Author + information
- Published online June 26, 2017.
- Dariush Mozaffarian, MD, DrPH∗ ()
- ↵∗Address for correspondence:
Dr. Dariush Mozaffarian, Tufts University, Friedman School of Nutrition Science and Policy, 150 Harrison Avenue, Boston, Massachusetts 02111.
Cardiovascular disease (CVD) is known to be the leading cause of death worldwide. Yet, global trends in different CVD subtypes, nonfatal CVD events, and corresponding morbidity are less well established. In this issue of the Journal, Roth et al. (1) estimate the global, regional, and national burdens of fatal and nonfatal CVD, overall and by 10 major subtypes, between 1990 and 2015. They also model the years lived with disability, and disability-adjusted life-years (DALYs) lost, due to CVD.
Several findings are notable. In 2015, 19.9 million CVD deaths occurred (one-third of all global deaths), and 423 million people had prevalent CVD (∼1 in 17 of the global population). In contrast to the conventional wisdom that CVD remains mainly a condition of wealthy nations, Roth et al. (1) found that, adjusted for age, far more cases of CVD are now occurring in countries with the lowest sociodemographic levels than with highest levels; with most CVD at middle sociodemographic levels in men and at middle and low sociodemographic levels in women. These findings confirm that the epidemiological “transition” away from infectious and maternal-child diseases and toward noncommunicable chronic diseases has already occurred globally—a sobering reality as countries around the world consider their priorities for health care, public health prevention, and economic growth.
For instance, estimated age-standardized CVD prevalence was highest in certain African and Middle Eastern nations; and lowest in several high-income Asian, South American, and Western nations. Among CVD subtypes, estimated ischemic heart disease mortality was highest in Central Asia and Eastern Europe, and lowest in high-income Asia Pacific nations (e.g., Japan). Estimated stroke mortality was highest in Oceania and central Sub-Saharan Africa. Between 1990 and 2015, estimated age-standardized CVD mortality remained relatively stable in Sub-Saharan Africa and Southeast Asia, and increased in Bangladesh and the Philippines. In contrast, significant declines occurred in all high- and some middle-income countries. In high-income Western nations, this decline appeared to plateau in more recent years—perhaps a harbinger of the advancing harms of the obesity and diabetes epidemics in these nations.
Other relevant findings included the burdens of hemorrhagic stroke, which produced similar or larger DALYs than ischemic stroke in all nations except in North America and Central and Eastern Europe. Also, global prevalence of atrial fibrillation was estimated at 33.3 million cases, causing nearly 200,000 estimated deaths per year.
Although these findings are striking, they represent modelled results. Given the scope of evaluated data and the complexity of approaches for their incorporation, several aspects of the methodology deserve close consideration. These include: 1) low data availability for many endpoints; and 2) the use of several modelling assumptions that were not incorporated into the statistical uncertainty of the estimated results (Table 1). For many major endpoints, a minority of global geographies had available data, including for conditions such as ischemic heart disease, atrial fibrillation, and many others. Little data (∼1% of global geographies) were available for “other” CVD, which nonetheless was estimated to represent 25% of prevalent CVD cases in sub-Saharan Africa.
In addition, to derive their estimates of the prevalence, mortality, and morbidity from total CVD and CVD subtypes globally, the authors made several major modelling assumptions (Table 1). Use of such assumptions is necessary and acceptable, based on the understandably limited, incomplete, and heterogeneous datasets available. Yet, the authors’ modelling methods include the implicit—and untenable—assumption that there is no uncertainty in each of these assumptions: essentially, that each is 100% accurate. In other words, the reported statistical uncertainly for each of the findings includes only sampling variability (e.g., based on sampled size of each dataset). Yet, each of the assumptions also introduces considerable uncertainty. For example, the authors redistributed death certificate–reported causes of deaths to other causes based on assumed errors in reporting, which would be inexact and vary across country and time. The incidence of ischemic heart disease before 2000 was uniformly upwards by 56% to account for troponin use in later years, but uptake and consistency of troponin usage would vary widely globally. Cases of heart failure were classified to a single assumed underlying cause; yet, the underlying cause of a patient’s heart failure can be notoriously difficult to determine even under the best of clinical circumstances.
For each of these and other major modelling decisions, the effects of differing assumptions or uncertainty in assumptions were not reported. The investigation would have been greatly strengthened by single and multiway probabilistic sensitivity analyses to quantify the influence of variations in and alternatives for each modelling decision. Without this, we have no way of knowing how these assumptions affect the final results—essentially, the modelling approach is a “black box.” Thus, for each finding, the validity of the central estimate should be considered uncertain, with a presented precision (uncertainty interval) that is likely to be substantially overestimated. The Global Burden of Disease study has the strengths of being comprehensive and regularly updated. As the complexity of the data sources and desired outputs has grown, new methods are needed to quantify and incorporate the uncertainty of the correspondingly complex and investigator-driven modelling decisions and assumptions.
What are the implications of the present findings? First, the careful effort to collect a variety of global datasets is important because it highlights specific gaps in knowledge for further surveillance efforts. Many nations and even world regions have limited information on some important causes of CVD, and the searches underlying this investigation can be considered a roadmap for future efforts to fill these data gaps. Second, even with the uncertain validity and overestimated precision described in the previous text, these results represent the best currently available estimates for many important CVD endpoints, including prevalence of and morbidity (DALYs) from specific nonfatal outcomes. Third, these findings provide clear confirmation that prevention of CVD can no longer be a priority of only wealthy nations. Even in low-income countries, CVD risk factors are widely prevalent in most urban areas and even many rural areas (2). Based on the present findings, middle-income nations are now facing an onslaught of CVD with tremendous corresponding health care costs for governments and businesses, lost workforce productivity, and personal and familial insolvency (3,4). With rising obesity and diabetes globally, these threats will only worsen.
The United Nations’ third Sustainable Development Goal includes a target to reduce premature mortality from noncommunicable diseases by one-third by 2030 (5). To achieve this aim, substantial health care system reform is needed in most countries to greatly increase rates of coverage, screening, and access to evidence-based cardiovascular treatments and preventive therapies, in particular for high blood pressure and dyslipidemia. An even larger population effect, together with greater cost-effectiveness or even cost-savings, must be achieved by effective policies and other systems innovations targeting lifestyle behaviors, especially smoking, suboptimal diet, and physical inactivity (6). Examples of relevant interventions include comprehensive antitobacco policies (7), government-supported strategies to reduce dietary sodium (8), and financial incentive and disincentive programs for industry and consumers to normalize costs of foods toward their true health and societal costs (9,10). Lifestyle-focused strategies to reduce CVD have additional cross-sectoral benefits of improving health and well-being for many other conditions, including lifestyle-related cancers, type 2 diabetes, obesity, and other related diseases. Both thoughtful health care efforts and precision population health must now be a priority for all nations to substantially reduce CVD by 2030.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Mozaffarian is an advisory board member for Omada Health. Deepak L. Bhatt, MD, MPH, served as Guest Editor-in-Chief for this paper.
- 2017 American College of Cardiology Foundation
- Roth G.A.,
- Johnson C.,
- Abajobir A.,
- et al.
- Jaspers L.,
- Colpani V.,
- Chaker L.,
- et al.
- ↵United Nations. Sustainable Development Goals: 17 Goals to Transform our World. 2015. Available at: http://www.un.org/sustainabledevelopment/. Accessed May 19, 2017.
- Mozaffarian D.,
- Afshin A.,
- Benowitz N.L.,
- et al.
- Webb M.,
- Fahimi S.,
- Singh G.M.,
- et al.
- Afshin A.,
- Penalvo J.L.,
- Del Gobbo L.,
- et al.