Association of Body Mass Index With Blood Pressure in Elderly Japanese American Men
The Honolulu Heart Program
Few data are available on the association between body mass index (BMI) and blood pressure in elderly individuals, particularly among minority groups. We studied the cross-sectional association of BMI with systolic and diastolic pressures in 1378 Japanese American men 60 to 82 years of age who were participants in the population-based Third Lipoprotein Examination of the Honolulu Heart Program conducted between 1980 and 1982. When the subjects were divided into 5-year age groups, mean BMI decreased linearly with increasing age. Mean systolic pressure rose from 134.8 mm Hg in the first quintile of BMI to 138 in the second and 141.8 in the third quintiles, with levels of 139.2 and 142 in the fourth and fifth quintiles, respectively (test for trend, P=.083). Mean diastolic pressure rose from 78.1 mm Hg in the lowest quintile of BMI to 83.9 in the highest (test for trend, P=.008). We performed multiple regression analysis, controlling for factors known to influence blood pressure values, including age, physical activity index, alcohol intake, current smoking status, and diabetes mellitus. The association between BMI and both systolic and diastolic pressures remained highly statistically significant in these analyses. These results show that obesity and high blood pressure continue to be highly correlated even in old age and suggest that it may be possible to modify rates of hypertension by changes in body weight.
The association of BP with cardiovascular disease is well documented in middle-aged individuals.1 2 3 Hypertension is associated with an increased risk for total mortality as well as CHD mortality and stroke mortality.1 2 3 SBP and DBP levels are significant risk factors for cardiovascular disease in the elderly as well.4 5 6 7
Obesity has been shown to be both an independent direct risk factor for cardiovascular disease8 9 and an indirect risk factor because of its effects on diabetes, hypertension, and hyperlipidemia.10 However, the effect of obesity on risks for hypertension, diabetes, and hypercholesterolemia has been shown to be more pronounced in individuals aged 20 to 45 years than among those aged 45 to 75.10 Some studies in elderly populations have also shown a correlation between body weight or BMI and SBP, particularly in men.11 12 Few data are available from elderly minority populations, particularly Asians. Studies from China showed that BMI is significantly associated with BP in young and middle-aged lean populations.13 14
Reports have conflicted regarding the association between obesity and cardiovascular disease in the elderly. The Framingham study showed no association between BMI and CHD in the elderly, but there was an increased risk of CHD based on higher relative weight at the beginning of the study.15 A previous report from the Honolulu Heart Program demonstrated a positive association between BMI and CHD in middle-aged men aged 51 to 59 years, but this association was not seen in the elderly, aged 65 to 74.5 However, the elderly men had significantly lower mean BMI compared with the middle-aged men.
The data collected in a detailed examination of a 30% random subsample of the Honolulu Heart Program cohort during 1980 to 82 provided an opportunity for us to explore the relationship between body fatness and BP in a relatively lean cohort of elderly Japanese American men aged 60 to 82 years.
The Honolulu Heart Program is a prospective epidemiological study of cardiovascular disease that is initiated and funded by the National Heart, Lung, and Blood Institute (National Institutes of Health). Participants were men of Japanese ancestry living on the island of Oahu, Hawaii, in 1965, who were born between 1900 and 1919 (ages 45 to 68 years at the time of the first examination). Eight thousand and six men participated in the first examination between 1965 and 1968; details of the selection process for this cohort have been published previously.16 Continuous morbidity and mortality surveillance through monitoring of hospital discharge records and death certificates has been performed since the beginning of the study. The entire cohort has undergone five examinations; in addition, a 30% random subsample of the men underwent another three examinations, the last of these being in 1980 to 82. This report is based on a cross-sectional analysis of data collected on 1378 men aged 60 to 82 years at this last subsample examination. The study was approved by the institutional review committee of Kuakini Medical Center; informed consent was obtained from all participants; and procedures followed were in accordance with institutional guidelines.
The examination included medical questionnaires with questions regarding smoking history, physical activity, and alcohol consumption. A detailed diet history was also obtained. Fasting serum cholesterol, high-density lipoprotein cholesterol, and triglyceride concentrations were measured. Height, weight, and skinfold thickness measurements were also obtained. Details of the data collection methods used in this examination have been previously published.17
Sitting BP was measured three times (twice by a nurse, once by a physician) on the left arm with a standard mercury sphygmomanometer and Velcro cuff; an average of the three measures was used in the analysis. SBP and DBP were defined as the points of the appearance and disappearance of Korotkoff sounds, respectively.
BMI (or Quetelet's index) was defined as weight (in kilograms) divided by height (in meters squared). This index has been shown to be a good measure of body fatness and is applicable to all populations.18
Subjects were divided into quintiles of BMI and 5-year age groups. We performed multiple regression analyses separately for SBP and DBP to establish their associations with BMI. To avoid potential bias caused by the effect of confounding variables that may be associated with both BP and BMI, we repeated the analyses excluding men with prevalent CHD, stroke, cancer, and diabetes mellitus as well as those on antihypertensive or cholesterol-lowering medications. The criteria for prevalent CHD19 and stroke20 have been published previously. Diabetes mellitus was defined by history or medication use.
The subjects were divided into quintiles on the basis of BMI (Table 1⇓). Mean BMI rose from 19.51 kg/m2 in the lowest quintile to 27.71 in the highest. Table 1⇓ also shows mean values (and SD for continuous variables) of potential confounding variables by quintiles of BMI. In general, men in the lowest quintile of BMI tended to be older and smoked more than men in higher quintiles. There was no clear pattern between quintiles of BMI for physical activity index or alcohol consumption nor for prevalent diabetes mellitus, CHD, stroke, or cancer.
When the subjects were divided into 5-year age groups, the mean BMI decreased linearly with age, from 24.08 kg/m2 in the 60-64 age group to 22.95 in the 75+ age group (test for trend, P=.012) (Table 2⇓). Mean SBP increased linearly with age, from 135.72 mm Hg in the 60-64 year age group to 146.25 in the 75+ age group (test for trend, P=.011), and DBP decreased with age from 82.18 mm Hg in the 60-64 year age group to 79.56 in the 75+ age group (test for trend, P=.016).
We examined mean SBP by quintile of BMI. Mean SBP rose from 134.8 mm Hg in the first quintile to 142 in the fifth quintile (test for trend, P=.083) (Fig 1⇓). This increase in SBP appeared to plateau at the third quintile of BMI. To ascertain whether age influenced the BMI-SBP relationship, we examined the trend for mean SBP values in each quintile of BMI within each 5-year age group. SBP increased with increasing BMI in all age groups; however, the most marked increase occurred in the older age groups, particularly in men 70 years and older (Fig 2⇓).
Mean DBP also rose by quintile of BMI, from 78.1 mm Hg in the lowest quintile to 83.9 in the highest (test for trend, P=.008) (Fig 1⇑). Again, to ascertain whether age influenced this relationship, we examined the trend for mean DBP by quintile of BMI for each 5-year age group. As with SBP, DBP increased by quintile of BMI in all age groups, and this was most evident in the oldest age group of 75+ years (Fig 2⇑).
We performed separate multiple regression analyses on SBP and DBP in association with BMI and other confounding factors known to influence BP and BMI, including age, physical activity index, alcohol consumption, current smoking, and diabetes mellitus (Table 3⇓). We found BMI to be independently and significantly associated with both SBP and DBP (P=.0001). Each unit increase in BMI (1 kg/m2) was associated with an increase of 1.15 mm Hg in SBP and 0.70 mm Hg in DBP, after adjusting for other variables. SBP was also significantly associated with age (P=.0001) and alcohol intake (P=.0009) but not with physical activity index, current smoking status, or diabetes mellitus. DBP was significantly negatively associated with age (P=.004), current smoking (P=.002), and diabetes mellitus (P=.03) but was not associated with physical activity index or alcohol consumption.
We repeated the same multiple regression analyses by age group for men 60 to 69 years of age and men 70 years and older. In both age groups, BMI was independently and significantly associated with both SBP and DBP.
To study the attributable risk of hypertension to BMI, we also examined the percentage of men with hypertension (defined as SBP ≥160 or DBP ≥95 mm Hg or use of antihypertensive medications) in the highest quintile of BMI, where the hypertension could be attributed to BMI. We found that 24.5% of the hypertension in men in the top quintile could have been eliminated if BMIs in these men were decreased to be similar to those values in the bottom quintile of BMI.
We repeated all analyses excluding men with medical conditions that might influence BP and BMI, such as prevalent CHD, stroke, cancer, and diabetes mellitus, as well as those men on antihypertensive or cholesterol-lowering medications to avoid potential bias (data not shown). After exclusions, 728 men remained to be studied. Results were similar; BMI again was significantly independently associated with both SBP and DBP.
Our study showed that BMI is significantly and independently associated with both SBP and DBP in elderly Japanese American men. This association appears to plateau at the middle and upper levels of BMI for SBP, except in the older ages (older than 70 years). In contrast, a previous publication including middle-aged men from this cohort showed a linear relationship between BMI and both SBP and DBP, with no plateauing at upper levels of BMI.21 Thus, it appears that the relationship of BMI and SBP changes somewhat at older ages in this population.
Previous reports from the Ni-Hon-San Study demonstrated that the apparent differences in mean BP values among three Japanese populations in Japan, Honolulu, and Northern California were possibly mediated by differences in weight. After statistical adjustment for weight, BP no longer differed significantly between the populations.22 A later report from the first examination (1965 to 1968) and the third examination (1971 to 1974) of the Honolulu Heart Program cohort found that BMI had a stronger correlation with both SBP and DBP than other factors known to influence BP.23 In addition, changes in weight between exams 1 and 3 were significantly positively associated with changes in SBP and DBP in both normotensive and hypertensive individuals.23 Another study in obese white middle-aged and elderly men showed that weight loss was associated with significant reductions in both SBP and DBP.24
There are ethnic differences in body weight caused by differences in body build. For the same height, Japanese Americans have a lower average weight than white Americans.25 The Ni-Hon-San study also showed that the prevalence of obesity was low in Japanese in Japan, Hawaii, and California (particularly in Japan) and these rates decreased with age.26 Other population-based studies have shown that mean BMI values increase with age until 69 years, after which they decrease.27 The Honolulu Heart Program cohort tends to be lean compared with whites of the same age, and BMI decreases with age from the age of 45 years.26 Few men in this cohort are frankly obese. Thus, the relationships seen in a more obese population might be different. However, these results indicate that even within the range of relatively ideal body weight, BP level is significantly related to BMI.
In industrialized nations, SBP rises continually with age, whereas DBP rises until middle age and then tends to level off.7 28 The plateau in the SBP-BMI association seen in this population may be partly due to selective mortality of those individuals with hypertension and partly due to reduction of BMI values with age, while SBP rises for other reasons. The data available for older age groups are limited. However, further study of the relationships between cardiovascular risk factors in older populations is warranted.
Since the study population was a random sample, the results can be generalized to the rest of the Honolulu Heart Program cohort of Japanese American men. It will be interesting to see whether similar results will be seen in other elderly populations as well as in women. Several factors may influence BMI and BP and act as potential confounders. We addressed this by adjusting for these variables in multivariate analyses as well as by repeating analyses excluding participants with medical conditions that may act as confounders, which did not change the main findings of this study.
As in any cross-sectional study, caution must be exercised in data interpretation because these data presented here do not address the temporal relationship between the measures. In this population, obesity and high BP continued to be significantly correlated even in older individuals. The prevalence of hypertension continued to increase with age despite a general decrease in BMI. Nevertheless, it may be possible to further modify rates of hypertension in the elderly by reducing body weight, particularly among those with high BMI values. Hypertension is one of the most common chronic conditions in the elderly population and remains strongly associated with morbidity and mortality. If lowering BMI is found to significantly decrease rates of hypertension in the elderly, this may have significant public health implications.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|CHD||=||coronary heart disease|
|DBP||=||diastolic blood pressure|
|SBP||=||systolic blood pressure|
This research was supported by contract number N01-HC-02901 from the National Heart, Lung, and Blood Institute, and contract number N01-AG-4-2149 from the National Institute on Aging, National Institutes of Health, Bethesda, Md.
Reprint requests to Kamal H. Masaki, MD, The Honolulu Heart Program, 347 N Kuakini St, Honolulu, HI 96817. E-mail email@example.com
- Received April 15, 1996.
- Revision received May 24, 1996.
- Revision received September 12, 1996.
Bots ML, Grobbee DE, Hofman A. High blood pressure in the elderly. Epidemiol Rev. 1991;13:294-314.
Hubert HB, Feinleib M, McNamara PM, Castelli WP. Obesity as an independent risk factor for cardiovascular disease: a 26-year follow-up of participants in the Framingham Heart Study. Circulation. 1983;67:968-977.
Curb JD, Marcus EB. Body fat, coronary heart disease, and stroke in Japanese men. Am J Clin Nutr. 1991;53:1612S-1615S.
Van Itallie TB. Health implications of overweight and obesity in the United States. Ann Intern Med. 1985;103:983-988.
Clarke RP, Schlenker ED, Merrow SB. Nutrient intake, adiposity, plasma total cholesterol, and blood pressure of rural participants in the (Vermont) Nutrition Program for Older Americans (Title III). Am J Clin Nutr. 1981;34:1743-1751.
Huang Z, Wu X, Stamler J, Rao X, Tao S, Friedewald WT, Liao Y, Tsai R, Stamler R, He H, Zhou B, Taylor J, Li Y, Xiao Z, Williams OD, Cen R, Zhang H. A north-south comparison of blood pressure and factors related to blood pressure in the People's Republic of China: a report from the PRC-USA Collaborative Study of Cardiovascular Epidemiology. J Hypertens. 1994;12:1103-1112.
Yano K, Reed D, McGee D. Ten-year incidence of coronary heart disease in The Honolulu Heart Program: relationship to biological and lifestyle characteristics. Am J Epidemiol. 1984;119:653-666.
Kagan A, Marmot MG, Kato H. The Ni-Hon-San study of cardiovascular disease epidemiology: population characteristics and epidemiology of stroke. In: Kesteloot H, Joossens JV, eds. Epidemiology of Arterial Blood Pressure. Boston, Mass: Martinus-Nijhoff; 1980:423-436.
Winkelstein W, Kagan A, Kato H, Sacks ST. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California: blood pressure distributions. Am J Epidemiol. 1975;102:502-513.
Reed D, McGee D, Yano K. Biological and social correlates of blood pressure among Japanese men in Hawaii. Hypertension. 1982;4:406-414.
Gordon T, Kagan A, Rhoads G. Relative weights in different populations. Am J Clin Nutr. 1975;28:304-305.
Curb JD, Marcus EB. Body fat and obesity in Japanese Americans. Am J Clin Nutr. 1991;53:1552S-1555S.
The Lipid Research Clinics Population Studies Data Book, Volume I: The Prevalence Study. Washington, DC: US GPO; 1979. US Department of Health, Education, and Welfare publication NIH 79-1527.
Curb JD, Taylor AA. Mild hypertension and atherosclerosis. In: Gotto AM Jr, Paoletti R, eds. Atherosclerosis Reviews, Volume 9. New York, NY: Raven Press Publishers; 1982:53-84.