Background and Purpose This article examines prevalence of isolated systolic hypertension (ISH) in a population of Japanese-American men living in Hawaii and compares rates of stroke among those with ISH, isolated diastolic hypertension (IDH), combined systolic/diastolic hypertension (SDH), and nonhypertension.
Methods Eight thousand six men aged 45 to 68 years participated in a baseline examination as part of a prospective study of coronary heart disease and stroke and were followed up for 20 years for incident disease and total mortality. Men were divided into four groups according to baseline blood pressure: ISH (systolic blood pressure [SBP] ≥160 mm Hg and diastolic blood pressure [DBP] <90 mm Hg); IDH (SBP <160 mm Hg and DBP ≥90 mm Hg); SDH (SBP ≥160 mm Hg and DBP ≥90 mm Hg); and nonhypertension (SBP <160 mm Hg and DBP <90 mm Hg).
Results Men with ISH had the highest rates of stroke, followed by men with SDH. Men with IDH had rates only slightly higher than men in the nonhypertension group. Relative risk adjusted for other risk factors varied by age group. For men aged 45 to 54 years, relative risks of stroke associated with ISH, IDH, and SDH compared with nonhypertensive subjects were 4.8, 1.4, and 4.3, respectively. For men aged 55 to 68 years, the same relative risks were 1.2, 1.8, and 1.7. Excluding men on antihypertensive medication at baseline did not alter results.
Conclusions ISH had a much greater impact on risk of stroke in younger than in older men. The results suggest that while ISH is more common in older men, in younger men it is associated with higher relative risk of stroke.
Mean systolic blood pressure (SBP) rises with increasing age in industrialized nations, while mean diastolic blood pressure (DBP) increases until middle age and then tends to level off.1 In an individual, elevation of SBP above the normal level with normal DBP is termed isolated systolic hypertension (ISH). The prevalence of ISH increases dramatically with increasing age in older individuals, and this type of hypertension has been strongly associated with stroke, other cardiovascular disease, and death in numerous epidemiological studies.2 3 4 5 More than 25 million people in the United States are currently age 65 or older, and as the proportion of elderly Americans continues to rise, ISH is becoming a significant public health problem.
However, recent findings from the Systolic Hypertension in the Elderly Program indicated that treatment of ISH with low-dose antihypertensive medication reduced risk of stroke by 36% during the 5-year follow-up period.6
ISH is usually defined as SBP ≥160 mm Hg and DBP <90 mm Hg. Prevalence of ISH in the United States by this definition is between 5% and 30% in various studies of the population older than 65 years, depending on the actual age range.
The pathophysiology of ISH is probably different from that of diastolic hypertension. Increasing arterial rigidity with increased peripheral resistance may contribute to the condition in elderly individuals, whereas high cardiac output may be the main factor in younger individuals with ISH.7 In less developed countries where atherosclerosis is rare, the increase in mean SBP with age seen in this country is not noted.8 9
In this report we examine the prevalence of ISH in a population of Japanese-American men living in Hawaii and compare rates of stroke among those with ISH, isolated diastolic hypertension (IDH), combined systolic and diastolic hypertension (SDH), and a nonhypertension group.
Subjects and Methods
The Honolulu Heart Program examined 8006 men from 1965 to 1968 as part of a prospective population-based study of coronary heart disease and stroke. Men had to be of Japanese ancestry, born between 1900 and 1919, and living on the island of Oahu in 1965 to be eligible for the study. Blood pressure was determined at the baseline examination (1965 to 1968). Sitting blood pressure was measured three times by means of a standard sphygmomanometer and cuff on the left arm after a 5-minute rest. The mean of the readings was used as the individual’s blood pressure value in these analyses.
Questions were asked regarding use of antihypertensive medication, smoking status, alcohol intake, and symptoms and history of coronary heart disease and stroke. Blood was drawn from nonfasting participants 1 hour after a 50-g glucose load and was analyzed for glucose and total cholesterol by Auto-Analyzer methods.10 An electrocardiogram was taken and a physical examination was performed by a physician, focusing on evidence of heart disease and stroke.
Since 1965, these men have been followed up for incident coronary heart disease and stroke via a continuous surveillance system consisting of regular monitoring of all hospital discharge records on the island of Oahu, death certificates, obituary notices in local newspapers, and medical examiners’ cases. All suspected cases of stroke were reviewed by the study neurologist. Records reviewed included hospital histories, physical examinations, neurological consultations, and discharge summaries. When performed, results and interpretations of cerebrospinal fluid examination, nuclear brain scans, electroencephalograms, duplex studies, and angiograms were also reviewed. Computed tomography (CT) scans became available in 1976 and have been widely used since 1978. The participant’s physician and family were contacted for additional information when needed.
The criteria for diagnosis of thromboembolic stroke were relatively sudden onset of a neurological deficit lasting at least 2 weeks or a clinical course typical of stroke with residual of less than 2 weeks with neuroimaging, surgical, or postmortem findings consistent with clinical stroke. Diagnosis of hemorrhagic stroke required neurological deficit or signs or symptoms of meningism with hemorrhagic lumbar puncture or hemorrhage on CT scan. Subarachnoid hemorrhage was differentiated from cerebral hemorrhage by clinical findings as well as the results of either nuclear or CT scans or angiography. Strokelike episodes attributable to other disease processes such as blood dyscrasias, neoplastic disease, head injuries, surgical accident, meningoencephalitis, fat embolism, epilepsy, or cardiac arrest were excluded.
The Honolulu Heart Program cohort is an extremely stable population. Out-migration has been less than 1/1000 per year, and some records have been obtained from cases occurring in the continental United States and in Japan, so that ascertainment is thought to be nearly complete.
For the current study men were divided into four groups based on their blood pressure level at the baseline examination: ISH (SBP ≥160 mm Hg and DBP <90 mm Hg); IDH (SBP <160 mm Hg and DBP ≥90 mm Hg); SDH (SBP ≥160 mm Hg and DBP ≥90 mm Hg); and nonhypertension (SBP <160 mm Hg and DBP <90 mm Hg). Prevalent cases of stroke and coronary heart disease were removed from analysis when stroke incidence rates were determined. Men were followed up from the date of their baseline examination through December 31, 1985. Cases of stroke included all cases defined as stroke, according to the above criteria, both nonfatal and fatal.
Mean levels of cardiovascular risk factors were adjusted for age by the general linear regression method.11 Age-specific incidence rates of stroke were obtained by person-years/life-table analysis.12 Relative risks, confidence intervals (CIs), and the testing of hypotheses, nonlinear effects, and interaction terms were calculated by means of Cox proportional-hazards analysis.13 The analysis was done both including and excluding men on antihypertensive medications.
As seen in Table 1⇓, of the 8006 men who attended the first examination, 7590 were free of prevalent coronary heart disease or stroke. Of these men, 5605 were assigned to the nonhypertension group as defined by this study, 183 had ISH, 1146 had IDH, and 656 had SDH. Men with IDH were the youngest (mean age, 53 years), followed by those in the nonhypertension group (mean age, 54 years). The SDH group had a mean age of 56 years, and the men with ISH were the oldest, with a mean age of 60 years. Mean blood pressures for the groups were as follows: nonhypertension group, 126/77 mm Hg; ISH group, 168/84 mm Hg; IDH group, 145/95 mm Hg; and SDH group, 174/102 mm Hg.
The nonhypertension group had the lowest levels of all risk factors except pack-years of smoking. The number of pack-years of smoking was similar in all groups except the ISH group, which had a significantly greater value (P<.01). Body mass index and cholesterol were lowest in the nonhypertension group, intermediate in the ISH group, and highest in the IDH and SDH groups. For the remaining risk factors of glucose, alcohol intake, and percentage taking antihypertensive medication, the nonhypertension group had the lowest levels, the IDH group intermediate levels, and the ISH and SDH groups the highest levels.
Men were divided into two age groups: 45 to 54 years and 55 to 68 years. As shown in Table 2⇓, 75.1% of the younger (middle-aged) group and 72.1% of the older group were in the nonhypertensive group. ISH was more than four times as common in the older group compared with the middle-aged group (4.4% versus 1.0%). A greater percentage of older men had SDH (11.1% versus 6.9%). Similar distributions were seen after excluding 625 men on antihypertensive medication.
The relative risks of stroke for men with ISH were 4.8 (CI, 2.1 to 11.0) for the middle-aged men and 1.2 (CI, 0.7 to 2.0) for the older men (Table 3⇓). SDH was associated with relative risks of 4.3 (CI, 2.9 to 6.3) in the middle-aged men versus 1.7 (CI, 1.2 to 2.4) in the older men. IDH was associated with relative risks of 1.4 (CI, 1.0 to 2.1) and 1.8 (CI, 1.3 to 2.5) in the middle-aged and older men, respectively. Removing men on antihypertensive medication from the analysis resulted in only slight and insignificant changes.
When strokes were divided into subtypes (thromboembolic and hemorrhagic), numbers of cases in each blood pressure group were small and CIs were wide. This was particularly true for hemorrhagic stroke. There were no cases of hemorrhagic stroke among middle-aged men with ISH, and therefore a relative risk could not be calculated for this group. For all other age and blood pressure groups, relative risks for both thromboembolic and hemorrhagic stroke were similar in magnitude to relative risks for total stroke.
For thromboembolic stroke in middle-aged men, the relative risks were 3.7, 1.3, and 6.3 in the ISH, IDH, and SDH groups, respectively. As with total stroke, the relative risk reached significance in the ISH and SDH groups. In older men the relative risks of thromboembolic stroke were 1.2, 1.8, and 1.6 in the ISH, IDH, and SDH groups, respectively. As with total stroke, the relative risks were significant in the ISH and SDH groups.
For hemorrhagic stroke in middle-aged men, the relative risk for individuals with ISH could not be determined because there were no cases. For the IDH and SDH groups, the relative risks were 1.2 and 5.6, respectively. The relative risk in the SDH group reached significance, as it did for total stroke. In older men, the relative risks of hemorrhagic stroke were 1.6, 1.7, and 2.4 for men with ISH, IDH, and SDH, respectively. The relative risk in men with SDH reached significance, as it did with total stroke. In men with IDH, the relative risk of total stroke was 1.8 compared with 1.7 for hemorrhagic stroke. The relative risk for total stroke in this group was significant, whereas the relative risk of hemorrhagic stroke did not reach significance. However, the number of hemorrhagic stroke cases in this age and blood pressure group was only 12.
When examined in Cox proportional-hazards models with other risk factors including age, serum cholesterol and glucose, alcohol intake, and pack-years of smoking, both ISH and SDH were significant predictors of stroke and showed inverse interactions with age. IDH showed no independent association with stroke and no interaction with age.
When men were divided by tertiles of DBP and similar models run in each tertile, the risk of stroke was positively related to SBP in each tertile. However, there was no clear-cut relation between risk of stroke and DBP when men were divided by tertiles of SBP (Figs 1⇓ and 2⇓).
This study examined the associations between several types of hypertension and the incidence of stroke in a cohort of middle-aged Japanese-American men followed up for 20 years. The most striking finding was the high risk of stroke found in the middle-aged men with ISH. However, it should be noted that ISH is rare among middle-aged men. There were only 43 men with ISH in the group aged 45 to 54 years, and there were six stroke events. This small sample size warrants caution in interpretation of the results, and confirmatory data from other trials would be useful. In contrast, the older men (aged 55 to 68 years) with ISH had, at most, a slightly elevated risk compared with normotensive subjects in their age group, and the 95% CI for that relative risk included one. In middle-aged men the existence of SDH was associated with a lower risk than that of ISH, whereas SDH put the older men at higher risk than either type of hypertension alone. IDH was associated with only slight and insignificant increases in risk over normotensive subjects in both middle-aged and older men. The incidence of stroke was highest in older men, and ISH was also more common in that group. However, the relative risk of stroke associated with ISH was much greater in middle-aged men. These findings persisted after excluding men who were on antihypertensive medication. Relative risks for hemorrhagic and thromboembolic strokes were similar in magnitude to relative risks for total stroke.
There has been growing interest in the factors that predispose the elderly to morbidity and mortality because the elderly constitute an increasing proportion of the US and worldwide populations. DBP has traditionally been used to define hypertension in both clinical and research settings. However, SBP and the prevalence of systolic hypertension, particularly ISH, increase dramatically with age. These factors have led to a number of reports on the association between systolic hypertension and disease in the elderly.2 3 4 5 14 15 16 Although the age range, the definition of ISH, and the control of confounding variables have varied among these studies, the general finding is that ISH is associated with an increased risk of stroke in the elderly, although in general, age-specific risk has not been presented.5 14 15 16 Our study confirms the greater importance of SBP in predicting future stroke. It also confirms the increased risk associated with ISH compared with the nonhypertensive group. The elevation of risk due to ISH is much greater in middle-aged men than in older men. Studies in other populations have previously shown indications of a relatively low relative risk of stroke associated with hypertension at older ages.17 18 Hypertension was relatively weakly associated with stroke mortality in upper-middle-class whites aged 65 to 74 years in a southern California retirement community (relative risk, 1.8).17 The relative risk of stroke associated with hypertension was reported to be considerably lower in older men (70 to 79 years) than in middle-aged men (50 to 59 years) in the Framingham Study.18
One possible explanation for the findings in the present study is that middle-aged men with ISH in this cohort may have very advanced pathology (atherosclerosis) compared with older men with ISH. This is suggested by the greater number of current smokers, higher SBPs, reduced likelihood of being on antihypertensive treatment, and greater body mass index among middle-aged men with ISH. Another explanation may be that there is a different and more malignant pathophysiology for ISH in younger individuals, leading to early selective mortality. This may also explain why diastolic hypertension in the presence of systolic hypertension may have different effects in middle-aged versus older men. One study that examined this possibility noted a high output state in men with ISH who were aged younger than 35 years.7 However, the youngest men in the Honolulu Heart Program were 45 to 54 years old. We noted that among men with ISH at baseline, the older men were more likely to be taking antihypertensive medication than the middle-aged men (25% versus 19% of men with ISH). This tendency is likely to have continued after the baseline examination, since younger men are less likely to see physicians and thus may not receive blood pressure screening. Older men, on the other hand, are more likely to be in the medical care system.19 A survey of primary care physicians in Honolulu in 1988 indicated that most would treat elderly patients who had elevated SBP with antihypertensive medication.20 However, although the increased risk of stroke associated with ISH may have been reduced more by subsequent treatment in the older group, it is difficult to believe that treatment in this age group was aggressive enough to eliminate risk. The incremental risk of treated hypertensive subjects who developed and were treated for hypertension after the baseline visit, as well as the risk of those who developed hypertension and were never treated, may also contribute to these findings. Since the incidence of hypertension is greater in older than in younger individuals, such changes in risk status within the group classified as normotensive at baseline may contribute to the apparent decreased risk in the older group. It is not possible to measure this effect since regular examinations and follow-up of the hypertension and treatment status were not part of the Honolulu Heart Program study design. However, it is unlikely that this could totally explain these findings.
The possibility of genetic physiological differences between our population (Japanese-American men living in Hawaii) and other populations studied cannot be excluded. However, it is apparent from this and other studies that the risk associated with hypertension may have complex relations with the aging process. Further investigation in closely monitored populations is warranted.
- Received April 18, 1994.
- Revision received September 26, 1994.
- Accepted October 12, 1994.
- Copyright © 1995 by American Heart Association
Kannel WB, Dawber TR, McGee DL. Perspectives on systolic hypertension: the Framingham Study. Circulation. 1980;61:1179-1182.
Colandrea MA, Friedman GD, Nichaman MZ, Lynd CN. Systolic hypertension in the elderly. Circulation. 1970;41:239-245.
Curb JD, Borhani NO, Entwisle G, Tung B, Kass E, Schnaper H, Williams W, Berman R. Isolated systolic hypertension in 14 communities. Am J Epidemiol. 1985;121:362-370.
Shekelle RB, Ostfeld AM, Klawans HL Jr. Hypertension and risk of stroke in an elderly population. Stroke. 1974;5:71-75.
Page LB. Epidemiological evidence on the etiology of human hypertension and its possible prevention. Am Heart J. 1976;1:527-534.
Page LB. Hypertension and atherosclerosis in primitive and acculturating societies. In: Hunt JC, ed. Hypertension Update. Bloomfield, NJ: HCS Press; 1979:5-61.
Yano D, Reed DM, McGee DL. Ten-year incidence of coronary heart disease in the Honolulu Heart Program: relationship to biologic and life-style characteristics. Am J Epidemiol. 1984;119:653-666.
Walker SH, Duncan DB. Estimation of the probability of an event as a function of several independent variables. Biometrika. 1967;54:167-179.
Kalbfleisch JD, Prentice RL. The Statistical Analysis of Failure Time Data. New York, NY: John Wiley & Sons, Inc; 1980:84-87.
Cox JR. Regression models and life tables. J R Stat Soc (B). 1972;34:187-202.
Kannel WB, Dawber TR, Sorlie P, Wolf PA. Components of blood pressure and risk of atherothrombotic brain infarction in the Framingham Study. Stroke. 1976;7:327-331.
Rabkin SW, Mathewson FA, Tate RB. Predicting risk of ischemic heart disease and cerebrovascular disease from systolic and diastolic blood pressures. Ann Intern Med. 1978;88:342-345.
O’Malley K, McCormack P, O’Brien ET. Isolated systolic hypertension: data from the European Working Party on High Blood Pressure in the Elderly. J Hypertens. 1988;6(suppl 1):S105-S108.
Vital and Health Statistics. Health Data on Older Americans: United States, 1992. Atlanta, Ga: Centers for Disease Control and Prevention/National Center for Health Statistics; 1992. Series 3;27:121.
Kaw KT, Barrett-Connor E, Suarez L, Criqui MH. Predictors of stroke-associated mortality in the elderly. Stroke. 1984;15:244-248.