Introduction and Overview
Eileen M. Crimmins, Samuel H. Preston, and Barney Cohen
According to the United Nations (UN) Population Division, life expectancy at birth in the United States in 1950 for males and females combined was 68.9 years (United Nations, 2009). At that time, relative to other countries or territories for which the United Nations collects and publishes data, the United States had the 12th highest life expectancy at birth in the world. Since then, life expectancy at birth in the United States has increased by slightly more than 10 years, to 79.2 years, a remarkable achievement. Yet during the same time period, many other countries around the world have done even better. If one were to redo the analysis using the most recently available data, life expectancy at birth in the United States would be tied for 28th place, just behind Korea, Luxembourg, Malta, and the United Kingdom, and more than 2 years behind Australia, Canada, France, Iceland, Italy, Japan, and Switzerland (United Nations, 2009). The decline in the relative position of the United States cannot be easily explained by higher rates of infant mortality in the United States than in other developed countries or by higher rates of violent deaths among young adults in the United States. Although both phenomena are evident, the vast majority of Americans (94 percent) survive to at least age 50 and when one compares international levels of life expectancy only from age 50 onward, the United States still ranks only 29th in the world, behind a surprisingly long list of other countries (see Chapter 9).
What are the reasons for the relatively poor performance of the United States at older ages? Are Americans too fat? Too stressed? Is the nation’s much maligned health care system to blame? Or are there other factors that can explain the country’s relatively low ranking in life expectancy?
Motivated by such questions and concerns, the National Institute on Aging (NIA) requested that the National Research Council (NRC) launch a major investigation to clarify patterns in the levels and trends in international differences in life expectancy above age 50 and to identify strategic opportunities for health-related interventions. NIA was also interested in the identification of areas for future high-priority research.
Responding to this request, the NRC appointed a panel of experts to prepare a report clarifying the state of scientific knowledge in this area. In addressing its charge, the Panel on Understanding Divergent Trends in Longevity in High-Income Countries confronted a large and burgeoning theoretical and empirical literature with contributions from virtually every field within the social and health sciences. In order to make sense of the vast amount of work, the panel decided to commission a set of background papers, each dealing with a topic relevant to the panel’s work. This volume contains those papers. The panel’s report, Explaining Divergent Levels of Longevity in High-Income Countries, is being published separately.
Taken collectively, the papers in this volume provide an assessment of the plausibility of the most obvious possible explanations that have been advanced to explain the poor position of the United States in terms of life expectancy above age 50. The authors, all of whom are at the forefront of work in their fields, provide state-of-the-art assessments of the research and identify gaps in measurement, data, theory, and research design where they exist.
For some topics, there is surprisingly little direct evidence that can address the basic question. A necessary prerequisite for investigating the importance of any potential explanation of differences in levels and trends in mortality between countries is the ability to examine comparable country-level information on the potential explanatory variables under consideration. Without such information it would be difficult, if not impossible, to draw conclusions with any degree of confidence. Fortunately, thanks to the HRS (the Health and Retirement Study) in the United States, ELSA (the English Longitudinal Study of Ageing) in the United Kingdom, and SHARE (the Survey of Health, Ageing and Retirement in Europe) across Europe and Israel, there are now comparable large-scale international surveys that contain important measures of many variables of relevance. However, the empirical basis for certain conclusions is significantly stronger in some cases than in others. For example, a lot is known about international differences in smoking patterns and levels of obesity, but far less about international differences in stress, physical exercise, and social networks.
The papers in this volume offer a wide variety of disciplinary and scholarly perspectives. Many different disciplines have made theoretical and empirical contributions to the study of mortality. The current collection is to some extent an amalgamation of concepts and insights—both old
and new—obtained from various disciplines, each with its own domain of interest and style of analyzing and presenting data. Some authors review research fields that use mature methodologies and standard approaches, while others report on new avenues of investigation that are in their infancies. In these latter cases, concepts, methods, and measures still need to be refined. Nevertheless, each of the papers in this volume conveys important ideas and information.
NATURE OF THE DIFFERENCES
To better understand some of the main features of the diverging trends in life expectancy across countries, the paper by Glei, Meslé, and Vallin (Chapter 2) examines mortality changes and differences in 10 countries where high-quality mortality and cause of death data are available. In some of them, life expectancy has increased rapidly in recent years; in others, progress is lagging as in the United States.
By basing their analysis on a solid foundation of high-quality statistics, the authors are able to explore a number of important empirical relationships and see whether they stand up to close scrutiny. They point out that the story for male life expectancy at age 50 (e50) is somewhat different than the story for female life expectancy at age 50. For the 10 countries examined, U.S. males have consistently ranked among the lowest in terms of e50. Consequently, even though they currently appear to be faring relatively poorly, the relative position of U.S. males has not deteriorated over the last 50 years.
In contrast, the relative rank of U.S. females has deteriorated over the last 30 years. Around 1980, the pace of gains in life expectancy at age 50 slowed among women in the United States as it did for women in Denmark and the Netherlands; for the other countries, the pace of gains increased. Consequently, over the last quarter-century, gains in e50 among U.S. women (2.4 years) were about half those in Australia, France, and Italy (4.5-5.2 years) and less than 40 percent of that of Japan (6.3). The authors identify similar important empirical relationships by examining the contributions to gains in e50 by age and sex over time.
The authors provide a careful examination of cause-of-death statistics for those countries for which detailed data are available. The purpose of the analysis is to identify particular causes of death that can explain the relatively poor performance in gains in e50 for the three countries with the least amount of progress, Denmark, the Netherlands, and the United States. Comparative analysis of cause of death is complicated by issues of variation in coding practices across countries and over time. Nevertheless, the authors are partly successful in being able to identify particular causes of death that are either contributing factors or that can be ruled out. And
although it is difficult to do justice to such careful analysis in one or two sentences, it appears that differences in female mortality due to lung cancer and respiratory diseases are an important part of the story. Such a finding is clearly consistent with the hypothesis that smoking was an important factor in slowing the mortality decline among women in those three countries.
In Chapter 3, Crimmins, Garcia, and Kim consider international patterns of morbidity and disability. These patterns shed a good deal of light on the factors that may be underlying mortality differences. The paper demonstrates that, in general, people age 50 and above in the United States have higher levels of self-reported disease and disability than those in the other countries investigated. Unusually high levels of prevalence in the United States are recorded for heart disease, stroke, and diabetes. Cancer registries show that the reported incidence of prostate cancer, breast cancer, and lung cancer is also highest in the United States. Colorectal cancer is the only disease for which the United States does not rank first in reported morbidity among the countries in the analysis. The United States also ranks first in self-reported diagnoses of hypertension and high blood cholesterol levels. On the other hand, it ranks at or near the bottom in measured hypertension and high blood cholesterol. A likely explanation of this apparent paradox is that the proportion of the population age 50 and above taking drugs to control hypertension and high cholesterol is highest in the United States.
As the authors point out, the higher prevalence of morbidity in the United States is consistent either with a higher incidence of disease or with a higher level of post-diagnostic survival. A higher reported incidence of disease could be produced by a higher true incidence or by more awareness of disease on the part of physicians and patients in the United States. Because the data systems that make possible these international comparisons are very new, they cannot yet support the longitudinal studies needed to sort out these issues of causality. Comparisons of morbidity-to-mortality patterns in this paper provide some insight, but the small number of countries involved makes it very difficult to identify relationships that are statistically significant. The high level of morbidity from major conditions in the United States is consistent with the adverse longevity of the United States. Given the location of the United States on these distributions, most of the cross-national relations reported in the chapter between morbidity and mortality are positive: higher morbidity is associated with higher mortality. Japan is often at the opposite end of both the morbidity and mortality distributions from the United States, contributing to the positive association. Finally, the authors analyze micro-level data on self-reported disease and show that the poor ranking of the United States in heart disease, stroke, and diabetes is maintained even after controlling for different levels of obesity and smoking.
The next five papers in the volume (Chapters 4-8) summarize what is known about some of the main behavioral health factors that are commonly believed to contribute to the observed international differences in life expectancy. As many of the authors point out in their papers, the search for internationally comparable data is often a demanding task.
Through the 1960s, the United States had much higher per capita tobacco consumption than any country in Western Europe (Forey et al., 2002), so investigating the impact of differential levels of smoking was an obvious candidate for the panel to investigate. The adverse health consequences of smoking have been known for more than 50 years: smoking harms almost every internal organ and increases the risk of dying from many different causes of death. Smoking is not only associated with cancer of directly exposed organs and tissue (esophagus, larynx, lung and bronchus, mouth, and throat), it is also associated with a range of cancers in indirectly exposed organs and tissues, including the bladder, brain, intestines, kidney, liver, pancreas, rectum, stomach, and uterus. In addition, it has been linked to a host of other respiratory diseases (asthma, bronchitis, emphysema, influenza, pneumonia, pulmonary fibrosis, and pulmonary tuberculosis), cardiovascular diseases (aortic aneurysms, cerebral vascular disease, coronary heart disease, and hypertension), and others. Even so, the full impact of smoking in many of the countries under consideration is still not fully understood because there have been few studies that contain large enough numbers of representative smokers and nonsmokers who are followed over a sufficiently long period of time to calculate definitive statistics.
In Chapter 4, Preston, Glei, and Wilmoth apply a new method for estimating the portion of total mortality attributable to smoking. Using the death rate from lung cancer as an indirect measure of smoking histories, the authors use macro-level statistical relationships to model the impact of smoking on mortality (see Preston, Glei, and Wilmoth, 2009). Their method is conceptually different from the well-known Peto-Lopez model (Peto et al., 1992) yet reaches remarkably similar conclusions with respect to the impact of smoking on mortality. The authors find that male mortality has been much more heavily influenced by smoking than female mortality but that the attributable fraction for women has been rising more rapidly. In 2003, the highest percentage of male deaths attributable to smoking occurred in Hungary (30 percent); among women, the highest fraction occurred in the United States (20 percent).
Life expectancy at age 50 has been powerfully influenced by smoking in many countries. In the United States, Preston, Glei, and Wilmoth (2009) estimate that male e50 would be 2.5 years longer if the smoking-attributable deaths were eliminated, female e50 would be 2.3 years longer. Among the 21 countries that the authors examined, if one were to remove the deaths that
are attributable to smoking, the relative ranking of the United States with regard to life expectancy would improve from 17th to 9th for women and from 15th to 12th for men. These results suggest that increases in smoking-attributable mortality have dampened the gains to e50 since 1950 among women in all 21 countries, although the impact varied: Among U.S. women, smoking reduced the gains by 1.6 years, the largest effect in any country, while the effect of smoking-attributable deaths on life expectancy in Spain and Portugal was negligible. The authors conclude that about two-thirds of the growing shortfall in life expectancy for U.S. women since 1950 can be attributed to the effects of smoking. For U.S. men since 1950, smoking has produced a modest deterioration in their position in international comparisons of life expectancy.
Given the centrality of smoking as an underlying cause of death, Pampel (Chapter 5) investigates the reasons behind the observed divergent patterns of smoking across high-income nations. Levels of smoking in 2000 varied widely between countries: from 19 percent in Sweden and the United States to 34 percent in Spain, 35 percent in Germany, and 38 percent in Greece (Cutler and Glaeser, 2006). Even larger differentials can be observed if the comparisons are restricted to males.
Pampel explores potential explanations for the current level of smoking in the United States relative to other high-income countries. In the past, researchers have stressed such factors as prices, policies, inequality, and national-level differences in beliefs about the harmfulness of tobacco (see, e.g., Cutler and Glaeser, 2006); in contrast, Pampel explores the hypothesis that international differences in smoking can best be understood from the vantage point of an epidemic that spreads from a relatively small part of a population to other parts, and then recedes, like other epidemics.
Pampel emphasizes the importance of diffusion theory to explain observed patterns of cigarette consumption by socioeconomic group. In the early stages of the epidemic, smoking emerges initially among the highest socioeconomic group. This group is most open to innovation and has the financial resources to afford to smoke. The epidemic then diffuses to lower socioeconomic groups, and it recedes first among men of high socioeconomic status. Pampel finds that cross-national comparisons of aggregate trends in prevalence and determinants of individual differences in smoking generally support the epidemic or diffusion model. A better understanding of these cross-national patterns of cigarette smoking may have important implications for researchers’ ability to project future mortality trajectories across countries.
It is well known that the prevalence of obesity has increased very dramatically in the United States since the 1970s, affecting all sex, race, and socioeconomic groups (Flegal et al., 2010). Because obesity is associated with a wide variety of chronic conditions, disability, and mortality, its rapid
growth in the United States, perhaps combined with a car-dominated sedentary lifestyle, is popularly perceived as a large part of the reason that people in the United States fair so poorly in cross-national health comparisons.
In Chapter 6, Alley, Lloyd, and Shardell address this popular perception by examining both international trends in obesity and the relationship between obesity and mortality. The authors conclude that although high levels of obesity in the United States are likely to be part of the explanation, they are unlikely to account for a very large fraction of the cross-national differences in life expectancy. The authors offer two reasons for reaching this conclusion. First, the obesity epidemic is not confined to the United States: rising levels of obesity are occurring in many other countries, although admittedly they lag somewhat behind the United States. Second, the association between obesity (or high body mass index) and mortality is not straightforward and relatively weak at older ages. In fact, there is a strong relationship only between mortality and very high weight levels (morbid obesity), the prevalence of which remains relatively low even in the United States. Nevertheless, the authors conclude that the importance of rising levels of obesity as a contributing factor to life expectancy is still not fully understood and likely to grow over time as obesity increases at younger ages and at higher weights.
Maintaining a certain level of physical activity as one ages is important for a variety of reasons, including the maintenance of good cardiovascular health, lower risk of falls and fractures, higher levels of cognition and positive well-being, and higher levels of social participation. Conversely, physical inactivity has been related to higher rates of mortality, to a lower quality of life, and to a higher risk of coronary heart disease, diabetes, fractures, hypertension, obesity, osteoporosis, various types of cancers, and more.
In Chapter 7, Steptoe and Wikman assess the evidence that national-level differences in physical activity contribute to observed variation in life expectancy across high-income countries. An accurate assessment of the extent to which physical activity contributes to variations in life expectancy is hard to achieve. Among other obstacles are a paucity of internationally comparable time-series data, a lack of common metrics, and questions about the relative quality of personal recall data versus data derived by objective measurement. In addition, there is no definitive theoretical framework to guide how to assess the effects of physical activity on life expectancy: this lack of a framework results in a lack of clarity about the most important variables and when and how to measure them. For example, should relatively more weight be placed on evidence about attaining the recommended level of physical activity currently or in the past? How important is a completely sedentary lifestyle? Even if these issues can be resolved, the links between physical exercise and other behaviors, such as smoking and diet, imply that multivariate analyses are necessary.
Following a review of the strengths and limitations of self-report and objective measures of physical activity, the authors present data from four different internationally comparable data sets that have recorded the frequency, duration, and intensity of physical activity. They show that the ranking of countries in terms of physical activity is only moderately consistent across studies, implying that conclusions regarding the relationship between physical activity and health outcomes must be drawn cautiously. Using data from the ELSA, HRS, and SHARE, the authors analyze the relationship between various measures of physical activity and inactivity and self-reported health and self-reported diabetes. Their results provide important, albeit rather preliminary, evidence that is consistent with the notion that physical activity contributes to cross-national variations in health. More definitive conclusions must wait for more sophisticated cross-national comparisons that use objective measures of physical activity and multivariate analyses of time trends in physical activity.
The final paper in this section, by Banks and his coauthors (Chapter 8), considers the possibility that international differences in the degree of social integration can account for international differences in health and mortality. It focuses on comparisons between England and the United States because of closely comparable and detailed longitudinal surveys that were conducted in the two countries. The authors recognize that this comparison is not ideal because the two countries share relatively similar mortality profiles.
Their analysis of the relationship between measures of social integration and health in the two countries suggests a relatively weak role for social integration in explaining national differences. Not only are measures of social integration quite similar in England and the United States, but also is the “toxicity” of different measures in the two countries. Although the authors show the significance of many cross-sectional relationships between social integration and health outcomes, the analysis of mortality shows relatively small or inconsistent links between mortality and social integration or network measures. The authors then use Gallup survey data to demonstrate that the international variation in measures of social integration is much greater than that between England and the United States, leaving open the possibility that social relations may play a larger role on other stages. The paper concludes with a set of thoughtful observations on how research in the area can be advanced.
THE U.S. HEALTH CARE SYSTEM
The low ranking of the United States in international comparisons of life expectancy is sometimes blamed on the poor performance of the U.S. health care system rather than on behavioral or social factors. The United States spends more money on health care than any other country in the
world, yet the country suffers from a number of well-documented problems, including a high level of inefficiency and waste in the system; a large number of people who are uninsured or underinsured, with accompanying reductions in access to high-quality or at least good health care; and the existence of persistent disparities in health care associated with economic status, education, ethnicity, geography, and race.
In Chapter 9, Preston and Ho present evidence on the relative performance of the U.S. health care system using death avoidance as the sole criterion. As the authors point out, given that the United States has historically had high levels of cigarette consumption and obesity, it is certainly possible that the country’s low longevity ranking could be compatible with a finding that the U.S. health care system is performing relatively well, at least in identifying and administering treatments for various diseases. The authors find that, by standards of other high-income countries, the United States does well in terms of screening for cancer, survival rates from cancer, survival rates after heart attacks, and medication of individuals with high blood pressure and high levels of cholesterol.
The authors consider in greater depth mortality from prostate cancer and breast cancer, diseases for which effective methods of identification and treatment have been developed and for which behavioral factors do not play a dominant role. They show that mortality reductions from prostate cancer and breast cancer have been significantly more rapid in the United States than in other high-income countries. They argue that these unusually rapid declines are attributable to wider screening and more aggressive treatment of these diseases in the United States. On the basis of their review and their detailed consideration of these two diseases, they conclude that the low longevity ranking in the United States is not likely to be the result of medical failures in the identification or treatment of the major diseases at older ages.
One important feature of the erosion of the U.S. survival advantage is that it has been pronounced for U.S. women relative both to U.S. men and to women in other high-income nations. This pattern suggests the potential importance of gender-specific explanatory factors. In Chapter 10, Goldman discusses one attractively simple but as yet unexplored hypothesis: that the widespread use of postmenopausal hormone therapy (HT) in the United States has adversely affected U.S. mortality trends. At least prior to 2002, HT had been widely prescribed to U.S. women at menopause, not only for the relief of unpleasant symptoms (e.g., hot flashes), but also for its presumed protection against cardiovascular diseases and loss of bone density. However, the author finds little evidence to support the theory that HT use has had a notable impact on all-cause mortality and presents findings questioning whether HT is a significant risk factor for coronary heart disease. Finally, although high, the prevalence of HT in the United States is not out
of line with rates of use in several other high-income countries that have experienced steady improvements in female life expectancy.
The next two papers in the volume review the extent to which various types of inequality influence mortality differentials. In Chapter 11, Avendano and his coauthors explore the hypothesis that lower life expectancy at age 50 in the United States, relative to several Western European countries, may be partly attributable to larger socioeconomic disparities in mortality in the United States. To explore this hypothesis, the authors compare U.S. mortality rates by level of education with similar mortality data for 14 European countries.
They find that at low levels of education U.S. men have higher mortality than men in Western Europe; for highly educated men, those in the United States and several other countries had comparable rates. The pattern for women was slightly different: U.S. women had higher mortality than Western European women at all levels of education, but the U.S. excess mortality was often larger among women with low levels of education. However, most Eastern European countries had higher mortality rates than the United States, particularly at the bottom of the educational distribution. In general, disparities in mortality by education in the United States were comparable to disparities in several Western European countries, including France and Norway, but smaller than inequalities in Eastern European countries. The authors conclude that a modest part of the difference between U.S. and European mortality rates for women is attributable to larger excess mortality at lower educational levels.
In Chapter 12, Wilmoth, Boe, and Barbieri consider how geographic differentials in life expectancy at age 50 have evolved in Europe, Japan, and the United States, using a variety of indicators of regional disparities. The authors consider states and counties in the United States and Europe as a whole, with and without Eastern Europe. They also study changes in the evolution of internal geographic disparities in Canada, France, Germany, and Japan. Their analysis adds valuable texture to the analysis in the rest of this volume, which is heavily focused on measures expressed as means.
One of their most informative analyses asks how different the changes in U.S. life expectancy would have been if the pace of change in the bottom half of the geographic distribution had been the same as that in the top half. Even though the United States was the only country that had a growing disparity between the top and the bottom halves of the distributions since 1980, the authors find that the growing disparity contributed little to the poor performance of U.S. women in terms of mortality: both halves of the distribution lagged relative to their European counterparts. Addi-
tional and dramatic evidence of the widespread difficulties of U.S. women is suggested by the fact that the life expectancy of the highest quintile of American women has been below that of the lowest quintile of Japanese women since 1980.
INTERNATIONAL CASE STUDIES
The United States is not the only country to have experienced a slowing in improvements in life expectancy at the oldest ages. Denmark and the Netherlands have recently experienced slowdowns in mortality decline comparable to that of the United States (Meslé and Vallin, 2006). Interestingly, however, progress in mortality decline among the elderly resumed in Denmark around 1995 and resumed in the Netherlands around 2002. In Chapter 13, Mackenbach and Garssen investigate the case of the Netherlands, searching for clues about what might account for the observed trend in mortality. A slight upturn in life expectancy at older ages was first observed in the Netherlands around 2002 and initially attributed to favorable climatic factors (milder than average winters, cooler summers). But when mortality decline continued, it became increasingly unlikely that milder temperatures could be the sole driving force.
In order to investigate this phenomenon in more detail, Mackenbach and Garssen examined evidence on causes of death. The authors find that the main contributors to the acceleration of the rise in life expectancy at age 65 were significant reductions in death from ill-defined conditions, stroke, diabetes, dementia, and pneumonia. The authors review a wide range of possible determinants of mortality to try to explain these patterns. Health care is the only category of determinants for which substantial changes appear to have occurred and for which changes are consistent with the observed pattern in mortality and changes in cause of death. A deliberate and sudden rise in health care expenditures around 2001 seems to have resulted in an increase in availability of health care for the elderly. A rapid increase in hospitalization rates and more liberal administration of life-saving treatments to elderly people appears to be the most plausible hypothesis for explaining the sudden reversal of old-age mortality trends in the Netherlands. This paper serves to reinforce the importance of examining differential access to and the quality of health care provided to older people.
Finally, Chapter 14 focuses on a more specific comparison of Denmark, one of the countries that the panel singled out as a life-expectancy laggard, to neighboring Sweden. Christensen and his colleagues show that Denmark’s life expectancy at birth dropped from 3rd highest among 20 European countries in the 1950s to 17th for males and 20th for females around 2000. The deterioration stopped in the mid-1990s but no catch-up occurred. Their analysis of cause-specific mortality data suggests that the
reason for the Danish deterioration was lifestyle factors, especially smoking and high alcohol consumption. The authors conclude that smoking and alcohol-related deaths accounted for virtually all of the disparity in life expectancy between Denmark and Sweden in 1997-2001, with smoking playing the larger role. There are also some indications that lower budgets for Denmark’s free national health care system, in comparison with other Nordic countries, may play a role in Denmark’s adverse position.
THE WAY AHEAD
Clearly, there is a need to continue to conduct research to better understand the factors underlying international differences in life expectancy at older ages. For the most part, the papers in this volume focus on the behavioral factors that are commonly believed to contribute to those differences. Because of the interaction and the multiple causal pathways between these various factors (e.g., obesity can lead to lack of physical exercise and poor health but poor health can also lead to lack of physical exercise and obesity), the exact amount that each factor contributes to the observed health differentials remains unknown. Yet one finding seems clear: having the highest level of cigarette consumption per capita in the developed world over a 40-year period (up to the mid-1980s) has left a very visible and continuing imprint on U.S. mortality.
The papers in this volume should be considered starting points. Although some questions have been answered, many others remain. Major advances in data collection have meant that high-quality cross-national research is becoming increasingly feasible. This area of research has already produced important insights, and it seems clear that its future is promising.
Cutler, D.M., and Glaeser, E.L. (2006). Why Do Europeans Smoke More Than Americans? NBER Working Paper 12124. Cambridge, MA: National Bureau of Economic Research.
Flegal, K.M., Carroll, M.D., Ogden, C.L., and Curtin, L.R. (2010). Prevalence and trends in obesity among U.S. adults, 1999-2008. Journal of the American Medical Association, 303(3), 235-241.
Forey, B., Hamling, J., Lee, P., and Wald, N. (Eds.). (2002). International Smoking Statistics (2nd edition). Oxford: Oxford University Press.
Gruer, L., Hart, C.L., Gordon, D.S., and Watt, G.C.M. (2009). Effect of tobacco smoking on survival of men and women by social position: A 28-year cohort study. British Medical Journal, 338(172), b480.
Meslé, F., and Vallin, J. (2006). Diverging trends in female old-age mortality: The United States and the Netherlands versus France and Japan. Population and Development Review, 32(1), 123-145.
Mokdad, A.H., Marks, J.S., Stroup, D.F., and Gerberding, J.L. (2004). Actual causes of death in the United States, 2000. Journal of the American Medical Association, 291(10), 1238-1245.
Peto, R., Lopez, A.D., Boreham, J., Thun, M., and Heath, C. (1992). Mortality from tobacco in developed countries: Indirect estimation from national vital statistics. Lancet, 339(8804), 1268-1278.
Preston, S.H., Elo, I.T., Hill, M.E., and Rosenwaike, I. (2003). The Demography of African Americans, 1930-1990. Dordrecht, The Netherlands: Kluwer Academic.
Preston, S.H., Glei, D.A., and Wilmoth, J.R. (2009). A new method for estimating smoking-attributable mortality in high-income countries. International Journal of Epidemiology, doi:10.1093/ije/dyp360.
Rogers, R.G., Hummer, R.A., Krueger, P.M., and Pampel, F.C. (2005). Mortality attributable to cigarette smoking in the United States. Population and Development Review, 31(2), 259-292.
United Nations. (2009). World Population Prospects: The 2008 Revision. Volume 1: Comprehensive Tables. New York: Author.