Professor Jessica Y. Ho
Is Cardiovascular Disease Slowing Improvements in Life Expectancy?
OECD and The King's Fund Workshop Proceedings
3. Causes of gains and losses in life expectancy in OECD countries
Abstract
Cardiovascular disease is an important contributor to the slowing life expectancy improvements seen in some countries. This cause of death is strongly linked to behavioural factors operating over the life course including smoking, obesity, diet, and physical inactivity, as well as access to and the quality of health care. Going forward, it is very important that we maintain the quality and timeliness of our vital registration systems and timely data releases. For most countries, there is an at least four‑year lag between the most recent cause-specific mortality data becoming available and the present year. These data are essential to identifying contemporary trends in cardiovascular disease mortality and their underlying drivers. Understanding the burden and causes of trends in cardiovascular disease mortality is further complicated by the linkages between cardiovascular disease and other causes of death, such as influenza and pneumonia, and drug overdose.
Life expectancy is one of the most commonly used summary measures of a country’s health and well-being. With few exceptions, life expectancy at birth has increased worldwide since the 1950s, with many low- and middle-income countries achieving impressive gains. In high-income countries, life expectancy increases have generally been smaller in magnitude, in part because they have already attained relatively high life expectancy levels. As most high-income countries have reduced infant and child mortality to very low levels, their life expectancy increases in recent decades have largely been driven by mortality improvements at the older adult ages and were fairly robust through the decade of the 2000s.
Since 2010, alarming trends in mortality have been observed for two high-income countries – the United States and the United Kingdom. Life expectancy at birth has declined for three consecutive years between 2014 and 2017 in the United States (Kochanek et al., 2016[1]; Xu et al., 2016[2]; Murphy et al., 2017[3]; Public Health England, 2018[4]), and improvements in life expectancy have virtually stalled in the United Kingdom in recent years (Public Health England, 2018[4]). Ho and Hendi (2018[5]) found that between 2014 and 2015, the majority of a set of 18 Organisation for Economic Cooperation and Development (OECD) countries experienced life expectancy declines for both men (11 countries) and women (12 countries). In most of these countries, the declines were related to increases in mortality at older ages (above age 65) following a particularly bad influenza year, and life expectancy recovered to its expected levels in the following year. The United States was an exception, with its declines strongly driven by increases in mortality at younger ages (below age 65) and drug overdose in connection with its ongoing opioid epidemic. Following the 2014‑15 declines, life expectancy continued to decline in the United States. In the United Kingdom, life expectancy increased in 2016 and 2017, but showed no further gains between 2017 and 2018 for both men and women (Public Health England, 2019[6]). Raleigh (2019[7]) compared life expectancy trends in two periods, 2011‑16 and 2006‑11, for selected European Union (EU) and OECD countries. This report documented a slowdown in life expectancy improvements among several of these countries and suggested that contributory factors to these trends may include slower reductions in cardiovascular disease (CVD) mortality and increases in mortality from dementia and Alzheimer’s disease, and in some countries, drug overdose.
This paper examines changes in life expectancy among a set of 32 OECD countries for which recent mortality data by cause of death are available. It aims to provide a comprehensive assessment of changes in life expectancy since 2010 and whether these trends have been similar or different across this set of countries. Trends between 2010 and the most recent year available for each country (ranging between 2014 and 2017) are compared to trends between 2000 and 2010. To account for differences in the observation window across countries, these changes are annualised for cross-country comparability. Data were sourced from the Human Mortality Database, the World Health Organization Mortality Database, and several countries’ vital statistics agencies (Finland, Italy, Norway, and Portugal) in order to provide coverage of the most recent mortality trends possible.
Recent trends in life expectancy at birth in OECD countries
Figure 3.1 shows the annualised change in life expectancy at birth since 2010 plotted against the life expectancy level in 2010 for each of the 32 OECD countries. Each circle corresponds to men or women in each country, and the size of the circle is proportional to the country’s male or female population size. The further a country lies above the x-axis, the larger the gains in life expectancy it has experienced since 2010. While some countries have made robust improvements, many are clustered very close to the x-axis.
Figure 3.1. Gains in life expectancy at birth since 2010 and levels of life expectancy at birth in 2010, men and women, 32 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
Figure 3.2. Gains in life expectancy at birth between 2000 and 2010 and levels of life expectancy at birth in 2010, men and women, 32 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
Figure 3.2 shows the life expectancy improvements these countries experienced in the preceding decade (i.e. the annualised change in life expectancy at birth between 2000 and 2010). In this figure, all the countries are located at much greater distances above the x-axis.
Figure 3.3 provides a more detailed look at the life expectancy trends since 2010 separately for men and women. The eight Central and Eastern European countries (Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovak Republic, and Slovenia) rank among the best performing countries for both men and women. However, these countries started out at considerably lower baseline levels of life expectancy. The bulk of this chapter focuses on the remaining 24 OECD countries, which have more similar mortality levels and histories. We can classify the remaining countries into three groups of eight: High, which consists of countries making the largest gains in life expectancy since 2010; Medium, which consists of countries making mid-sized gains; and Low, which consists of countries making the smallest gains since 2010. Most of these countries fall into the same groups for men and women, but for ten of the countries, men and women are classified into different (although adjacent) groups. The country groupings, and the average gains in life expectancy for each group, are shown in Table1.1 in Chapter 1. While the number and exact composition of these groups is somewhat arbitrary, the essence of the main point is clear – some countries have made fairly strong gains in life expectancy since 2010, while others have made almost none.
Figure 3.3. Gains in life expectancy at birth since 2010 and levels of life expectancy at birth in 2010, men and women, 32 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
For example, men in the countries in the Low group gained 0.12 years on average since 2010, compared to 0.27 years on average between 2000 and 2010. Countries in the Medium group gained 0.20 and 0.28 years since 2010 and between 2000 and 2010, respectively. In contrast, countries in the High group performed just as well since 2010 as in the prior decade (0.30 years in each period). For women, the pattern is highly similar, but the magnitudes of the average gains in both periods and all three country groups are somewhat smaller in size compared with men.
What is particularly striking is that Japan and Switzerland, two of the countries in the High group, had very high levels of life expectancy to begin with, consistently ranking among the world leaders in life expectancy. This would suggest that these countries are not approaching a limit to life expectancy and that stagnation in mortality improvements is not concentrated among countries with little possibility of achieving higher life expectancy levels (in other words, the High group is not composed of the countries with the lowest life expectancy levels, and the Low group is not composed of the countries with the highest life expectancy levels). In fact, the United States, which is often ranked last in terms of life expectancy among its high-income peer countries (Crimmins, Preston and Cohen, 2011[8]; Ho, 2019[9]; Ho and Preston, 2010[10]; Woolf and Aron, 2013[11]), falls into the Low group for both men and women.
In the subsequent sections, we focus on making two sets of comparisons. The first is between groups in the most recent period (e.g. what differentiates High-performing countries from those that were Low- or Medium-performing?). The second is within groups across time (e.g. for countries in the Low group, which age groups and causes of death contribute most to slower gains since 2010 compared to a decade ago?).
Age group contributions
Figure 3.4 displays the contribution of broad age groups to gains in life expectancy at birth for men and women in each of the three groups of countries (Low, Medium, and High) since 2010. The longer the length of the bar, the more that age group is contributing to improvements in life expectancy. First, across the board – for all three groups and for both men and women – mortality reductions at ages 65‑84 have been the most important source of gains in life expectancy (accounting for 40‑50% of these gains), followed by mortality reductions at ages 45‑64. Second, greater reductions in mortality in these two age groups explain the majority (about 74‑78% among men and 63‑67% among women) of the High group’s larger gains compared to the Medium and Low groups. The rest of the difference was driven by younger age (i.e. below age 45) mortality among men (25%) and by older age (i.e. at ages 85+) mortality among women (about a tenth).
Figure 3.4. Contribution of age groups to gains in life expectancy at birth since 2010, men and women, 24 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
Figure 3.5 demonstrates the importance of mortality reductions in each age group for life expectancy gains since 2010 and between 2000 and 2010. Focusing on the Medium and Low groups, which both achieved much slower progress in the recent period, we see that the bars for the 65‑84 and 45‑64 age groups are much shorter in the recent period than they were in the earlier period. In other words, much of the reason why life expectancy gains have been slower since 2010 is that mortality reductions have been considerably smaller between ages 45‑84 for these two groups of countries.
Figure 3.5. Contribution of age groups to gains in life expectancy at birth between 2000 and 2010 and since 2010, men and women, 24 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
Cause of death contributions
Mortality rates for most causes of death have continued to decline since 2010. However, we have seen notable increases in mortality from a few causes of death: accidental poisoning (a key component of drug overdose), mental and nervous system diseases (including Alzheimer’s disease), and perinatal conditions. For two causes – respiratory diseases (other than influenza and pneumonia) and infectious diseases – mortality increased only among women. Mortality from influenza and pneumonia also increased among women after 2010 in 5 out of 8 countries in the Low group and 6 out of 8 countries in the Medium group, contributing to losses in life expectancy. These mortality increases are not limited to a select, small set of countries. For example, 23 out of the 24 countries experienced increases in mortality from mental and nervous system disorders, and 14 (58%) and 17 (71%) experienced increases in accidental poisoning death rates for men and women, respectively (Table 3.1).
Table 3.1. Countries (number and %) experiencing increases in mortality from specific causes of death since 2010, men and women, 24 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
Cardiovascular disease (CVD) has been implicated in the general slowdown in life expectancy improvements (Raleigh, 2019[7]) and particularly in the United Kingdom (Public Health England, 2018[4]; Office for National Statistics, 2018[12]; Steel et al., 2018[13]). Figure 3.6 shows trends in CVD mortality since 1995 among men and women in these 24 countries. Compared to about a quarter century ago, these countries have experienced enormous reductions in CVD mortality. On average, CVD mortality today is roughly half of what it was in 1995. However, it is important to note that the bulk of those reductions occurred before 2010. Countries that experienced particularly slow reductions in CVD mortality for both men and women since 2010 (relative to the rates experienced between 2000 and 2010) include the United States, Italy, Canada, Ireland, Austria, and the Netherlands.
Figure 3.6. CVD mortality, 1995 to most recent year, men and women, 24 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
This is more apparent in Table 3.2, which shows each country’s annualised rate of reduction in CVD mortality since 2010 compared to the prior decade. All countries achieved reductions in CVD mortality during both periods, but for nearly all of them (the exceptions being Japanese and Icelandic men and Greek women), progress was greater in the earlier than the most recent period. The larger the value in the Difference column, the greater the discrepancy in reducing CVD mortality between the two periods. These differences tended to be larger for the Low and Medium groups than for the High groups. For example, the average annualised rate of reduction in CVD mortality among men in the Low group was about 5.99 deaths per 100 000 greater on average between 2000 and 2010 than since 2010; this number was 3.59 for the Medium group and 3.33 for the High group. For women, the corresponding figures were 3.34 (Low), 3.39 (Medium), and 2.26 (High).
Table 3.2. Reductions in CVD mortality between 2000 and 2010, and since 2010, men and women, 24 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
CVD is one among several important cause of death categories. Next, we identify how several major causes of death have contributed to changes in life expectancy at birth using Arriaga’s decomposition (Arriaga E, 1989[14]). The results from this analysis are shown in Figure 3.7 and Figure 3.8. Figure 3.7 focuses on comparisons between groups in the most recent period, while Figure 3.8 make comparisons within groups across time. The length of each bar is the total average gain in life expectancy experienced by the countries in a given group, and each of its components is a cause of death category. Causes of death to the right of the zero line contributed to gains in life expectancy, while causes of death to the left of zero line contributed to losses in life expectancy.
Beginning with men (Figure 3.7), we see that for all groups, CVD made the largest contributions to life expectancy improvements, followed by cancer. Mental and nervous system disorders, on the other hand, contributed to life expectancy losses. The three causes of death contributing most to differences between the Medium and High groups were: cancer, accidental poisoning, and external causes. Countries in the High group made larger reductions in cancer and external cause mortality than countries in the Medium group. CVD, cancer, and external causes were the three causes contributing most to difference between the Low and High groups, with each accounting for about 20% of this difference. It is notable that countries in the High group continued to experience reductions in mortality from accidental poisoning, but countries in both the Low and Medium groups experienced mortality increases from this cause of death.
Among women, CVD was also the cause of death contributing most to gains in life expectancy, followed by cancer. Mental and nervous system disorders played an even more important role in the substantial slowdowns in life expectancy gains for women than they did for men. Cancer, other chronic diseases, and external causes were the top three contributors to the difference between the Medium and High groups. For the difference between the Low and High groups, these were CVD, cancer, and mental and nervous system disorders. Influenza and pneumonia were also important contributors to these differences, accounting for 15% and 13% of the smaller life expectancy gains among women for countries in the Low and Medium groups, respectively, compared to countries in the High groups.
Figure 3.7. Contributions of causes of death to gains in life expectancy at birth since 2010, men and women, 24 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
Finally, we turn our attention to which causes of death each group of countries has been less successful in reducing mortality from since 2010 compared to a decade ago (Figure 3.8). As we saw earlier, gains have been smaller in the most recent period than in the earlier period for Medium and Low countries, but similar in magnitude in both periods for the High group. The causes that contribute most to the smaller gains in the most recent period are the same for the Medium and Low groups. For men, countries in these groups are now making much smaller reductions from CVD and external causes compared to a decade prior, and they are experiencing substantial increases in the burden of mortality from mental and nervous system disorders. For women, countries in the Medium and Low groups are now making much smaller reductions from CVD. They are experiencing losses rather than gains in life expectancy from influenza and pneumonia, especially in the Low and Medium groups, and even larger losses from mental and nervous system disorders. Among women, smaller reductions in influenza and pneumonia since 2010 accounted for 20% of the difference in life expectancy gains between the two periods for countries in the Low group and 13% of the difference for countries in the Medium group.
Figure 3.8. Contributions of causes of death to gains in life expectancy at birth between 2000 and 2010 and since 2010, men and women, 24 OECD countries
Source: Data from the Human Mortality Database and World Health Organization Mortality Database.
Conclusions
This paper has illustrated which age groups and causes of death explain why some high-income countries are not experiencing the robust life expectancy gains enjoyed by others, and also why countries are not performing as well as they did a decade ago. The middle (45‑64) and older (65‑84) adult ages are currently the age groups that constitute the key sources of life expectancy improvements since 2010. However, these are also the age groups in which mortality reductions have slowed in Low- and Medium-performing countries. In contrast, High-performing countries have continued to post strong mortality reductions in these age groups. Furthermore, they have continued to experience stronger mortality improvements at the younger ages (below age 45) for men and at the oldest ages (above age 85) for women, which is not the case for their counterparts in the Low and Medium groups.
CVD is both an important source of life expectancy gains and an area where some countries are struggling to make reductions, including matching their progress from a decade ago. The same can be said of cancer. Other causes of death that have emerged as important contributors to the slowing of life expectancy gains include: accidental poisoning (drug overdose), mental and nervous system disorders (including Alzheimer’s disease), external causes, and, among women, influenza and pneumonia. Two key features of these trends are that countries in the Low and Medium groups are experiencing heavy burdens of mortality from mental and nervous system disorders, both compared to a decade prior and compared to countries in the High groups; and that while countries in the Low and Medium groups are losing years of life to accidental poisoning, the countries in the High groups appear to be largely exempt from this adverse trend.
As with any cross-national and over time comparison, we must interpret these results with caution and acknowledge that differences in cause of death coding across countries and over time may be present. It is likely that there have been considerable changes in diagnostic practices, death certification, and cause of death coding over the past two decades. It is most likely that an increasing number of deaths now being assigned to dementia, Alzheimer’s disease, and other mental and nervous system disorders would have been coded as due to CVD, respiratory disease, or influenza and pneumonia in the past. These cause of death categories are also showing up as key areas where mortality reductions have slowed.
Clearly, CVD is an important contributor to the stalled life expectancy improvements seen in some countries. This is a cause of death that is strongly linked to behavioural factors operating over the life course including smoking, obesity, diet, and sedentary lifestyle, as well as access to and quality of health care. Going forward, it is very important that we maintain the quality and timeliness of our vital registration systems and timely data releases. For most countries, there is an at least four‑year (and often greater) lag between the most recent cause-specific mortality data becoming available and the present year. These data are essential for identifying contemporary trends in CVD mortality, as well as the broader drivers underlying these trends.
Understanding the burden and causes of trends in CVD mortality is further complicated by the linkages between CVD and other causes of death, particularly influenza and pneumonia, and also drug overdose. Worldwide, there have been a series of severe influenza years since 2010. As populations’ age and deaths become increasingly concentrated at the older ages, the influence of elevated influenza mortality on life expectancy may become more evident. For example, the majority of high-income countries experienced declines in life expectancy between 2014 and 2015, and influenza and pneumonia, along with respiratory and cardiovascular diseases, were key contributors to these life expectancy declines (Ho and Hendi, 2018[5]). Differentiating among these causes of death is often a complicated process. Influenza and pneumonia can trigger cardiovascular events like heart attacks, and in turn, individuals with CVD may be more susceptible to dying from influenza or pneumonia. Thus, the underlying cause of death on the death certificate may end up being assigned to any one of these categories – influenza, pneumonia, CVD, or respiratory diseases, and studying trends in CVD mortality alone will not give us the full picture if a large proportion of those deaths are related to influenza.
Drug use and overdose are also linked to CVD, which is the leading cause of death among users of illicit drugs. Illicit drug use has been linked to several cardiovascular conditions including atrial fibrillation, cardiomyopathy, endocarditis, stroke, cardiac arrest, myocardial infarction (De Los Ríos et al., 2012[15]; Kadri et al., 2019[16]; Kevil et al., 2019[17]; Tseng et al., 2018[18]). Some of these deaths may be coded as CVD deaths, while others may be coded as drug overdose. Without understanding these connections, we may be arriving at an incomplete picture of why CVD mortality is no longer declining so rapidly. This is particularly important since mortality from drug overdose is increasing in several high-income countries (Ho, 2019[9]). While this epidemic is most advanced in the United States, high levels of drug overdose mortality have also been observed in other OECD countries, including several Nordic and Anglophone countries. Many of these countries are experiencing substantial increases in drug overdose mortality between the ages of 45 and 85 (Ho, 2019[9]) – which, as we saw earlier, is exactly the age range where mortality gains are failing to keep pace with prior decades.
References
[14] Arriaga E (1989), “Changing trends in mortality decline during the last decades”, in Ruzicka L, W. (ed.), Differential mortality: Methodological issues and biosocial factors, Clarendon Press, Oxford.
[8] Crimmins, E., S. Preston and B. Cohen (2011), International Differences in Mortality at Older Ages: Dimensions and Sources, National Academies Press, http://dx.doi.org/10.17226/12945.
[15] De Los Ríos, F. et al. (2012), Trends in substance abuse preceding stroke among young adults: A population-based study, http://dx.doi.org/10.1161/STROKEAHA.112.667808.
[9] Ho, J. (2019), “The Contemporary American Drug Overdose Epidemic in International Perspective”, Population and Development Review, Vol. 45/1, pp. 7-40, http://dx.doi.org/10.1111/padr.12228.
[5] Ho, J. and A. Hendi (2018), “Recent trends in life expectancy across high income countries: Retrospective observational study”, BMJ (Online), Vol. 362, http://dx.doi.org/10.1136/bmj.k2562.
[10] Ho, J. and S. Preston (2010), “US mortality in an international context: Age variations”, Population and Development Review, Vol. 36/4, pp. 749-773, http://dx.doi.org/10.1111/j.1728-4457.2010.00356.x.
[16] Kadri, A. et al. (2019), “Geographic Trends, Patient Characteristics, and Outcomes of Infective Endocarditis Associated With Drug Abuse in the United States From 2002 to 2016”, Journal of the American Heart Association, Vol. 8/19, p. e012969, http://dx.doi.org/10.1161/JAHA.119.012969.
[17] Kevil, C. et al. (2019), “Methamphetamine Use and Cardiovascular Disease”, Arteriosclerosis, thrombosis, and vascular biology, Vol. 39/9, pp. 1739-1746, http://dx.doi.org/10.1161/ATVBAHA.119.312461.
[1] Kochanek, K. et al. (2016), Mortality in the United States, 2016 Key findings Data from the National Vital Statistics System, https://www.cdc.gov/nchs/data/databriefs/db293_table.pdf#1. (accessed on 6 February 2020).
[3] Murphy, S. et al. (2017), Mortality in the United States, 2017 Key findings Data from the National Vital Statistics System, https://www.cdc.gov/nchs/data/databriefs/db328_tables-508.pdf#1. (accessed on 6 February 2020).
[12] Office for National Statistics (2018), Changing trends in mortality: an international comparison - Office for National Statistics, ONS, Titchfield, https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/lifeexpectancies/articles/changingtrendsinmortalityaninternationalcomparison/2000to2016 (accessed on 6 February 2020).
[6] Public Health England (2019), Health Profile for England, Health Profile for England, https://www.gov.uk/government/publications/health-profile-for-england-2019 (accessed on 6 February 2020).
[4] Public Health England (2018), A review of recent trends in mortality in England, http://www.facebook.com/PublicHealthEngland (accessed on 6 February 2020).
[7] Raleigh, V. (2019), “Trends in life expectancy in EU and other OECD countries : Why are improvements slowing?”, OECD Health Working Papers, No. 108, OECD Publishing, Paris, https://dx.doi.org/10.1787/223159ab-en.
[13] Steel, N. et al. (2018), “Changes in health in the countries of the UK and 150 English Local Authority areas 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016”, The Lancet, Vol. 392/10158, pp. 1647-1661, http://dx.doi.org/10.1016/s0140-6736(18)32207-4.
[18] Tseng, Z. et al. (2018), “Prospective Countywide Surveillance and Autopsy Characterization of Sudden Cardiac Death: POST SCD Study”, Circulation, Vol. 137/25, pp. 2689-2700, http://dx.doi.org/10.1161/CIRCULATIONAHA.117.033427.
[11] Woolf, S. and L. Aron (eds.) (2013), U.S. Health in International Perspective, National Academies Press, Washington, D.C., http://dx.doi.org/10.17226/13497.
[2] Xu, J. et al. (2016), “Mortality in the United States, 2015”, NCHS data brief 267, pp. 1-8, https://www.cdc.gov/nchs/products/databriefs/db267.htm.