Mind the gaps: age and cause specific mortality and life expectancy in the older population of South Korea and Japan

This study obtained data of the South Korean and Japanese population by five-year age groups from the World Population Prospects (WPP) 2019, and age- and cause-specific deaths from the Global Burden of Disease (GBD) 2017 Study [16]. The World Population Prospects and GBD data are publicly available on websites of the World Population Prospects (https://​population.​un.​org/​wpp/​) and the Global Health Data Exchange (http://ghdx.healthdata.org). The GBD 2017 study follows the Guidelines for Accurate and Transparent Health Estimates Reporting (GATHER), which includes recommendations on documentation of data sources, estimation methods, and statistical analysis. Detailed methods for the GBD 2017 study are provided in other publications [17]. The GBD 2017 study organises causes of death in a hierarchical list containing four levels. At the highest level (Level 1), all disease burdens are divided as three mutually exclusive and collectively exhaustive categories: communicable, maternal, neonatal, and nutritional (CMNN) diseases; non-communicable diseases (NCDs); and injuries. Level 2 distinguishes the Level 1 category into 21 cause groups, and Level 3 disaggregates these causes further, into 169 cause groups. The GBD data is available in the form of the number of deaths in point estimates and 95% uncertainty intervals. This study particularly used the mean values of the number of specific causes of death at Level 3 for both sexes in South Korea and Japan between 1997 and 2017. Among 169 causes of death, the top 20 causes of death for persons ages 60 years and older based on South Korea in 2017 were filtered out, and the remaining causes were merged into the ‘Others’ category. Figure 1 shows the top 20 causes of death in South Korea and Japan in 1997 and 2017. Although the ‘Others’ category slightly increased from 1997 to 2017 in both countries, the top 20 causes accounted for almost 80% of the total number of deaths in both countries for the two periods. In addition, the top 20 causes based on South Korea in 2017 were selected to explore the causes of death among the elderly in contemporary developed societies and investigate how South Korea caught up with the old-age life expectancy of Japan between 1997 and 2017.

A descriptive analysis of age structural changes in South Korea and Japan between 1997 to 2040 was carried out with the WPP 2019 data. Then, the multiple decrement table by five-year age groups for both countries between 1997 and 2017 was constructed to compute life expectancy (LE). A broad mathematical assumption of the multiple decrement life table is that an individual surviving to a certain age is the product of all independent risk-of-death probabilities. Although sampling variation is not an essential issue when calculating LE at national levels, we used Monte Carlo simulations using the probability of dying from an abridged life table to generate a binomial distribution of death numbers. The simulation was performed 10,000 times to generate the LE distribution, where the mean value was used as LE and the 2.5th and 97.5th percentiles of the distribution were used as the 95% confidence interval of LE. Calculation techniques for multiple decrement life tables and the standard error of LE are described elsewhere [18, 19].

In order to compare cause-specific mortality rates, this study uses age-standardised mortality rates per 1000 persons, with a direct method of standardisation further calculated to eliminate the effect of different age structures among the different population structures across two countries and times. The directly standardised mortality rate is given by the following formula; where d_it is the number of deaths for each specific cause of mortality t at age i; p_i denotes the number of persons at age i in the observed population; N_r is the total number of persons in the reference population; p_ir denotes the number of persons at age i in the reference population. The population to be used as a reference was derived from the average of the age distributions of South Korean and Japan in 1997 and 2017 at each age group. In addition, the Chi-square test was calculated for testing the significance of differences in age-standardised mortality rates (ASMR) of the two countries between 1997 and 2017.

Pollard’s actuarial method of decomposing life expectancy was then used to estimate the age- and cause-specific contributions to changes in life expectancy [20, 21]. This method was selected because it allows for simultaneously decomposing both different age groups and different causes of death. The decomposition method was calculated to examine the difference in life expectancy of (1) South Korea between 1997 and 2017 and (2) Japan between 1997 and 2017, which is given by the formula;

where \( {e}_x^{2017} \) and \( {e}_x^{1997} \) are the life expectancies at age x in 2017 and 1997 for the the decomposition of life expectancy within countries; n denotes the number of causes of death; \( i{Q}_x^{2017} \) and \( i{Q}_x^{1997} \) is the mortality rate of the i cause of death at age interval x with the weight ω_x. The values \( x{P}_{60}^{2017} \) and \( x{P}_{60}^{1997} \) denote the probability of surviving from 60 years old to 60 + x years of age between different times.

The same decomposition method was further calculated to examine the difference in life expectancy between (3) South Korea and Japan in 1997, and (4) South Korea and Japan in 2017, which is given by the following formula;

where j corresponding to Japan and k corresponding to South Korea for the decomposition of life expectancy between countries with the weight ω_x between different countries. This study based all life expectancy estimates on the population aged 60-64.

Figure 2 depicts the UN medium variant projection estimates of age structural changes in South Korean and Japan [1]. The proportion of elderly population over 60 years of age increased in both countries from 1997 to 2017 and is projected to further grow in 2040, as shown in the population pyramids in Fig. 2a (top grey bars). Although Japan had larger proportions of the population over 60 years of age in 1997 and 2017, the proportion in 2040 is expected to be more or less equal between the two countries, 40% in South Korea and 42% in Japan. In terms of the pace of population ageing, while Japan seems to follow a gradually increasing trend in proportions of the population over 60 years old, South Korea seems to follow a steadily increasing trend between 1997 and 2040 (see Fig. 2b). Particularly, the projected increase from 2017 to 2040 in the proportions of the population over 60 years old in South Korea is striking, as South Korea is expected to catch up with Japan’s four decades of growth just in the next two decades. Furthermore, the proportions of the population under 15 years old (bottom grey bars in Fig. 2 a) in South Korea shrank between 1997 and 2017 at a much faster pace than in Japan. Hence, in comparison with Japan, South Korea experienced faster population ageing, accompanied by precipitous falls in birth rates; and hence the country is very likely to face much more pressing demographic challenges in the next 20 years.

Although both countries experienced increases in old-age life expectancy between 1997 and 2017, South Korea rapidly caught up with life expectancy in Japan(see Table 1). Life expectancy at age 60 was 20.09(±0.04) years for South Korea and 23.87(±0.02) years for Japan in 1997, whereas the gap narrowed in 2017, as life expectancy increased to 25.09(±0.03) years for South Korea and 26.40(±0.01) years for Japan. The gaps in old-age life expectancy between the two countries therefore decreased by 2.47 from 3.78 in 1997 to 1.31 in 2017.

Table 2 shows estimated age-standardised mortality rates (ASMRs) per 1000 persons for the population aged 60 years and older in South Korea and Japan between 1997 and 2017. Between 1997 and 2017, age-standardised all-cause mortality rates decreased from 47.9 to 27.4 in South Korea and from 31.2 to 23.4 in Japan. The comparative mortality figure (CMF) in Table 2 is defined as the expected number of deaths in the standard population compared with those observed [22]. The CMF shows that in 1997 all-cause mortality of the elderly population in South Korea was 54% higher than in Japan, but the figure decreased to 17% in 2017. The last column shows the changes in ASMRs of each cause from 1997 to 2017, where negative figures indicate falls and positive signs indicate rises in ASMRs over the periods. ASMRs from stroke and ischemic heart disease largely declined in both countries. The highest ASMRs at old ages in 1997 were observed in stroke in both countries, but they were Alzheimer’s disease and dementia for both countries in 2017. On the other hand, ASMRs rose from a few causes between 1997 and 2017. While only one cause in Japan 2017, pancreatic cancer, had higher ASMRs than those in 1997, South Korea experienced increased ASMRs from four causes: Parkinson’s disease, lower respiratory tract disease, self-harm and falls. Interestingly, the increased figure in Japan was only categorised in non-communicable disease within the IHME’s Level 1 category; however, the increased figures of the four causes in South Korea were categorised within all the three Level 1 classifications, Communicable, Non-Communicable disease and Injuries. Moreover, ASMRs of lower respiratory tract disease in Japan were higher than South Korea in both years, but the two countries had opposite trends, from 2.6 to 1.9 in Japan, from 0.9 to 1.5 in South Korea. The Chi-square test showed that difference in ASMRs between 1997 and 2017 was statistically significant in South Korea, which was primarily derived from mortality declines in non-communicable disease.

Life expectancy at 60 years increased by 5 years in South Korea and 2.53 years in Japan between 1997 and 2017. Table 3, columns 1 and 2 show the age-specific contributions to the changes in old age life expectancy in South Korea and Japan between 1997 and 2017. The contributions to the 5 year life expectancy gains in South Korea were from across all old age groups, but the largest contribution was observed in the 70–79 age group in South Korea. Life expectancy gains for Japanese old adults also occurred across all old age groups, but the largest improvement was observed in the 75–84 age group. Table 4, columns 1 and 2 show the cause-specific contribution to the change in life expectancy in South Korea and Japan between 1997 and 2017. Reduced mortality from stroke and ischaemic heart disease explained most of the increase in life expectancy and mortality reductions from cancers also explained a large part of the increase in both countries. In relation to the ASMRs, increased mortality from lower respiratory tract disease, self-harm and falls slightly led to decreasing life-expectancy of South Korean old adults between 1997 and 2017. In addition, the increased mortality from Parkinson’s disease and pancreatic cancer in South Korea and Japan, respectively, between 1997 and 2017 led to the marginal decreases in life expectancy of South Korean and Japanese old adults. Figure 3 depicts age- and cause-specific contributions of aggregate figures of 20 causes within the Level 1 category to the changes in life expectancy of Korean and Japanese old adults between 1997 and 2017. It shows that large improvements were found in the 70–79 age groups in South Korea while improvements were pronounced in the 75–84 age groups in Japan. The figure also clearly illustrates that, among 20 causes, reduced mortality from non-communicable disease greatly contributed to the increases in life expectancy at 60 years in both countries between 1997 and 2017; however, the contribution of communicable diseases to the increases was larger in Japan.

Between 1997 and 2017, life expectancy gaps at 60 years of age between South Korea and Japan decreased by 2.47 from 3.79 in 1997, and 1.32 in 2017. Table 3, column 5 shows the age-specific contributions to the changes in gaps in life expectancy at 60 years between the two countries during 1997 and 2017, where negative figures indicate falls in gaps in old-age life expectancy. The age-specific contributions to the decreased gaps were found across all age groups, except for a small increase in the 90–94 age group, and the decreased gaps were especially large in the 70–79 years age group. Table 4, column 5 shows the cause-specific contributions to the decreased gap in life expectancy at 60 years in the two countries over the time period of 1997–2017, where positive signs indicate rises in gaps in old age life expectancy. Among 20 causes, cardiovascular disease (stroke, ischaemic heart disease and hypertensive-heart disease) and cancer deaths (stomach, liver, lung, pancreatic cancers), which were leading causes of the decline in ASMRs in South Korea between 1997 and 2017, contributed to the decreased gap in old-age life expectancy by 1.34 and 0.41, respectively. On the other hand, the three causes that showed increased ASMRs in South Korea between 1997 and 2017, lower respiratory tract disease, self-harm and falls, contributed to increased gaps in old age life expectancy between the two countries over the period. Moreover, changes in mortality pattern from two additional causes, chronic kidney disease and Alzheimer/Dementia, increased the gaps between the two countries over the period. However, the increased figure in chronic kidney disease (0.02 years) does not mean an increase in the gap, as the gap between these countries became almost zero in 2017, from a negative figure in 1997. In contrast, Alzheimer’s disease and dementia, which was the highest ASMRs for both countries in 2017, posed negative impacts on the gap between the two countries by 0.09 years during 1997–2017. Figure 4 depicts age- and cause-specific contributions of the aggregate figures of 20 causes within the IHME’s Level 1 category to the gaps in life expectancy at 60 years between the two countries during 1997 and 2017. Among the 20 causes, decreased gaps by non-communicable diseases were profound between 1997 and 2017 across all old age groups. However, bars indicating injury-associated causes slightly expanded from 1997 to 2017, especially in the age groups after 75. Moreover, although the gaps in communicable diseases decreased from − 0.13 to − 0.04, it should be noted that ASMRs of lower respiratory tract disease at an old age followed opposite trends in South Korea and Japan. In other words, ASMRs of lower respiratory tract disease among South Korean old adults is likely to outstrip Japan in the future. Overall, the result showed that causes from non-communicable diseases (cardiovascular disease and cancers) greatly contributed to decreasing the gaps in old-age life expectancy between South Korea and Japan; however, certain causes from communicable disease and injuries, in line with increased mortality patterns in South Korea, led to increasing the gaps.