In 1993, Michael McGinnis and William Foege published work in the Journal of the American Medical Association on the actual causes of death in the United States (McGinnis and Foege, 1993). A key finding showed that about 40 percent, or nearly half of all deaths, could be explained by factors that were preventable given current knowledge, many of which were behavioral in nature. These preventable factors included tobacco, diet and activity patterns, alcohol, microbial agents, firearms, sexual behaviors, motor vehicles, and illicit drug use. McGinnis and Foege gathered these data from the literature and other sources using conservative rounded estimates or actual counts of factors when they existed. This seminal work garnered a great deal of attention and highlighted the significant proportion of deaths due to preventable factors. Although this work focused on all deaths, McGinnis explained at the 2013 workshop the importance of paying particular attention to deaths that are “clearly early,” both from an analytical as well as a societal perspective. Ultimately, McGinnis argued, understanding these early deaths, as well the causes of morbidity and loss of quality of life, are important to public health and policy. He stated, “Unfortunately, we are not spending our resources in a very effective or efficient manner if we want to improve the health of the population.”
Since that initial work in the early 1990s, Steven Schroeder published a provocative review paper (Schroeder, 2007) that focused on document-
ing and explaining the factors that explain the poor health of people in the United States, using international rankings of health status to assess the impact of five domains—genetics, social circumstances, environmental exposures, behavioral patterns, and health care shortfalls. Using the data assembled by McGinnis and Foege (1993), Schroeder found that “medical care played a relatively minor role” (p. 1222) in reducing early deaths, accounting for only 10 percent of premature deaths (p. 1225) and suggesting that “if the entire U.S. population had access to excellent medical care—which it does not—only a fraction of these deaths could be prevented” (p. 1222). The review concluded that “the greatest opportunity to improve health and reduce premature deaths lies in personal behavior” (p. 1222). The prevalence of causes of death associated with personal behaviors has been variously estimated. One study concluded that approximately 40 percent of all deaths in the United States are associated with four health behaviors: tobacco use, unhealthy diet, physical inactivity, and problem drinking (Mokdad et al., 2004, 2005).
More recent work employing a broad perspective that considered both downstream, proximate determinants of health and more upstream distal ones has informed several other major health studies, including the World Health Organization’s Commission on the Social Determinants of Health (2008), the “Marmot Review” of health in the United Kingdom (Marmot, 2010), and the Robert Wood Johnson Foundation’s Commission to Build a Healthier America in the United States (Braveman and Egerter, 2008).
Two recent National Research Council (NRC) and Institute of Medicine (IOM) reports have put these findings into sharp relief, reporting that the United States spends more on health care than any other country, but its health outcomes are “no better” than those of other rich nations. In fact, the situation is far worse: Americans are in poorer health and die earlier than people in other high-income countries (National Research Council, 2011; National Research Council and Institute of Medicine, 2013).
The 2011 NRC report International Differences in Mortality at Older Ages: Dimensions and Sources presented a descriptive analysis of causes of death but also moved from “description to identifying the underlying determinants of the observed differences, a necessary first step toward ultimately developing an integrated model of causal processes” (National Research Council, 2011, p. 143). The NRC Panel on Understanding Divergent Trends in Longevity in High-Income Countries, the group that conducted the analysis, examined a number of possible risk factors and considered how differences among countries in exposure to these risk factors might
account for observed disparities in improvement in life expectancy. For some factors, comparable cross-country data exist on the current levels of risk; for others, surprisingly little direct evidence can be brought to bear. The report concluded that few countries are conducting systematic surveillance of health risk factors, so directly comparable data even for the present often are not available for a large number of countries, and for a substantial number of countries, data are available for almost no risk factors for the 50-year period examined for this study. For example, much is known about current international differences in smoking patterns and levels of obesity, but far less about international differences in stress, physical exercise, or social networks. Moreover, very little is known about changes over time and across countries in lifetime exposures and behaviors for most risk factors. The dynamic nature of the relationship between mortality and some of the major risk factors is another important consideration.
The panel adopted the strategy of establishing the strength of the evidence for a number of the most commonly proffered explanations for differences in life expectancy between the United States and other high-income countries—for example, that these differences are the result of a particularly inefficient U.S. health care system or that they are a function of poor health behaviors in the United States, particularly with respect to smoking, overeating, and failing to exercise sufficiently. The panel also considered differences among countries in levels of social integration and in socioeconomic inequality. Ultimately, all of these potential risk factors will need to be examined in an integrated framework across the entire life course, taking account of the effects of differences in socioeconomic status, behavioral risk factors, and social policy, as well as effects across particular cohorts and periods.
The 2013 NRC/IOM report cited above, U.S. Health in International Perspective: Shorter Lives, Poorer Health, considered these issues over the entire life course, concluding that even U.S. newborns have a shorter life expectancy than those in other high-income countries and that premature death is not just associated with the diseases of aging. In attempting to account for this pervasive pattern, the NRC/IOM Panel on Understanding Cross-National Differences Among High-Income Countries found clues in almost every class of health determinants it considered, beginning with the obvious difference between health care in the United States and other countries—the lack of universal health insurance coverage. But the U.S. health care system has many other differences, such as a weaker foundation in primary care and greater barriers to access and affordable care.
The report narrowed in on individual behaviors and found that although Americans are currently less likely to smoke (due to successful tobacco control efforts) and may drink alcohol less heavily than adults in peer countries, they have a greater propensity for a variety of other unhealthy behaviors. As summarized in the report, Americans consume more calories per capita, abuse more drugs, are less likely to fasten seat belts, have more traffic accidents involving alcohol, and own more firearms than do people in other high-income countries.
The report also concluded that socioeconomic conditions matter greatly to health. Such factors as income inequality and high rates of relative poverty, relative lack of social mobility, and failure to keep pace with improving educational outcomes in peer countries and emerging economies were seen to play a role in producing less favorable health outcomes in the United States than in other countries (OECD, 2011). The NRC/IOM panel recognized that relevant scientific evidence was at once vast (e.g., for understanding the determinants of health) and scant (e.g., for establishing causality for the U.S. health disadvantage). It called for an international collaboration to collect prospective data and devise innovative study designs to pursue new lines of scientific inquiry that get at the roots of the U.S. health disadvantage.
One way to approach these issues is to examine recent data on the risks and causes of premature mortality, including the important scientific and technological advances affecting the prevalence of particular causes of premature mortality over the past two decades. At the workshop, Richard Suzman of the National Institute on Aging argued that despite the mounting research findings and the high-profile reports summarized above, public health and particularly the substantial contributions of social and behavioral factors to premature mortality have not received the attention they deserve.
The Committee on Population of the National Research Council at the National Academies conducted two workshops to address the data sources, science, and future research needs to understand the causes of premature mortality in the United States on behalf of the National Institute on Aging of the National Institutes of Health. The first workshop was held in September 2013, with a follow-up workshop held in September 2014. The purpose of the workshops, organized in cooperation with the Institute of Medicine, was to examine previous work in the field in light of newly
emerging data sources, particularly data generated as part of the work of the NRC Panel on Understanding Divergent Trends in Longevity in High-Income Countries (National Research Council, 2011) and the NRC/IOM Panel on Understanding Cross-National Differences Among High-Income Countries (National Research Council and Institute of Medicine, 2013).
Workshop participants considered the state of the science of measuring the determinants of the causes of premature death and, to a limited extent, the related constructs of disability and health loss. They discussed the availability and quality of data sources and identified future courses of action to improve the understanding of the causes of premature death. Presenters shared their approaches to and results of measuring premature mortality and specific risk factors, with a particular focus on those factors most amenable to improvement through public health policy.
The agendas were developed in accordance with the statement of task (see Box 1-1) and are included in Appendix A. Because a workshop is not a consensus activity, no recommendations or other consensus findings are
Statement of Task
An ad hoc steering committee will organize a public workshop on updating the classification of the determinants of the real causes of premature death in the United States. The purpose of the workshop, which will be organized in cooperation with the Institute of Medicine, will be to review previous work in the field in light of new data generated as part of the work of the NRC Panel on Understanding Divergent Trends in Longevity in High-Income Countries (National Research Council, 2011) and the NRC/IOM Panel on Understanding Cross-National Differences Among High-Income Countries (National Research Council and Institute of Medicine, 2013). The committee will develop an agenda, select and invite speakers and discussants, and moderate the discussions. The workshop will feature presentations that will consider the state of the science of measuring the determinants of the causes of premature death, assess the availability and quality of data sources, and chart future courses of action to improve the understanding of the causes of premature death. The products of the workshop will be a verbatim transcript in 508-compliant format and a published individually authored workshop summary prepared by a designated rapporteur.
offered in this report. This report offers a summary of the views expressed during the workshop meetings; therefore, all views expressed are those of the workshop presenters or other workshop attendees.
This report summarizes the presentations and discussions at both the 2013 and 2014 workshops. Chapter 2 presents a summary of the data sources and a description of the methodological approaches and issues that presenters described in their work on premature mortality. Chapter 3 summarizes presenters’ current data from large international studies of premature mortality or sources of variation in mortality and risk within the United States. Finally, Chapter 4 offers a look toward the future with a summary of suggested improvements offered by some participants for data sources and methodology and for next steps for research.