniel Kim, DrPH; Danny Colombara, PhD; David C. Schwebel, PhD; Derrick Tsoi, BS; Dhaval Kolte, MD; Elaine Nsoesie, PhD; Emma Nichols, BA; Eyal Oren, PhD; Fiona J. Charlson, PhD; George C. Patton, MD; Gregory A. Roth, MD; H. Dean Hosgood, PhD; Harvey A. Whiteford, PhD; Hmwe Kyu, PhD; Holly E. Erskine, PhD; Hsiang Huang, MD; Ira Martopullo, MPH; Jasvinder A. Singh, MD; Jean B. Nachega, PhD; Juan R. Sanabria, MD; Kaja Abbas, PhD; Kanyin Ong, PhD; Karen Tabb, PhD; Kristopher J. Krohn, MPH; Leslie Cornaby, BS; Louisa Degenhardt, PhD; Mark Moses, MHS; Maryam Farvid, PhD; Max Griswold, MA; Michael Criqui, MD; Michelle Bell, PhD; Minh Nguyen, BS; Mitch Wallin, MD; Mojde Mirarefin, MPH; Mostafa Qorbani, PhD; Mustafa Younis, DrPH; Nancy Fullman, MPH; Patrick Liu, MPH; Paul Briant, BS; Philimon Gona, PhD; Rasmus Havmoller, PhD; Ricky Leung, PhD; Ruth Kimokoti, MD; Shahrzad Bazargan-Hejazi, PhD; Simon I. Hay, DSc; Simon Yadgir, BS; Stan Biryukov, BS; Stein Emil Vollset, DrPH; Tahiya Alam, MPH; Tahvi Frank, BS; Talha Farid, MD; Ted Miller, PhD; Theo Vos, PhD; Till Bärnighausen, MD; Tsegaye Telwelde Gebrehiwot, MPH; Yuichiro Yano, MD; Ziyad Al-Aly, MD; Alem Mehari, MD; Alexis Handal, PhD; Amit Kandel, MBBS; Ben Anderson, MD; Brian Biroscak, PhD; Dariush Mozaffarian, MD; E. Ray Dorsey, MD; Eric L. Ding, ScD; Eun-Kee Park, PhD; Gregory Wagner, MD; Guoqing Hu, PhD; Honglei Chen, PhD; Jacob E. Sunshine, MD; Jagdish Khubchandani, PhD; Janet Leasher, OD; Janni Leung, PhD; Joshua Salomon, PhD; Jurgen Unutzer, MD; Leah Cahill, PhD; Leslie Cooper, MD; Masako Horino, MPH; Michael Brauer, ScD; Nicholas Breitborde, PhD; Peter Hotez, PhD; Roman Topor-Madry, PhD; Samir Soneji, PhD; Saverio Stranges, PhD; Spencer James, MD; Stephen Amrock, MD; Sudha Jayaraman, MD; Tejas Patel, MD; Tomi Akinyemiju, PhD; Vegard Skirbekk, PhD; Yohannes Kinfu, PhD; Zulfiqar Bhutta, PhD; Jost B. Jonas, MD; Christopher J. L. Murray, DPhil.
Affiliations of US Burden of Disease Collaborators: Institute for Health Metrics and Evaluation, University of Washington, Seattle (Mokdad, Ballestros, Echko, Glenn, Olsen, Mullany, Lee, Ferrari, Carter, Zipkin, Troeger, Fitzmaurice, Santomauro, Colombara, Tsoi, Nsoesie, Nichols, Charlson, Roth, Whiteford, Kyu, Erskine, Martopullo, Ong, Krohn, Cornaby, Degenhardt, Moses, Griswold, Nguyen, Mirarefin, Fullman, Liu, Briant, Hay, Yadgir, Biryukov, Vollset, Alam, Frank, Vos, Brauer, James, Murray); University of Louisville, Louisville, Kentucky (Khan, Farid); Kermanshah University of Medical Sciences, Kermanshah, Iran (Ahmadi); Karolinska Institutet, Stockholm, Sweden (Ahmadi, Havmoller); School of Public Health, University of Queensland, Brisbane, Australia (Ferrari, Santomauro, Charlson, Whiteford, Erskine); Queensland Centre for Mental Health Research, Brisbane, Australia (Ferrari, Santomauro, Charlson, Whiteford, Erskine); Hematology-Oncology and Stem Cell Transplantation Research Center, and Hematologic Malignancies Research Center, Tehran University of Medical Sciences, Tehran, Iran (Kasaeian); Competence Center Mortality-Follow-Up of the German National Cohort, Federal Institute for Population Research, Wiesbaden, Hessen, Germany (Werdecker); Public Health Medicine, School of Nursing and Public Health, University of KwaZulu-Natal, Durban, South Africa (Sartorius); UKZN Gastrointestinal Cancer Research Centre, South African Medical Research Council, Durban, South Africa (Sartorius); University of Colorado, Aurora (Serdar); Departments of Criminology, Law & Society, Sociology, and Public Health, University of California, Irvine (Sykes); Division of Hematology, Department of Medicine, University of Washington, Seattle, and Fred Hutchinson Cancer Research Center, Seattle (Fitzmaurice); Montefiore Medical Center, Bronx, New York (Rehm); Department of Health Sciences, Northeastern University, Boston, Massachusetts (Kim); University of Alabama at Birmingham (Schwebel, Singh, Akinyemiju); Division of Cardiology, Brown University, Providence, Rhode Island (Kolte); Division of Epidemiology & Biostatistics, Graduate School of Public Health, San Diego State University, San Diego, California (Oren); Murdoch Childrens Research Institute, Department of Paediatrics, University of Melbourne, Melbourne, Victoria, Australia (Patton); Albert Einstein College of Medicine, Bronx, New York (Hosgood); Cambridge Health Alliance, Cambridge, Massachusetts (Huang); Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pennsylvania (Nachega); Stellenbosch University, Cape Town, Western Cape, South Africa (Nachega); Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland (Nachega); Joan C. Edwards School of Medicine, Marshall University, Huntington, West Virginia (Sanabria); Case Western Reserve University, Cleveland, Ohio (Sanabria); Department of Infectious Disease Epidemiology, London School of Hygiene & Tropical Medicine, London, England (Abbas); School of Social Work, University of Illinois at Urbana-Champaign (Tabb); National Drug and Alcohol Research Centre, University of New South Wales, Sydney, Australia (Degenhardt); Harvard T.H. Chan School of Public Health, Harvard University, Boston, Massachusetts (Farvid, Ding, Cahill); Harvard/MGH Center on Genomics, Vulnerable Populations, and Health Disparities, Mongan Institute for Health Policy, Massachusetts General Hospital, Boston (Farvid); University of California, San Diego, La Jolla, California (Criqui); Yale University, New Haven, Connecticut (Bell, Biroscak); VA Medical Center, Washington, DC (Wallin); Neurology Department, Georgetown University, Washington, DC (Wallin); Hunger Action Los Angeles, Los Angeles, California (Mirarefin); Non-communicable Diseases Research Center, Alborz University of Medical Sciences, Karaj, Alborz, Iran (Qorbani); Jackson State University, Jackson, Mississippi (Younis); University of Massachusetts, Boston (Gona); State University of New York, Albany, Rensselaer, New York (R. Leung); Simmons College, Boston, Massachusetts (Kimokoti); College of Medicine, Charles R. Drew University of Medicine and Science, Los Angeles, California (Bazargan-Hejazi); David Geffen School of Medicine, University of California at Los Angeles (Bazargan-Hejazi); Oxford Big Data Institute, Li Ka Shing Centre for Health Information and Discovery, University of Oxford, Oxford, United Kingdom (Hay); Center for Disease Burden, Norwegian Institute of Public Health, and Department of Global Public Health and Primary Care, University of Bergen, Bergen, Norway (Vollset); Pacific Institute for Research & Evaluation, Calverton, Maryland (Miller); School of Public Health, Curtin University, Perth, Australia (Miller); Department of Global Health and Population, Harvard T.H. Chan School of Public Health, Harvard University, Boston, Massachusetts (Bärnighausen, Salomon); Africa Health Research Institute, Mtubatuba, South Africa (Bärnighausen); Institute of Public Health, Heidelberg University, Heidelberg, Germany (Bärnighausen); Jimma University, Jimma, Oromia, Ethiopia (Gebrehiwot); Department of Preventive Medicine, Northwestern University, Chicago, Illinois (Yano); Washington University in St Louis, St Louis, Missouri (Al-Aly); College of Medicine, Howard University, Washington, DC (Mehari); University of New Mexico, Albuquerque (Handal); University at Buffalo, Buffalo, New York (Kandel); University of Washington, Seattle (Anderson, Sunshine, J. Leung, Unutzer); University of South Florida, Tampa (Biroscak); Friedman School of Nutrition Science and Policy, Tufts University, Boston, Massachusetts (Mozaffarian); University of Rochester Medical Center, Rochester, New York (Dorsey); Department of Medical Humanities and Social Medicine, College of Medicine, Kosin University, Busan, South Korea (Park); Department of Environmental Health, Harvard T.H. Chan School of Public Health, Harvard University, Boston, Massachusetts (Wagner); Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha, Hunan, China (Hu); Michigan State University, East Lansing (Chen); Department of Nutrition and Health Science, Ball State University, Muncie, Indiana (Khubchandani); College of Optometry, Nova Southeastern University, Fort Lauderdale, Florida (Leasher); School of Public Health, University of Queensland, Brisbane, Australia (J. Leung); Dalhousie University, Halifax, Canada (Cahill); Mayo Clinic, Jacksonville, Florida (Cooper); Bureau of Child, Family & Community Wellness, Nevada Division of Public and Behavioral Health, Carson City (Horino); University of British Columbia, Vancouver, Canada (Brauer); The Ohio State University, Columbus (Breitborde); College of Medicine, Baylor University, Houston, Texas (Hotez); Institute of Public Health, Faculty of Health Sciences, Jagiellonian University Medical College, Kraków, Poland (Topor-Madry); Faculty of Health Sciences, Wroclaw Medical University, Wroclaw, Poland (Topor-Madry); Dartmouth College, Hanover, New Hampshire (Soneji); Department of Epidemiology & Biostatistics, Schulich School of Medicine & Dentistry, Western University, London, Canada (Stranges); Department of Population Health, Luxembourg Institute of Health, Strassen, Luxembourg (Stranges); Oregon Health & Science University, Portland (Amrock); Department of Surgery, Virginia Commonwealth University, Richmond (Jayaraman); White Plains Hospital, White Plains, New York (Patel); Norwegian Institute of Public Health, Oslo, Norway (Skirbekk); Columbia University, New York, New York (Skirbekk); Centre for Research and Action in Public Health, University of Canberra, Canberra, Australia (Kinfu); Centre of Excellence in Women and Child Health, Aga Khan University, Karachi, Pakistan (Bhutta); Centre for Global Child Health, The Hospital for Sick Children, Toronto, Canada (Bhutta); Department of Ophthalmology, Medical Faculty Mannheim, Ruprecht-Karls-University Heidelberg, Heidelberg, Germany (Jonas).
Author Contributions: Dr Murray had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Murray, Mokdad, Troeger, Bell, Vos, Sykes, Dorsey, Cooper, Ahmadi, Kinfu, Younis, James.
Acquisition, analysis, or interpretation of data: Murray, Mokdad, Olsen, Mullany, Echko, Glenn, Ballesteros, Sartorius, Serdar, Troeger, Fitzmaurice, Rehm, Santomauro, Hosgood, Patton, Roth, Kyu, Martopullo, Nichols, Charlson, Degenhardt, Farvid, Criqui, Mirarefin, Huang, Gona, Havmoeller, Nachega, Hay, Biryukov, Vollset, Miller, Vos, Bärnighausen, Yano, Biroscak, Mozaffarian, Ding, Hu, Leasher, J. Leung, Salomon, Unützer, Horino, Bhutta, Al-Aly, Nsoesie, Schwebel, Kasaeian, Werdecker, Sykes, Colombara, Tsoi, Kolte, Sunshine, R. Leung, Erskine, Krohn, Whiteford, Oren, Kim, Singh, Ferrari, Khan, Carter, Sanabria, Abbas, Ong, Tabb, Moses, Cornaby, Fullman, Liu, Bazargan-Hejazi, Kimokoti, Breitborde, Hotez, Brauer, Dorsey, Cooper, Topor-Madry, Stranges, Amrock, Jayaraman, Patel, Akinyemiju, Ahmadi, Nguyen, Wallin, Qorbani, Yadgir, Frank, Farid, Alam, Kandel, Anderson, Park, Wagner, Chen, Khubchandani, Cahill, Soneji, James, Gebrehiwot.
Drafting of the manuscript: Murray, Mokdad, Olsen, Mullany, Glenn, Ballesteros, Roth, Vos, Horino, Krohn, Sanabria, Fullman, Ahmadi, Frank.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Mokdad, Glenn, Troeger, Fitzmaurice, Rehm, Santomauro, Roth, Kyu, Martopullo, Nichols, Charlson, Hay, Biryukov, Vos, J. Leung, Kasaeian, Sykes, Colombara, Tsoi, Erskine, Ferrari, Carter, Ong, Moses, Cornaby, Liu, Nguyen, Qorbani, Yadgir, Frank, Khubchandani.
Obtained funding: Murray, Mokdad.
Administrative, technical, or material support: Murray, Mokdad, Olsen, Echko, Nachega, Hay, Mozaffarian, Unützer, R. Leung, Krohn, Whiteford, Singh, Akinyemiju, Ahmadi, Wallin, Alam, Park, James.
Supervision: Murray, Mokdad, Olsen, Mullany, Criqui, Bell, Vos, Salomon, Stranges, Ahmadi, Alam.
Other (created custom analyses, tables, and figures, and fact checked the analysis): Echko, Ballesteros, Lee, Glenn.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Singh reports receipt of grants and personal fees from Savient, Takeda, and Crealta/Horizon; personal fees from Regeneron, Merz, Iroko, Bioiberica, Allergan, UBM LLC, WebMD, and the American College of Rheumatology during the conduct of the study; personal fees from DINORA; and nonfinancial support from Outcomes Measures in Rheumatology outside the submitted work. Dr Degenhardt reports receipt of grants from Mundipharma, Indivior, and Seqirus outside the submitted work. Dr Bell reports receipt of grants from the US Environmental Protection Agency and the National Institutes of Health (NIH) during the conduct of the study, and honorarium and/or travel reimbursement from NIH, the American Journal of Public Health, Columbia University, Washington University, Statistical Methods and Analysis of Environmental Health Data Workshop, North Carolina State University, and Global Research Laboratory and Seoul National University. Dr Mozaffarian reports receipt of personal fees from Acasti Pharma, GOED, DSM, Haas Avocado Board, Nutrition Impact, Pollock Communications, Boston Heart Diagnostics, and Bunge; other for serving on a scientific advisory board from Omada Health and Elysium Health; a grant from NIH and the National Heart, Lung, and Blood Institute; a grant from the Bill and Melinda Gates Foundation; and royalties from UpToDate outside the submitted work.
Funding/Support: This study was supported in part by the Intramural Program of the National Institutes of Health, the National Institute of Environmental Health Sciences, and in part by the Bill and Melinda Gates Foundation.
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.
Additional Contribution: We thank Ashkan Afshin for assistance with revisions to the risk factor section of this article.
Create a personal account or sign in to: