Changes in Stroke Mortality Rates for 1950 to 1997
A Great Slowdown of Decline Trend in Japan
Background and Purpose— Stroke is the second leading cause of death in Japan. We sought to examine changes in stroke mortality rates by age, sex, and birth cohorts from 1950 to 1997.
Methods— The numbers of deaths from stroke and the population by age and sex in subjects aged 55 to 79 years between 1950 and 1997 were obtained from national vital statistics. Poisson regression analyses for annual percentage changes were used to explore these trends.
Results— Stroke mortality rates decreased since the mid-1960s for both men and women. Four periods of the decline are identified. In the first period (1950 to 1964), the age-adjusted annual percent changes averaged 0.9% for men (P<0.05) and −0.6% for women (P>0.05). In the second period (1965 to 1974), the annual change averaged −5.0% for men and −5.2% for women (both P<0.01). In the third period (1975 to 1989), the annual change averaged −8.6% for men and −8.7% for women (both P<0.01). In the fourth period (1990 to 1997), the annual change averaged −1.2% for men (P>0.05) and −3.0% for women (P<0.01). In the fourth period, the slowdown of the decline in stroke mortality was most evident in the older age groups of men.
Conclusions— The findings suggested that along with an increasing aging population, the slowdown in the decline of stroke mortality rates, especially for men, is of considerable concern. Efforts to control stroke should be vigorously continued in the 2000s.
The Japanese have the longest life expectancy in the world, which reached 77 years for men and 84 years for women in 1997. This prolongation of life expectancy is partly ascribed to the rapid decrease in stroke mortality rates. For example, age-adjusted stroke deaths have greatly decreased from 1048 per 100 000 to 203 per 100 000 for men aged 55 to 79 years and from 807 per 100 000 to 108 per 100 000 for women from 1950 to 1997 in Japan. This decline is probably due to a number of factors, including efforts made to control risk factors (eg, treatment for hypertension, discouragement of cigarette smoking, reduction in salt intake, etc), to improve treatment to reduce case-fatality rates, and to increase survival. However, stroke remains the second most common cause of death in Japan; in 1997, it accounted for 16% of total deaths.1,2 It affects mainly the elderly and results in a significant amount of disability.3 The present study examined the change in stroke mortality rates by age, sex, and birth cohort from 1950 to 1997. This approach may help us to understand the long-term trend in stroke mortality rates and may provide important hypotheses for further studies.
Age- and sex-specific stroke mortality data for subjects aged 55 to 79 years between 1950 and 1997 were obtained from national records of vital statistics.1,2 During this period, deaths from stroke were classified according to the International Classification of Disease (ICD) 6th and 7th editions (codes 330 to 334) from 1950 to 1967 and ICD 8th and 9th editions (codes 430 to 438) from 1968 to 1997. The certification of death from stroke was not greatly affected by the change in the coding system.4,5
To examine the change in stroke mortality rates by years, 4 periods were considered according to the general descriptive trends, as shown in Table 1: 1950 to 1964, 1965 to 1974, 1975 to 1989, and 1990 to 1997. Poisson (log-linear) regression analysis method was used to estimate annual percent changes in stroke mortality rates by age and period6 (SPSS Advanced Statistics: Poisson Log-Linear Regression, SPSS Advanced Statistics 7.5, 1997; SPSS Inc). This method allows for estimation of trends across individual calendar years to obtain average annual percentage changes. The Poisson regression procedure fits a model of the following form:
where year is given as 0, 1, 2, … 47 (year 0 is 1950, year 1 is 1951, and so on to 1997). Equation 1 is for the estimate of stroke mortality rate and observed years for a single age group. Equation 2 is for the regression of stroke mortality rates on observed years in all groups, with adjustment for age, because there are differences in stroke mortality rates by age. In equation 2, age (as a covariate) is given as 1, 2, 3, 4, and 5 for age groups 55 to 59, 60 to 64, 65 to 69, 70 to 74, and 75 to 79 years. The exponential of the estimate of β*(year) in equation 1 or β1*(year) in equation 2 is the relative risk per passing of each calendar year. For example, for men in period 1 (1950 to 1964), we estimated β1 to be 0.00907. The exponential of this is 1.00911, an average increase in stroke mortality rates of 0.9% per year among subjects aged 55 to 79 years, as shown in Table 2.
To examine changes by birth cohort in stroke mortality, analyses were made for twelve 5-year birth cohorts: 1881 to 1885 through 1936 to 1940. These classifications were based on mortality data for 5-year age groups from the periods of 1950 to 1995. For example, those who died in 1950 at age 65 to 69 years were born in 1881 to 1885; those who died in 1960 at age 70 to 74 years were born in 1886 to 1890; and those who died in 1995 at age 55 to 59 were born in 1936 to 1940.
Age- and sex-specific annual changes in mean blood pressure (BP) were calculated using the weighted mean method, and annual change in prevalence of smoking was calculated using the formula ln(R)=[ln(R0/R1)]/t, where R is the annual rate of change, R0 and R1 are the actual rates at the start and end of the time period of interest, and t is the duration of the period. The period of 1966 to 1997 was estimated because data for smoking were obtained for only this period. Finally, the Pearson correlation analysis method was used to examine the association of stroke mortality rates with mean BP and prevalence of smoking from 1966 to 1997. All data analyses were conducted using SPSS for PC version 8.0, 1998 (SPSS).
Changes in Stroke Mortality Rates by Periods
Table 1 shows the trends of stroke mortality rates by age and sex from 1950 to 1997. Stroke mortality rates decreased sharply from the mid-1960s to 1997, with the exception of men aged 75 to 79 years. Stroke mortality rates increased with age, and the rate for men was higher than that for women in all age groups. The absolute changes in the rates slowed down greatly in recent years. Table 2 shows that in the first period (1950 to 1964), the age-adjusted annual percentage change in stroke mortality rates for all age groups was 0.9% for men (P<0.05) and −0.6% for women (P>0.05). In the second period (1965 to 1974), the annual percentage change was −5.0% for men and −5.2% for women (each P<0.01). In the third period (1975 to 1989), the annual change was −8.6% for men and −8.7% for women (both P<0.01). In the fourth period (1990 to 1997), however, the annual change was only −1.2% for men (P>0.05) and −3.0% for women (P<0.01). In the fourth period, in men there was a significant decline in the mortality rate for ages 55 to 59 only (P<0.01) but nonsignificant changes in other groups aged 60 and older (P>0.05). In women, although the decline trends by age groups remained significant at P<0.01 in the fourth period, the absolute decline trends were greatly lowered compared with the third period (range from −7.3% to −9.5% in the third period and −2.1% to −3.5% in the fourth period).
Changes in Stroke Mortality Rates by Birth Cohorts
Figures 1 and 2⇓ and Table 3 and Table 3A depict age-specific male and female stroke mortality rates on a logarithmic scale by 5 age groups and 12 birth cohorts. First, as expected, mortality rates increased with age in all birth cohorts. Second, the figures show that by age group, this decline trend started in the oldest age group of men and then gradually spread to the youngest men (Figure 1). For women, the decline in mortality rates started earlier than it did for men in all age groups (Figure 2). In addition, the decline trend leveled off in the recent cohorts in both sexes; this slowdown was most evident for older men. Data for the birth cohorts and mortality rates are shown in Table 3 and Table 3A.
Changes in Two Main Risk Factors for Stroke
Several studies have examined risk factors for stroke death in Japan.7–10 High BP and cigarette smoking were suggested to be the most important risk factors among a number of lifestyle-related factors.9 Decline trends in mean BP and prevalence of cigarette smoking were observed in Japan. For example, in men aged 40 to 59 years, mean systolic BP was 145.2 mm Hg in 1963, followed by 137.4, 135.3, and 134.3 mm Hg in 1980, 1990, and 1997, respectively. The corresponding values for diastolic BP were 86.3, 84.9, 84.5, and 83.8 mm Hg in 1963, 1980, 1990, and 1997, respectively.3,7 Prevalence of cigarette smoking in men aged 40 to 59 years was 86%, 80%, 70%, 63%, and 58% in 1966, 1970, 1980, 1990, and 1997, respectively. The corresponding values for smoking rates in women were 23%, 19%, 14%, 13% and 12%, respectively.3,7 The decline trends in mean BP and prevalence of cigarette smoking were significantly correlated with the decline in stroke mortality rates by age and sex (Table 4).
The study analyses advance previous observations and demonstrated 4 distinct periods of trends in stroke mortality since 1950.7–10 The results highlighted the important health issue of the great slowdown of this trend from 1990 for men and women.
The study was based on data from national vital statistics. The main characteristics are (1) complete data on the whole population, rather than on selected groups of patients. The data that were used cover all 47 prefectures of Japan. (2) The data are relatively reliable, because the ICD system has been used in Japan for >100 years, since 1899.1,11 Since 1949, annual vital statistics have been reported officially to the World Health Organization1,11,12 and have provided data for a number of studies related to international comparison analyses.9,12,13 (3) During the period under review, although the frequency of CT scanning may have led to changes in accurate diagnosis of forms of stroke, such as hemorrhagic,14,15 the advent of CT scanners is most unlikely to have affected the apparent mortality rates for all strokes combined, which is the principal parameter considered in this report.
However, some limitations should be considered. First, no extensive validation of death certificate data on stroke has been reported in Japan, although some studies from Japan and western countries have reported that death certificate diagnosis of stroke is sufficiently accurate for descriptive epidemiological studies.14,15 Second, 4 revisions to the ICD system were used during the study period. There are no reports on the comparability of the 4 revisions in Japan, although studies from the United States suggested that death certification for stroke has changed very little from ICD codes 330 to 334 (ICD 6 and 7) to ICD codes 430 to 438 (ICD 8 and 9).4,5 Third, only a single cause of death is coded and analyzed in the mortality statistics. This obscures deaths due to multiple causes, especially in those with competing (or shared) risk of diseases, such as coronary heart disease and cancer.16,17 These potential limitations may result in an underestimate or overestimate of stroke mortality rates during the study period. It is, however, unlikely that the accuracy of diagnosis and the coding system has deteriorated in Japan. Any major changes in how stroke death is recorded should have manifested as a period of death effect or as a sudden jump between 2 consecutive years before and after a new edition, and there would be a similar effect for men and women. In fact, no such sudden jump was observed, and the changes in the slowdown of decline trends in stroke death were much greater for men than for women (particularly in period 4, Table 2). Although careful interpretation is necessary, the long-term trends are unlikely to be biased.
On a global scale, stroke mortality rates have decreased rapidly in most developed and developing countries since the mid-1960s or 1970s.7,12,13 A great slowdown in the decline in the stroke mortality rate was reported from the United States.18 In the present study, we observed the decline in stroke mortality rates starting from the mid-1960s in Japan (period 2). However, the decline slowed greatly in recent years. The reasons for the slowdown are not clear. Changes in risk factors of stroke would mostly account for these changing patterns of stroke death.
Risk factors for stroke can be generally classified into 1 of 2 groups: (1) unpreventable, such as age, sex, etc, and (2) preventable, such as high BP, smoking, excess salt intake, etc. The most recent report by Ueshima et al,9 based on an average follow-up period of 14 years from the 1980 national cardiovascular survey in Japan (NIPPON DATA 80), indicated that stroke death was significantly associated with hypertension and cigarette smoking. Thus, changes in the 2 major risk factors should, at least in part, explain the changes in stroke mortality rates. Mean BP and prevalence of cigarette smoking decreased from the 1960s to the 1990s for both men and women.3,7,19–22 Table 4 shows the significant correlations of the decline in stroke mortality rates with the decline in mean BP and prevalence of smoking from 1966 to 1997, although it cannot be interpreted further as cause-effect association from this data, because of the limitation of the ecological analysis. It should be noted that smoking prevalence is still very high for Japanese men and is increasing for younger Japanese women.3,20–22 There is no doubt that efforts to control smoking and high BP should be vigorously continued as the primary measures to lower stroke mortality rates.
The overall changes in stroke mortality rates were similar among age groups from period 2 to period 4 (Table 2), and the greatest slowdown of the decline trend was seen from 1990 to 1997. These changes were confirmed by birth cohort analysis (the almost parallel lines for age, Figures 1 and 2⇑). However, there were differences in the changes for sex. The slowdown trend was particularly observed in men compared with women (Table 2). These results suggest that similar risk factors consistently influenced the risk of stroke but to a different extent in men and women. It is unclear why so much slowdown of the decline trend has occurred in recent years and why the slowdown is greater in men aged 60 or older. Further studies will be required to determine whether a new risk factor has come in play since 1990 or whether control of the main, established risk factors has slipped. The slowdown of decline trends in elderly men implies that young persons, who have not reached the high-risk age groups, may experience a slowdown of the decline similar to that for the elderly. Thus, the great burden of this disease may well continue in the 2000s much at the present level.
In conclusion, the present study sets out both the dramatic decline in stroke mortality rates in Japan since 1965 and the marked slowdown in the decline since 1990. The causes of this very unwelcome recent development require urgent attention, particularly in view of the rapid increase in the elderly population.
- Received September 11, 2000.
- Revision received April 19, 2001.
- Accepted April 27, 2001.
Statistics and Information Department, Ministry of Health, and Welfare. Vital Statistics: 1950 to 1997. Tokyo, Japan: Ministry of Health and Welfare, Japan; 1980 to 1999.
Statistics and Information Department, Ministry of Health, and Welfare. Age-Adjusted Death Rates by Prefecture Special Report on Vital Statistics: 1995. Tokyo, Japan: Ministry of Health and Welfare, Japan; 1998.
Ministry of Health and Welfare. Guide for Lifestyle Related Diseases [in Japanese]. Tokyo, Japan: Mitsuiseimei Kouseijigyodan; 1998: 102.
Klebba AJ, Dolman AB. Comparability of mortality statistics for the seventh and eighth revisions of the International Classification of Diseases adapted for use in the United States. Vital Health Stat. 1975; 2: 1–93.
Klebba AJ. Estimates of selected comparability ratios based on dual coding of 1976 death certificates by the eighth and ninth revisions of the International Classification of Diseases. Natl Center Health Stat Mnthly Vital Stat Rep. 1980; 28 (suppl).
Armitage P, Berry G. Statistical Methods in Medical Research, 3rd ed. Oxford, UK: Blackwell Scientific Publications; 1994.
World Health Organization. Profile of Cardiovascular Diseases, Diabetes Mellitus and Associated Risk Factors in the Western Pacific Region. Manila, Philippines: WHO Regional Office for the Western Pacific; 1999.
Ueshima H, Tatara K, Asakura S. Mortality trends in stroke, blood pressure, the treatment rate of hypertension between 1961 and 1980 in Japan: slowdown of the decreasing rate of stroke mortality of men aged 30s and 40s [in Japanese]. Jpn J Public Health. 1984; 31: 589–599.
Ueshima H. Trends in Circulatory Diseases and Risk Factors in Japan (NIPPON DATA). Osaka, Japan: Asian Pacific Society of Atherosclerosis and Vascular Disease; April 1999. Abstract.
Ueda K, Omae T, Hirota Y, Takeshita M, Katsuki S, Tanaka K, Enjoji M. Decreasing trend in incidence and mortality from stroke in Hisayama residents, Japan. Stroke. 1981; 12: 154–160.
Shigematsu I. Epidemiology in Japan and future problems. Asian Med J. 1988; 31: 620–627.
World Health Organization. World Health Statistics Annual: 1995. Geneva, Switzerland: WHO; 1997.
Bonita R, Stewart A, Beaglehole R. International trends in stroke mortality: 1970–1985. Stroke. 1990; 21: 989–992.
Iso H, Jacobs DRJ, Goldman L. Accuracy of death certificate diagnosis of intracranial hemorrhage and nonhemorrhagic stroke: the Minnesota Heart Survey. Am J Epidemiol. 1990; 132: 993–998.
Rothman KJ, Greenland S. Modern Epidemiology, 2nd ed. Philadelphia, Pa: Lippincott-Raven Publishers; 1998.
Benfante R, Yano K, Hwang LJ, Curb JD, Kagan A, Ross W. Elevated serum cholesterol is a risk factor for both coronary heart disease and thromboembolic stroke in Hawaiian Japanese men: implications of shared risk. Stroke. 1994; 25: 814–820.
Cooper R, Sempos C, Hsich SC, Kovar MG. Slowdown in the decline of stroke mortality in the United States, 1978–1986. Stroke. 1990; 21: 1274–1279.
Tominaga S, Aoki K. The UICC Smoking Control Workshop. Nagoya, Japan: The University of Nagaya Press; 1981: 33.
Liu LJ, Ueshima H, Okayama A, Nakamura Y, Choudhury SR, Kita Y, Hayakawa T, Kadowaki T, Watanabe M. Differences in dietary habits and relative body weight according to smoking status: results from the 1990 National Cardiovascular Survey in Japan. CVD Prev. 1998; 1: 282–289.