Ten-Year Prognosis of Stroke and Risk Factors for Death in a Japanese Community
The Hisayama Study
Background and Purpose— There have been very few population-based cohort studies of long-term prognosis and risk factors for death after stroke. We examined the 10-year prognosis, causes, and risk factors of death after stroke in a Japanese cohort.
Methods— During a 26-year follow-up of a cohort of 1621 subjects ≥40 years of age, 333 subjects developed first-ever stroke and were prospectively followed up for 10 years after onset. During these 10-year follow-up periods, 268 of the 333 stroke patients died. Of those, 239 (89.2%) underwent autopsy.
Results— The risk of death was greatest in the first year after first-stroke onset in both sexes (men, 40.3%; women, 43.7%). Thereafter, the survival curves decreased gradually, and risk of death reached 80.7% for men and 80.2% for women by the end of the 10-year follow-up. The 30-day case fatality rate was substantially greater in patients with cerebral hemorrhage (63.3%) or subarachnoid hemorrhage (58.6%) than in patients with cerebral infarction (9.0%). The risk of dying after the first stroke was twice the risk for stroke-free subjects. The most common cause of death was the index stroke in the first year. Thereafter, the impact of the first stroke gradually decreased, while that of recurrent stroke increased. Multivariate analysis revealed that age, lower body mass index, and hemorrhagic stroke were significant risk factors for death after stroke.
Conclusions— Our findings suggest that the risk of death after first-ever stroke is high, in part because of the larger proportion of hemorrhagic stroke in Japanese relative to stroke victims in Western countries.
In Japan, stroke is the third-leading cause of death.1 Stroke is also a major cause of disability2 and cognitive dysfunction in the elderly.3,4 Because the elderly population in Japan has been increasing rapidly, stroke-related problems in Japanese health care have become important in recent years. Information on the risk of death after stroke and on stroke predictors would be helpful in coping with these problems. Although the literature concerning fatal prognosis after stroke is extensive, most studies are based on selected series of patients referred to a hospital.5–9 Such patients are often not representative of all stroke cases, because patients with the most severe stroke die quickly, before reaching the hospital, and patients with mild stroke may not go to the hospital at all. A prospective study of a defined population can clearly assess this problem, but very few population-based cohort studies have been able to define accurately the pathological type of stroke and evaluate its long-term prognosis.10–12
Since 1961, we have been carrying out a prospective cohort study of cardiovascular disease in the town of Hisayama,
See Editorial Comment, page 2347
Japan.13,14 The most outstanding features of this study are that causes of death were verified by autopsy and that brain lesions of most stroke patients were examined morphologically during autopsy or by brain imaging such as CT and MRI. The purposes of this study are to determine the 10-year survival rate after first-ever stroke in the community of Hisayama; to compare the observed risks of death after first-ever stroke with those of age- and sex-matched stroke-free subjects selected from the same community; to determine the major causes of death during different time periods up to 10 years after a stroke; and to determine risk factors for death during this period.
Subjects and Methods
Hisayama is a suburb of Fukuoka on Kyushu Island in southern Japan. In 1961, a total of 1621 men and women ≥40 years of age who had never suffered a stroke were recruited from the residents (88.1% of the total population of this age range) as a cohort. These subjects were then followed up for 26 years, until 1987. A detailed description of the study methods was published previously.12–14 Briefly, we collected information about new cardiovascular events through a daily monitoring system established by the study team, local practitioners, and the town government. When we suspected new neurological symptoms, the study physicians carefully evaluated the subject and tried to obtain information by further diagnostic examination, including lumbar puncture, cerebral angiography, or recent brain CT or MRI. During the 26-year study period, only 2 subjects were lost to follow-up, and 852 subjects died. Of those who died, 704 (82.6%) underwent autopsy examination.
Stroke Cases and Controls
Stroke, defined as a sudden onset of a nonconvulsive and focal neurological deficit persisting for >24 hours, was classified as cerebral infarction, cerebral hemorrhage, subarachnoid hemorrhage, or undetermined type.14,15 The diagnosis of stroke and determination of its pathological type were based on clinical history, neurological examination, all available clinical data (including brain CT or MRI), and autopsy findings. During follow-up, we identified 340 new stroke events and divided them into 244 cases of cerebral infarction (120 men, 124 women), 60 cases of cerebral hemorrhage (40 men, 20 women), 29 cases of subarachnoid hemorrhage (6 men, 23 women), and 7 cases of undetermined type (3 men, 4 women). After exclusion of the 7 cases of undetermined type, 333 were enrolled in the study (166 men, 167 women; mean age, 73±10 years; range, 46 to 97 years). As a reference group, we randomly selected individuals free from stroke at baseline among residents of Hisayama who participated in health checkups in 1973 or 1974, which was the middle of the study period. Each stroke case had 2 sex- and age-matched (±2 years) controls. The reference group consisted of 666 individuals (332 men, 334 women; mean age, 72±8 years; range, 46 to 97 years).
We followed up the stroke patients from the onset of the stroke until death or 1997 and the reference group for 10 years from 1973 or 1974 to 1983 or 1985. No cases or controls were lost to follow-up.
During the 10-year follow-up, 268 of the 333 stroke patients died; of them, 239 (89.2%) underwent autopsy examination. We reviewed all the available clinical information and interviewed the attending physicians and the families of the deceased subjects. For each individual, the underlying disease was chosen as the cause of death. Diseases linked to the underlying cause of death were classified into the following 4 categories established in the Oxfordshire Community Stroke Project.7 First-stroke deaths were due to the direct effects of a brain lesion or to complications of immobility resulting from the first stroke. These included deaths from bronchopneumonia even years after the stroke if stroke-related impairments were thought to be in some way responsible and there was no other, more likely, cause of death. Recurrent stroke deaths were attributed directly to brain lesions or complications of immobility after a recurrent stroke (ie, with symptoms that led to early death and were associated with an increase in disability). If a death from stroke-related impairments occurred after a recurrent stroke, it was attributed to the recurrent stroke rather than the first stroke. Cardiovascular deaths were due to definite or probable cardiac causes, ruptured aortic aneurysms, or peripheral vascular disease. Sudden deaths were regarded as cardiovascular unless an alternative explanation was found at autopsy. Nonvascular deaths were those unrelated to any stroke disability and clearly were due to a nonvascular cause, eg, cancer, bronchopneumonia, accidents, or suicide.
Risk Factors for Death
To elucidate the risk factors for death in stroke cases, we collected the following data from the regular health checks performed within 2 years before the onset of stroke: age at onset; sex; alcohol consumption; smoking habits; glucose intolerance; average of 3 systolic and diastolic blood pressures; hypertension (≥140/90 mm Hg or use of antihypertensive agents); body mass index; abnormal ECG findings, including left ventricular hypertrophy (Minnesota code 3-1) and/or ST depression (Minnesota code 4-1, 4-2, 4-3); atrial fibrillation (Minnesota code 8- 3); and serum total cholesterol. The categories used in the definition of glucose intolerance were described in a previous report.16
SAS software (version 6.12) was used for the statistical analysis. Survival curves within the 10 years were calculated with the Kaplan-Meier product limit technique. The odds ratio and 95% confidence intervals (CIs) for stroke cases compared with reference subjects were calculated by the χ2 test. We estimated the age-adjusted and multivariate relative risk (RR) of each potential risk factor for death by using the β coefficients available in the Cox proportional-hazards analysis.
Absolute Risks for All Patients
Figure 1 shows Kaplan-Meier survival curves by sex for all 333 stroke patients. To similar degrees in both sexes, the risk of death was greatest in the first year (men: 40.3%; 95% CI, 32.9 to 47.8; women: 43.7%; 95% CI, 36.1 to 51.2) and particularly during the first 30 days after stroke (men: 24.7%; 95% CI, 18.1 to 31.3; women: 21.6%; 95% CI, 15.3 to 27.8%). Beyond the first year, the survival curves for the patients decreased similarly to those for reference subjects in both sexes. The risk of death reached 66.9% (95% CI, 59.7 to 74.0) for men and 67.7% (95% CI, 60.6 to 74.8) for women up to 5 years and 80.7% (95% CI, 14.9 to 27.2) and 80.2% (95% CI, 74.2 to 86.3) up to 10 years, respectively.
Absolute Risks for Subgroups
Stratification by age showed that patients <65 years of age had a worse prognosis during the early period after stroke than did older patients (Figure 2). Among other age groups, the risk of death increased the older the patient was, regardless of the poststroke period. The survival curves comparing prognoses among the different pathological types of stroke are shown in Figure 3. The fatality rate during the early period was substantially greater for patients with hemorrhagic stroke (cerebral hemorrhage: 63.3%; 95% CI, 51.1 to 75.5; subarachnoid hemorrhage: 58.6%; 95% CI, 40.7 to 76.5) than for patients with cerebral infarction (9.0%; 95% CI, 5.4 to 12.6).
RR Compared With Controls
The risk of dying after first stroke was twice that of reference subjects (Table 1). In the first year after stroke, patients had a 12-fold RR of death, which declined to 2- or 3-fold over the subsequent years up to 5 years and to 1.6-fold in the next 5 years. Patients who survived at least 30 days had a 1.9-fold higher risk of dying over the next 10 years than reference subjects.
Cause of Death
A histogram of causes of death during different time intervals from the onset of first-ever stroke is shown in Figure 4. During the first 30 days after first stroke, 83.1% of deaths were due to the direct neurological effects of the index stroke, and another 6.5% and 2.6% were due to recurrent stroke and other cardiovascular diseases, respectively. Thereafter, the frequency of first stroke gradually decreased, whereas that of recurrent stroke increased. Over the 10-year follow-up, 41.0% of deaths were due to first stroke and 21.6% to recurrent stroke. Only 5.6% of deaths were due to other cardiovascular causes.
Risk Factors for Death Over 10 Years
Table 2 shows the multivariate prediction model for death 10 years after first-ever stroke among the 333 patients and the 244 patients diagnosed with cerebral infarction using baseline risk factors that appeared to be significant in the univariate analysis. The significant prognostic factors for death among stroke patients were age (RR for a 10-year increment, 1.55; 95% CI, 1.34 to 1.79), body mass index (RR, 0.95; 95% CI, 0.91 to 0.98), and hemorrhagic stroke, including cerebral hemorrhage and subarachnoid hemorrhage (RR, 2.84; 95% CI, 2.17 to 3.72). Among patients with cerebral infarction, significant predictors of death were age (RR, 2.03; 95% CI, 1.87 to 2.45), body mass index (RR, 0.91; 95% CI, 0.85 to 0.95), hypertension (RR, 1.57; 95% CI, 1.14 to 2.17), and atrial fibrillation (RR, 1.62; 95% CI, 1.11 to 2.36).
The strengths of our study are that it is community based and includes almost all stroke cases developed in a cohort, thus eliminating the bias of case selection encountered in clinical series of hospital cases; a large proportion of the stroke patients underwent CT or MRI examination or autopsy to determine the pathological type; an autopsy was performed in almost 90% of deceased patients; and RRs and absolute risks of death were estimated up to 10 years after first-ever stroke.
In our study, the 5-year risk of death after first stroke was 67% for men and 68% for women, exceeding the 40% to 54% rates reported in previous community-based studies.6–8,10 The higher fatality rates of our stroke patients are attributed to the older age of our subjects and the larger proportion (26%) of hemorrhagic stroke. These findings were similar to those of a previous study performed in another part of Japan.11 In particular, our patients with subarachnoid hemorrhage showed a higher fatality rate (59%) in the early period compared with other studies in which the risks were 37% to 46%.6–8,10 This can be explained partly by the fact that we identified almost all instantaneous or sudden deaths by subarachnoid hemorrhage, even in cases lacking a correct diagnosis before death, because autopsies were performed in 83% of the deaths among our inception cohort.
Cause of Death
Our data confirm that the RR of death in the year after first-ever stroke for which the index stroke is mainly responsible is very high (odds ratio, 12.2; 95% CI, 7.99 to 18.55), and patients continue to die with a 2- to 3-fold excess risk in subsequent years up to 10 years. These late deaths are not due mainly to the index stroke but to recurrent stroke. Compared with Western populations,7,8 our subjects are characterized by a relatively small proportion of other cardiovascular events among causes of death during the late period after stroke. This probably reflects the lower risk of atherosclerotic diseases such as coronary heart and peripheral vascular diseases in Japanese.
Risk Factor for Death
As in previous studies,7,8,11 we found that advanced age was associated with a greater risk of death in the long-term period after stroke. This suggests that older patients may have more advanced underlying diseases. Meanwhile, in our subjects, the risk of death during the early period was worse in younger patients (<65 years of age) than in older patients. This phenomenon was attributed to a larger proportion of hemorrhagic stroke in this younger age group than among older patients.
In our study, the survival curves of hemorrhagic stroke dropped rapidly during the course of the first year after index stroke, particularly during the early period, and almost reached a plateau during the period after 1 year, whereas during the 10 years overall, the risk of dying from hemorrhagic stroke was 2.8-fold higher than the risk of dying from cerebral infarction. These findings imply that the risk of death from hemorrhagic stroke was determined in the acute period but that its adverse effect on fatal prognosis remained for the full 10-year follow-up period.
There are conflicting reports of the impact of hypertension on survival after stroke.8–11,17,18 In most of those reports,8,9,17 blood pressure was taken after the stroke; thus, the measurements may have been influenced by the acute event and thereby may be difficult to evaluate accurately. As in our study, prospective cohort studies in Framingham (Mass)10 and Manitoba (Canada)18 demonstrated a significant association between hypertension before stroke and increased risk of subsequent death.
Among our subjects, lower body mass index was a significant risk factor for death among patients with cerebral infarction and among total stroke patients. Our previous study showed that low body mass index was a risk factor for death from pneumonia in the elderly.19 Thus, being lean may reflect insufficient nourishment and decreased resistance to bacterial infection. Several previous reports have indicated excess mortality rates from cardiovascular disease among lean hypertensive subjects.20–22 Lean hypertensive individuals tend to have more severe end-organ damage20 and may suffer from higher peripheral vascular resistance than those who are obese21 and may well carry stronger genetic determinants of cardiovascular disease than obese hypertensive subjects.22
Arial fibrillation was another independent risk factor for death among our patients with cerebral infarction, as in other population-based studies.17,23 Atrial fibrillation often induces congestive heart failure and cerebral embolism, leading to early death.
There are several potential limitations to the findings in our study. First, we collected stroke cases that developed among an inception cohort during 26 years of follow-up. Thus, their prognoses may have changed during the long-term observation period. Secular trends in the prognosis of stroke patients should be taken into account, and we will do so in another study. Second, the severity of the index stroke was not taken into account in the evaluation of risk factors for fatality. Thus, our estimates of the effects of risk factors are probably conservative. Finally, we could not provide information on the rehabilitation of stroke patients and evolution of the level of their functional impairment. Nonetheless, we believe that our findings contribute to a better understanding of the prognosis of stroke patients and of stroke predictors in Japanese who are considered to be at higher risk of stroke than are people in other countries.
Our findings confirm the importance of primary prevention of stroke, because Japanese are characterized by a larger proportion of hemorrhagic stroke and thereby a higher fatality rate after stroke. The maintenance of good nutrition may be important to improve the long-term prognosis after stroke. In addition, management of hypertension and atrial fibrillation is considered to be useful to avoid early death after cerebral infarction.
- Received April 17, 2003.
- Revision received June 4, 2003.
- Accepted June 25, 2003.
Statistics and Information Department, Minister’s Secretariat, Ministry of Health, Labour and Welfare of Japan. Vital Statistics of Japan 2001, Volume 3. Tokyo, Japan: Health and Welfare Statistics Association; 2003.
Yoshitake T, Kiyohara Y, Kato I, Ohmura T, Iwamoto H, Nakayama K, Ohmori S, Nomiyama K, Kawano H, Ueda K, et al. Incidence and risk factors of vascular dementia and Alzheimer’s disease in a defined elderly Japanese population: the Hisayama Study. Neurology. 1995; 45: 1161–1168.
Schmidt EV, Smirnov VE, Ryabova VS. Results of the seven-year prospective study of stroke patients. Stroke. 1988; 19: 942–949.
Matsumoto N, Whisnant JP, Kurland LT, Okazaki H. Natural history of stroke in Rochester, Minnesota, 1955 through 1969: an extension of a previous study, 1945 through 1954. Stroke. 1973; 4: 20–29.
Dennis MS, Burn JPS, Sandercock PAG, Bamford JM, Wade DT, Warlow CP. Long-term survival after first-ever stroke: the Oxfordshire Community Stroke Project. Stroke. 1993; 24: 796–800.
Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Burvill PW, Anderson CS, Stewart-Wynne EG. Five-year survival after first-ever stroke and related prognostic factors in the Perth Community Stroke Study. Stroke. 2000; 31: 2080–2086.
Sacco RL, Shi T, Zamanillo MC, Kargman DE. Predictors of mortality and recurrence after hospitalized cerebral infarction in an urban community: the Northern Manhattan Stroke Study. Neurology. 1994; 44: 626–634.
Sacco RL, Wolf PA, Kannel WB, McNamara PM. Survival and recurrence following stroke: the Framingham study. Stroke. 1982; 13: 290–295.
Kojima S, Omura T, Wakamatsu W, Kishi M, Yamazaki T, Iida M, Komachi Y. Prognosis and disability of stroke patients after 5 years in Akita, Japan. Stroke. 1990; 21: 72–77.
Kiyohara Y, Ueda K, Hasuo Y, Wada J, Kawano H, Kato I, Sinkawa A, Ohmura T, Iwamoto H, Omae T, Fujishima M. Incidence and prognosis of subarachnoid hemorrhage in a Japanese rural community. Stroke. 1989; 20: 1150–1155.
Katsuki S. Epidemiological and clinicopathological study on cerebrovascular disease in Japan. Prog Brain Res. 1966; 21B: 64–89.
Tanizaki Y, Kiyohara Y, Kato I, Iwamoto H, Nakayama K, Shinohara N, Arima H, Tanaka K, Ibayashi S, Fujishima M. Incidence and risk factors for subtypes of cerebral infarction in a general population: the Hisayama Study. Stroke. 2000; 31: 2616–2622.
Cerebrovascular Diseases: Prevention, Treatment, and Rehabilitation. Geneva, Switzerland: World Health Organization; 1971. Technical report series No. 469.
Omae T, Ueda K. Diabetic nephropathy in a general adult population of Hisayama. In: Abe H, Hoshi M, eds. Diabetic Microangiopathy. Tokyo, Japan: University of Tokyo Press; 1983: 317–328.
Petty GW, Brown RD, Whisnant JP, Sicks JD, O’Fallon WM, Wiebers DO. Survival and recurrence after first cerebral infarction: a population-based study in Rochester, Minnesota, 1975 through 1989. Neurology. 1998; 50: 208–216.
Iwamoto H, Kiyohara Y, Kato I, Ohmura T, Nakayama K, Ohmori S, Nomiyama K, Yoshitake T, Ueda K, Fujishima M. Mortality from major causes of death and its risk factors in the elderly: 26-year follow-up study in Hisayama [in Japanese with English abstract]. Jpn J Geriatr Soc. 1994; 31: 671–676.
Goldbourt U, Holtzman E, Cohen-Mandelzweig L, Neufeld HN. Enhanced risk of coronary heart disease mortality in lean hypertensive men. Hypertension. 1987; 10: 22–28.
Carman WJ, Barrett-Connor E, Sowers M, Khaw K. Higher risk of cardiovascular mortality among lean hypertensive individuals in Tecumseh, Michigan. Circulation. 1994; 89: 703–711.
Sandercock P, Bamford J, Dennis M, Burn J, Slattery J, Jones L, Boonyakarnkul S, Warlow C. Atrial fibrillation and stroke: prevalence in different types of stroke and influence on early and long term prognosis (Oxfordshire Community Stroke Project). BMJ. 1992; 305: 1460–1465.