- Email: [email protected]
Factors That Affect Blood Pressure Variability
AJH
1997;10:1281–1289
Factors That Affect Blood Pressure Variability A Community-Based Study in Ohasama, Japan Yutaka Imai, Akiko Aihara, Takayoshi Ohkubo, Kenichi Nagai, Ichiro Tsuji, Naoyoshi Minami, Hiroshi Satoh, and Shigeru Hisamichi
Factors that affect blood pressure (BP) variability, ie, standard deviation (SD) and variation coefficient (VC: SD/average ambulatory BP) of ambulatory BP, were examined in a communitybased sample in northeastern Japan. Screening and ambulatory BPs were measured in 823 subjects >20 years of age, and the effects of age and BP on the SD and the VC were examined. In bivariate regression analysis, the SD of ambulatory BP was positively correlated with age and the ambulatory BP. The VC was also correlated with age. Both the SD and the VC were strongly correlated with the magnitude of the nocturnal decline in BP. Ambulatory BP was positively correlated with age and negatively correlated with heart rate and the SD of heart rate. Multivariate analysis demonstrated that the nocturnal decline in BP showed the strongest association with the SD and the VC of 24-h BP. However, age and BP were still independently and positively associated with the SD and the VC of ambulatory BP. Furthermore, pulse pressure and BMI were independently and
positively associated with the SD and the VC of ambulatory BP. Since the SD and the VC of 24-h ambulatory BP were determined mainly by the nocturnal decline in BP, this variable appears to be an index of the circadian variation in BP and not an index of short-term BP variability. Pulse pressure, an index of arterial stiffness, was a relatively strong predictor of the SD and the VC of BP. In addition, the SD of heart rate, an index of baroreflex function, decreased with increasing age. Findings suggest that the increase in BP variability in hypertensive and elderly subjects may be explained, in part, by a disturbance of baroreflex function associated with an increase in arterial stiffness due to aging and hypertension. Am J Hypertens 1997;10:1281–1289 © 1997 American Journal of Hypertension, Ltd.
Received September 18, 1996. Accepted May 5, 1997. From the Department of Medicine (YI, AA, NM), Department of Public Health (TO, IT, SH), and Department of Environmental Health Science (HS), Tohoku University School of Medicine, Sendai, and Ohasama Hospital (KN), Iwate, Japan. This work was supported by a Research Grant from the Miyagi Prefecture Kidney Association, by a Research Grant from Takeda Medical Research Foundation, by a Research Grant for Cardiovascular Disease (No. 4C-3, 5C-2) and a Research Grant entitled ‘‘Eval-
uation of the effect of drug treatment on hypertension and other chronic disease conditions in the elderly’’ from the Ministry of Health and Welfare, Kosei-Kagaku Kenkyuhi, 1996 and 1997, and Rojin Hoken Jigyo Suishin Hojokin, 1996, of the Ministry of Health and Welfare and by a Research Grant for Scientific Research (07670746 and 07670420) from the Ministry of Education, Science, and Culture of Japan. Address correspondence and reprint requests to Yutaka Imai, MD, The Second Department of Internal Medicine, Tohoku University School of Medicine, 1-1 Seiryomachi, Aobaku, Sendai, 980, Japan.
© 1997 by the American Journal of Hypertension, Ltd. Published by Elsevier Science, Inc.
KEY WORDS:
Ambulatory blood pressure, blood pressure variability, heart rate, standard deviation, variation coefficient, age, pulse pressure, general population.
0895-7061/97/$17.00 PII S0895-7061(97)00277-X
1282
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
IMAI ET AL
H
ypertension is usually diagnosed based on measurements of blood pressure (BP) recorded during a visit to a physician, ie, ‘‘office’’ or ‘‘casual’’ BP. Casual BP measurements are of limited value because they do not reflect the circadian variation in BP, the ‘‘white-coat effect,’’ regression to the mean, observer bias, and other factors. The average of multiple BP measurements obtained during 24-h ambulatory BP monitoring provides a reliable estimate of cardiovascular risk.1–7 Ambulatory BP monitoring can reflect the variability in BP as well as provide the average BP over a specific period. Recent studies have suggested that the BP variability predicts target organ damage and is of prognostic value in patients with hypertension.3,6,8,9 However, data on factors that affect BP variability in the general population are lacking. We analyzed data from an ambulatory BP monitoring study performed in a cohort in northeastern Japan to determine the factors that affect BP variability. METHODS Study Population The present study is part of a longitudinal observational study of subjects who participated in our ambulatory BP monitoring project in Ohasama Town, Iwate Prefecture, Japan, initiated in 1986. Ohasama Town, characterized as a rural community, had a total population of 8040 in 1991. An ambulatory BP monitoring study was conducted of the Uchikawame and Kamegamori regions of Ohasama between 1986 and 1991. The socioeconomic and demographic characteristics of Uchikawame region have been reported elsewhere.10 The geographic and socioeconomic characteristics of Kamegamori are similar to those of Uchikawame. The most common cause of death among the residents of this town was cerebrovascular disease, followed by cancer and heart disease. The standardized mortality ratio (SMR) of the population of Ohasama Town for the period from 1988 to 1992 was 0.98 for all-cause mortality, 1.31 for cerebrovascular disease, 0.99 for cancer, and 0.59 for heart disease, when compared with mortality in the Japanese population at large. The study was approved by the Institutional Review Board of Tohoku University School of Medicine. Study participants had to be at least 20 years old and had to work near or stay in their homes during the daytime because public health nurses visited subjects to attach ambulatory BP monitoring devices during the daytime on work days (from Monday through Friday). We excluded 1040 of the 2728 individuals aged 20 or older because they worked away from home. Of the remaining 1688 individuals, 23 bedridden and 90 hospitalized residents were excluded. Thus, 1575 individuals were eligible for the study. This group consisted mainly of farmers, housewives,
and retirees. Of the eligible 1575 individuals, 383 declined to participate in the study for various reasons. Thus, 1192 subjects (395 men aged 61.7 6 13.4 years and 797 women aged 57.8 6 13.5 years) participated in the ambulatory BP monitoring projects, representing 75.7% of the total number of eligible individuals. There were no significant differences in the level of education or in the SMR for the period from 1988 to 1992 (P 5 .76 for all-cause mortality) between study subjects and residents who were excluded or declined to participate, suggesting that a selection bias was unlikely and that the subjects were representative of the general population in this community. Of the 1192 subjects, 823 (274 men aged 57.6 6 13.2 years and 549 women aged 53.5 6 12.6 years) were not receiving antihypertensive drugs. Information regarding antihypertensive medication was obtained from questionnaires sent to each household and from medical records at Ohasama Prefectural Hospital. The average ambulatory BP was 124.4 6 13.2/73.6 6 8.0 mm Hg in the 274 men and 118.4 6 12.5/69.3 6 7.3 mm Hg in the 549 women. Individuals with a history of stroke, coronary heart disease, diabetes mellitus, or other major complications were eligible unless they were taking antihypertensive medications or met the exclusion criteria described above. We initially analyzed ambulatory BP monitoring in these 823 untreated subjects. Subjects were classified according to age (Table 1). Data obtained in 706 subjects (217 men aged 60.2 6 12.2 years and 489 women aged 54.7 6 11.6 years) in whom ambulatory BP and screening BP data and body mass index (BMI) were available were analyzed by multiple regression analysis. Blood Pressure Measurements During November and December of both 1987 and 1989, a physician and public health nurses conducted health education classes regarding ambulatory BP monitoring in Uchikawame and Kamegamori. Of the 796 households in these regions, representatives of 637 (80%) attended the classes. Public health nurses visited the remaining households to provide information on ambulatory BP monitoring. Ambulatory BP was recorded using an ABPM 630 (Nippon Colin, Komaki, Tokyo, Japan), a fully automatic device preset to measure BP every 30 min. Although BP was measured simultaneously by the cuff-oscillometric method and the microphone method, we used only data obtained by the cuff-oscillometric method. Subjects aged 20 years or older in the Uchikawame region underwent a medical examination during July or August of either 1988 or 1989. Those in the Kamegamori region underwent an examination in July or August of 1990 or 1991. Blood pressure was measured by nurses or technicians twice consecutively with the subjects seated after at least 2 min of rest at 9 am and
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
BLOOD PRESSURE VARIABILITY
TABLE 1. SUBJECTS CLASSIFIED ACCORDING TO YEARS OF AGE Subjects
<29
30–39
40–49
50–59
60–69
>70
Men Women Total
7 21 28
23 65 88
39 105 144
63 173 236
91 133 224
51 52 103
12 noon or at 1 pm and 4 pm using an automatic BP measuring device (USM 700F, UEDA Electric Work Co. Ltd., Tokyo, Japan) based on the microphone method. We used a standard arm cuff to obtain ambulatory and screening BPs because the circumference of subjects’ arms was , 34 cm in most cases. Three subjects whose arm circumference exceeded 35 cm were excluded from the study. The BP measurement devices used have been previously validated.11,12 Analysis of Data We analyzed only the ambulatory BP data that were obtained during more than 8 h during the waking period (daytime) and more than 4 h during the time the subject was in bed (nighttime). Daytime and nighttime periods were determined from the subject’s diary. If the 24-hour ambulatory BP monitoring data were incomplete, the average 24-hour ambulatory BP was calculated as follows: 24-h average ambulatory BP 5 (daytime average 3 waking hours 1 nighttime average 3 sleep hours)/24, where the ‘‘sleep’’ hours were those spent in bed. The mean monitoring time was 23.4 6 1.6 h; we obtained 46.5 6 3.8 measurements in 823 subjects. Daytime and nighttime average ambulatory BPs were also calculated. The quality of sleep at night was disregarded because the study is based on a large population survey. Artifacts that occurred during monitoring were defined according to previously described criteria13 and were omitted from analysis. We calculated the standard deviations (SD) of 24-h, daytime, and nighttime ambulatory BP and the variation coefficients (VC) of BP (SD/average ambulatory BP). The SD and VC of BP measured every 30 min for the 24-h, daytime, and nighttime period were used as indexes of BP variability. The variability of BP during a 24-h period is usually defined as circadian BP variation and reflects mainly the day–night BP variation. In the present study we did not focus on circadian BP variation but on every 30-min BP variability (shortterm BP variability). Short-term BP variability, including sporadic and random variations as well as physiological variations, should be examined by beat-tobeat measurements of BP.14 However, for conceptual reasons, BP variability based on 30-min measurements is defined as short-term BP variability in the present study. Parameters were first analyzed by bivariate linear regression analysis and then by multivariate
1283
stepwise linear regression analysis to obtain the bestfit model using the most important independent variables influencing BP variability (SD or VC of ambulatory BP) as the dependent variables. Analysis was performed with the SAS PHREG procedure.15 Age, sex, screening BP, ambulatory BP (averaged for 24 h, daytime and nighttime), heart rate (HR, averaged for 24 h, daytime and nighttime), SD of HR, pulse pressure, magnitude of nocturnal fall in BP (daytime average ambulatory BP 2 nighttime average ambulatory BP), and difference between screening and ambulatory BP levels (the white-coat effect) were used as independent variables. Data are reported as the mean 6 SD. A P , .05 was accepted as statistically significant. RESULTS Bivariate Linear Regression Analysis The results of bivariate regression analysis of factors affecting the BP variability are shown in Table 2. Relationship Between Age and BP Variability The SDs of both the 24-h systolic BP (SBP) and 24-h diastolic BP (DBP) increased significantly with age but the increase was more marked with SBP (Figure 1A, Table 2). The VC of 24-h SBP also increased significantly with increasing age, but the age-dependent increase in the VC of DBP was less than the increase in the SD (Figure 1B, Table 2). The SD and the VC of daytime SBP and DBP increased with age (SBP: SD, r 5 0.419, P , .0001; VC, r 5 0.332, P , .0001; DBP: SD, r 5 0.297, P , .0001; VC, r 5 0.221, P , .0001). This trend was also observed for nighttime SBP and DBP (SBP: SD, r 5 0.323, P , .0001; VC, r 5 0.208, P , .0001; DBP: SD, r 5 0.226, P , .0001; VC, r 5 0.139, P , .0001). Age-Dependent Changes in Other Variables Both the 24-h SBP (r 5 0.351, P , .0001) and the 24-h DBP (r 5 0.147, P , .0001) increased with increasing age. The 24-h HR and the SD of 24-h HR decreased with increasing age (HR, r 5 20.226, P , .0001; SD of HR, r 5 20.187, P , .0001). Pulse pressure increased with increasing age (r 5 0.408, P , .0001). The magnitude of the nocturnal fall in SBP (r 5 20.073, P , .04) and DBP (r 5 20.070, P , .005) decreased with increasing age. The difference between the screening SBP and the ambulatory SBP (white-coat effect) increased with increasing age (SBP: r 5 0.093, P , .02). Relationship Between Ambulatory BP Level and BP Variability The SD of 24-h SBP increased with increase in 24-h SBP. The SD of 24-h DBP also increased with increase in 24-h DBP (Figure 2A, Table 2). However, the VCs of 24-h SBP and 24-h DBP were not related to increases in BP (Figure 2B, Table 2). These trends were also observed for the SD and VC of daytime BP (SBP, SD: r 5 0.429, P , .0001; VC: r 5 0.072, P , .04; DBP,
1284
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
IMAI ET AL
TABLE 2. BIVARIATE REGRESSION ANALYSIS OF FACTORS AFFECTING BLOOD PRESSURE VARIABILITY 24-H SBP
24-H DBP
SD (mm Hg)
Variable Nocturnal fall in BP (mm Hg) Pulse pressure (mm Hg) 24-h BP (mm Hg) Screening BP (mm Hg) Age (years) BMI (kg/m2) Sex Casual BP 2 24 h BP (mm Hg) 24-h HR (beats/min) SD of 24-h HR (beats/min)
VC (%)
SD (mm Hg)
VC (%)
r
P
r
P
r
P
r
P
0.617 0.473 0.458 0.338 0.332 0.23 20.106 0.036 20.04 0.063
.0001 .0001 .0001 .0001 .0001 .0001 .002 .3356 .03 .07
0.603 0.137 0.068 0.133 0.221 0.18 20.212 0.114 20.024 0.12
.0001 .0001 .0001 .0004 .0001 .0001 .0001 .003 .5 .0005
0.629 0.449 0.35 0.316 0.232 0.237 20.061 0.015 0.016 0.116
.0001 .0001 .0001 .0001 .0001 .0001 .08 .7 .6 .008
0.5 0.192 20.117 0.06 0.136 0.152 20.187 0.16 20.008 0.192
.0001 .0001 .0008 .107 .0001 .0001 .0001 .0001 .8 .0001
BP, blood pressure; SBP, systolic blood pressure; DBP, diastolic blood pressure; SD, standard deviation; VC, variation coefficient; HR, heart rate, BMI, body mass index. Sex: men 5 1, women 5 0.
SD: r 5 0.197, P , .0001; VC: r 5 20.172, P , .0001) and nighttime BP (SBP, SD: r 5 0.318, P , .0001; VC: r 5 20.030, P 5 .39; DBP, SD: r 5 0.221, P , .0001; VC r 5 20.126, P , .0003). To exclude the influence of age on the relationship between BP variability and ambulatory BP, we analyzed data for subjects classified according to age. The
FIGURE 1. Age-dependent changes in 24-h blood pressure variability. A: SD, standard deviation. B: VC, variation coefficient (SD/average ambulatory blood pressure).
SD of 24-h SBP increased with increases in the ambulatory SBP in each age group (Figure 3, #39 years: r 5 0.409, P , .0001; $40 and #59 years: r 5 0.395, P , .0001; $60 years: r 5 0.390, P , .0001). The DBP also demonstrated these trends (#39 years: r 5 0.297, P , .0001; $40 and #59 years: r 5 0.395, P , .0001, age $60 years: r 5 0.334, P , .0001). The VC of 24-h SBP
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
BLOOD PRESSURE VARIABILITY
1285
FIGURE 2. Blood pressure– dependent changes in 24-h blood pressure variability. A: SD, standard deviation. B: VC, variation coefficient (SD/average ambulatory blood pressure).
did not increase with increasing SBP (#39 years, r 5 0.029, P 5 .75; $40 and #59 years, r 5 20.015, P 5 .77; $60 years, r 5 0.0003, P 5 .996) and the VC of 24-h DBP decreased with increasing DBP (#39 years, r 5 20.209, P , .03; $40 and #59 years, r 5 20.156, P , .003; $60 years, r 5 20.132, P , .02). BP Dependent Changes in Other Variables Pulse pressure was significantly correlated with the 24-h SBP (r 5 0.875, P , .0001) and DBP (r 5 0.584, P , .0001). The magnitude of the nocturnal decline in BP (SBP: r 5 0.171, P , .0001; DBP: r 5 0.230, P , .0001) and the BMI (SBP: r 5 0.188, P , .0001; DBP, r 5 0.214, P , .0001) were positively correlated with the 24-h BP. The white-coat effect was negatively correlated with the 24-h BP (SBP: r 5 20.171, P , .0001; DBP, r 5 20.172, P , .0001). Daytime and nighttime BPs were also negatively correlated with the white-coat effect. Multivariate Stepwise Linear Regression Analysis Stepwise linear regression analysis showed that age, sex, BMI, pulse pressure, white-coat effect, and magnitude of the nocturnal fall in BP were significantly and independently related to the SD of 24-h SBP (Table 3). BP variability was significantly greater in women. Although the magnitude of the nocturnal decline in BP was a strong predictor of the SD of 24-h SBP, multivariate stepwise linear regression analysis
demonstrated that pulse pressure, age, sex, BMI, and white-coat effect were still independently correlated with the SD of the 24-h SBP. Similarly, multivariate regression analysis showed that the magnitude of the nocturnal decline in DBP, pulse pressure, white-coat effect, 24-h HR, age, BMI, and sex were significantly correlated with the SD of 24-h DBP (Table 4). Pulse pressure was a better predictor of the SDs of 24-h SBP and DBP than were age, sex, BMI, and white-coat effect. Pulse pressure was the strongest predictor of the SDs of daytime SBP and DBP (SD of daytime SBP: b-coefficient 5 0.246, SE 5 0.019, P , .0001, R2 5 0.27; SD of daytime DBP: b-coefficient 5 0.197, SE 5 0.015, P , .0001, R2 5 0.227). The following factors were independently correlated, in descending order of magnitude, with the SD of daytime SBP: pulse pressure, age, sex, and BMI. The following factors were independently correlated, in descending order of magnitude, with the SD of daytime DBP: pulse pressure, magnitude of nocturnal fall in BP, daytime HR, age, BMI, and daytime DBP. Age was the best predictor of the SD of nighttime SBP (b-coefficient 5 0.073, SE 5 0.01, P , .0001, R2 5 0.104). The SD of nighttime SBP was independently correlated, in descending order of magnitude, with age, nighttime average of SBP, magnitude of nocturnal decline in BP, nighttime HR, and BMI, although their predictive value was mini-
1286
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
IMAI ET AL
FIGURE 3. Blood pressure– dependent changes in 24-h blood pressure (SD, standard deviation) according to age. (A) Patients $39 years, (B) Patients $40 and #59 years, (C) Patients $60 years.
mal. The SD of nighttime DBP was independently correlated, in descending order of magnitude, with the nighttime DBP, magnitude of the nocturnal decline in BP, nighttime HR, age, BMI, and pulse pressure, although their predictive value was minimal. DISCUSSION In the present study, the SD and VC of BP measured every 30 min for 24-h, daytime, and nighttime periods
were used as indexes of short-term BP variability. Short-term BP variability, including sporadic and random variations as well as physiological variations, should be examined by beat-to-beat measurements of BP.15 DiRienzo et al15 reported that when the intervals between the intermittent measurements were 5, 10, or 15 min, the 24-h SD of BP obtained by intermittent measurements corresponded to the 24-h SD of BP obtained by continuous measurements. When the in-
TABLE 3. MULTIVARIATE STEPWISE LINEAR REGRESSION ANALYSIS OF THE SD OF 24-H SBP Variable Nocturnal fall in SBP (mm Hg) Pulse pressure (mm Hg) Age (years) Sex BMI (kg/m2) Casual SBP 2 24 h SBP (mm Hg)
Coefficient
SE
P
Partial R2
0.238 0.197 0.089 21.180 0.130 0.015
0.010 0.016 0.009 0.224 0.034 0.007
.0001 .0001 .0001 .0001 .0003 .0338
0.370 0.170 0.042 0.018 0.010 0.003 Model R2 0.612
BMI, body mass index; SE, standard error; SBP, systolic blood pressure. The independent variables entered in the model for multivariate regression analysis were cited in the text. The variables that showed P $ .05 in the multivariate relations were excluded from the table. Sex: men 5 1, women 5 0.
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
BLOOD PRESSURE VARIABILITY
1287
TABLE 4. MULTIVARIATE STEPWISE LINEAR REGRESSION ANALYSIS OF THE SD OF 24-h DBP Variables Nocturnal fall in DBP (mm Hg) Pulse pressure (mm Hg) Casual DBP 2 24 h DBP (mm Hg) 24 h HR (beats/min) Age (years) BMI (kg/m2) Sex
Coefficient
S.E.
P
Partial R2
0.218 0.127 0.032 0.043 0.035 0.084 20.353
0.011 0.011 0.007 0.009 0.006 0.022 0.142
.0001 .0001 .0001 .0001 .0001 .0001 .0130
0.326 0.153 0.022 0.012 0.016 0.012 0.004 Model R2 0.544
BMI: body mass index; S.E.: standard error; DBP: diastolic blood pressure; HR: heart rate. The independent variables entered in the model for multivariate regression analysis were cited in the text. Sex: men 5 1, women 5 0. The variables which showed P $ .05 in the multivariate relations were excluded from the table.
terval between intermittent measurements was 30 or 60 min, there was a marked difference in the 24-h SD between intermittent and continuous measurements in some subjects.15 However, indirect ambulatory BP monitoring using a device preset to measure BP every 30 min is widely used in clinical practice. Therefore, we examined the characteristics of BP variability based on every 30-min measurements. Bivariate Linear Regression Analysis It is generally accepted that BP variability is correlated with the BP level,16 –21 although Pickering et al did not confirm this relationship.22 In the present study, the SD of BP was significantly and positively correlated with the ambulatory BP. Mancia et al reported that the SD, but not the VC, of BP was greater in hypertensive subjects than in age-matched normotensive subjects.21 In the present study, we also found that the VC of SBP did not increase with increasing SBP, suggesting that an increase in BP variability is a function of the BP level. BP variability is also reportedly correlated with age.18 –24 However, the effect of age on BP variability does not appear to be independent of the concomitant increase in BP level. Brennan et al25 reported that there was no difference in BP variability between young and elderly subjects with the same BP level. However, the present results showed an age-dependent increase in the VC of SBP, an index of BP variability corrected by the ambulatory BP level. The increase in the SD of SBP associated with increasing ambulatory SBP was observed in all age groups, suggesting that the BP level and age independently affected BP variability. The increase in BP variability in hypertensive and elderly subjects may be partly explained by the diminished baroreflex sensitivity associated with increased arterial stiffness due to aging and hypertension,26 –31 although other factors (probably central in nature32) may also be involved.18,21,32 Conway et al33 found that subjects with elevated baroreflex sensitivity showed
greater HR variability. Although the effect of age on the SD of HR is controversial,10,18,25,34 in the present study, the SD of HR decreased with age, suggesting that aging was associated with diminished baroreflex function. Disturbed baroreflex function is related to exaggerated pressor responses to the mental and physical stimuli and causes orthostatic and postprandial hypotension,35–38 resulting in an increase in BP variability. Another possible explanation for the age-dependent increase in SBP variability is that any change in stroke volume in elderly individuals is likely to induce a greater change in BP because the arterial wall is less compliant in the elderly. Age- and BP-dependent increases in SBP variability were greater than the corresponding changes in DBP variability in the present study. A change in stroke volume, which is mediated mainly by a change in the activity of the sympathetic nervous system and is caused by changes in mental and physical activities, directly reflects SBP, but not DBP, variability. Zachariah et al previously found that there was a lesser degree of variability in DBP compared with SBP.39 The present findings are consistent with their results. Mental and physical activity are generally decreased in elderly subjects, resulting in a lesser degree of BP variability. However, age-dependent increases in BP variability were observed in the present study, not only for the 24-h BP and the daytime BP but also for the nighttime BP, indicating that BP variability is mediated by changes in mental and physical activity and by changes in intrinsic vasomotor activity. The increased variability in vasomotor activity in association with decreased vascular compliance in elderly subjects may result in a greater variability in BP in elderly individuals than in younger individuals.
1288
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
IMAI ET AL
Multivariate Stepwise Linear Regression Analysis To exclude the influence of circadian BP variation from 24-h BP variability and to confirm the independent contributions of age and BP to BP variability, we performed multivariate stepwise linear regression analysis. We also performed multivariate stepwise regression analysis using daytime and nighttime BP variability as dependent variables. The magnitude of the nocturnal decline in BP, which is an index of circadian BP variation but not of short-term BP variability, was the strongest predictor of the SD of 24-h BP, suggesting that the SD of 24-h BP is not an appropriate index of short-term BP variability. Previous studies examining the relationship between BP variability and prognosis in hypertensive subjects have usually used the SD of 24-h BP as an index of BP variability.6,8,9 Thus, the reported relationship between BP variability and prognosis may depend mainly on the circadian BP variation. Although the nocturnal decline in BP was the strongest predictor of the 24-h BP, pulse pressure was still a strong predictor of 24-h BP variability. The contributions of age, SBP, and DBP to BP variability were smaller than that of pulse pressure. Pulse pressure increases with increasing age, especially after the age of 60 years,10 because of the associated decrease in compliance of the large elastic arteries. Thus, an increase in pulse pressure reflects aging and other factors that contribute to stiffness. Decreased compliance of the large elastic artery may lead to disturbed baroreceptor reflex function, which in turn results in increased variability in SBP in response to changes in the stroke volume in elderly subjects. It is interesting that the SD of HR, an index of baroreflex function, was not identified as an independent variable for SBP variability when pulse pressure was included in the multivariate analysis. These findings suggest that decreased compliance of the large elastic artery was a major independent contributor to BP variability. Age was the strongest predictor of the SD of nighttime SBP, whereas nighttime DBP was the strongest predictor of the SD of nighttime DBP. Daytime SBP variability may have resulted from changes in stroke volume due to a change in sympathetic nerve activity and to decreased compliance of the large elastic artery. However, changes in stroke volume are minimal during sleep; therefore, age and the BP level per se may have a greater effect on BP variability at night. It is interesting that BMI was an independent variable for 24-h, daytime, and nighttime BP variability and for SBP and DBP. Cardiac autonomic dysfunction, especially vagal dysfunction, has been observed in obese subjects,36,37 which may explain the increased variability in BP in obese subjects in the present study. The contribution of the white-coat effect for the SD of
the 24-h SBP was minimal in the model, although the association was statistically significant (Table 3). Study Limitations There are a number of possible limitations to the present study. Our data may not be applicable to other geographic or racial groups. We excluded treated hypertensive subjects from the cohort, which could potentially have led to underestimation of BP variability. In addition, we did not assess the effects of sleep disturbances on nocturnal BP variability. Although this problem could be addressed in a future study by monitoring the subjects’ sleep, we believe that the large number of subjects in the present study should have eliminated the influence of sleep disturbance. It must be emphasized that multiple logistic regression analysis can only suggest associations between dependent and independent variables and cannot prove a cause-and-effect relationship. Finally, we would like to emphasize again that the BP variability in the present study represents 30-min BP variability and not beat-to-beat BP variability. In summary, multivariate stepwise linear regression analysis showed that pulse pressure, an index of aging and arterial stiffness, was a strong predictor of BP variability in the overall study population. Because the nocturnal fall in BP was the major determinant of 24-h ambulatory BP, the SD of 24-h BP should not be used as an index of short-term BP variability. Future studies are needed to address the relationship between BP variability and cardiovascular events. REFERENCES 1.
Sokolow M, Werdegar D, Kain HK, Hinman AT: Relationship between level of blood pressure measured casually and by portable recorders and severity of complications in essential hypertension. Circulation 1966; 34:279 –307.
2.
Julius S, Ellis CN, Pascual AV, et al: Home blood pressure determination. Value in borderline (‘‘labile’’) hypertension. JAMA 1974;229:663– 666.
3.
Perloff D, Sokolow M, Cowan R: The prognostic value of ambulatory blood pressures. JAMA 1983;249:2792– 2798.
4.
Conway J: Home blood pressure recording. Clin Exp Hypertens 1986;A8:1247–1294.
5.
Devereux RB, Pickering TG, Harshfield GA, et al: Left ventricular hypertrophy in patients with hypertension: importance of blood pressure response to regularly recurring stress. Circulation 1983;68:470 – 476.
6.
Parati G, Pomidossi G, Albini F, et al: Relationship of 24-hour blood pressure mean and variability to severity of target-organ damage in hypertension. J Hypertens 1987;5:93–98.
7.
White WB, Schulman P, McCabe EJ, Dey HM: Average daily blood pressure, not office blood pressure, determines cardiac function in patients with hypertension. JAMA 1989;261:873– 877.
8.
Frattola A, Parati G, Cuspidi C, et al: Prognostic value
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
of 24-hour blood pressure variability. J Hypertens 1993; 11:1133–1137.
BLOOD PRESSURE VARIABILITY
1289
25.
Pickering TG, James GD: Ambulatory blood pressure and prognosis. J Hypertens 1994;12(suppl 8):S29 –S33.
Brennan M, O’Brien E, O’Malley K: The effect of age on blood pressure and heart rate variability in hypertension. J Hypertens 1986;4(suppl 6):S269 –S272.
26.
10.
Imai Y, Nagai K, Sakuma M, et al: Ambulatory blood pressure of adults in Ohasama, Japan. Hypertension 1993;22:900 –912.
McGarry K, Laher M, Fitzgerald D, et al: Baroreflex function in elderly hypertensives. Hypertension 1983; 5:763–766.
27.
11.
Imai Y, Abe K, Sasaki S, et al: Determination of clinical accuracy and nocturnal blood pressure pattern by new portable device for monitoring indirect ambulatory blood pressure. Am J Hypertens 1990;3:293–301.
Gribbin B, Pickering TG, Sleight P, Peto R: Effect of age and high blood pressure on baroreflex sensitivity in man. Circulation 1971;29:424 – 431.
28.
Shimada K, Kitazumi T, Ogura H, et al: Differences in age-independent effects of blood pressure on baroreflex sensitivity between normal and hypertensive subjects. Clin Sci 1986;70:489 – 494.
29.
Floras JS, Hassan MO, Jones JV, et al: Consequences of impaired baroreflexes in essential hypertension: effects on pressor responses, plasma noradrenaline and blood pressure variability. J Hypertens 1988;6:525–535.
30.
Avolio AP, Chen SG, Wang RP, et al: Effects of aging on changing arterial compliance and left ventricular load in a northern Chinese urban community. Circulation 1983;68:50 –58.
31.
O’Rourke M: Arterial stiffness, systolic blood pressure, and logical treatment of arterial hypertension. Hypertension 1990;15:339 –347.
32.
Mancia G, Parati G, Pomidossi G, et al: Arterial baroreflexes and blood pressure and heart rate variabilities in humans. Hypertension 1986;8:147–153.
9.
12.
13.
Imai Y, Abe K, Sekino H: Evaluation of automatic devices for blood pressure measurement, in Koshikawa S, Nagasama T (eds): Annual Review Nephrology 1992. Chugai-Igakusha, Tokyo, 1991, pp 55– 61 (in Japanese). Imai Y, Nihei M, Abe K, et al: A finger volume-oscillometric device for monitoring ambulatory blood pressure: laboratory and clinical evaluation. Clin Exp Hypertens 1987;A9:2001–2025.
14.
SAS Institute Inc: SAS/STAT User’s Guide, Release 6.03 Edition. SAS Institute Inc., Cary, NC, 1988.
15.
DiRienzo M, Grassi G, Pedotti A, Mancia G: Continuous vs intermittent blood pressure measurements in estimating 24-hour average blood pressure. J Hypertens 1983;5:264 –269.
16.
Clement DL: Blood pressure variability in hospitalized patients. Acta Clin Belg 1977;32:163–167.
17.
Goldberg AD, Raftery EB, Cashman MM, Stott FD: Study of untreated hypertensive subjects by means of continuous intra-arterial blood pressure recordings. Br Heart J 1978;40:656 – 664.
33.
Conway J, Boon N, Vann Jones J, Sleight P: Mechanisms concerned with blood pressure variability throughout the day. Clin Exp Hypertens 1985;A7:153– 157.
18.
Mancia G, Ferrari A, Gregorini L: Blood pressure variability in man: its relation to high blood pressure, age and baroreflex sensitivity. Clin Sci 1980;59(suppl 1): 401S– 404S.
34.
Kawano A, Shimada K, Matsubayashi K, et al: Cardiovascular regulatory function in elderly patients with hypertension. Hypertension 1989;13:401– 407.
19.
Watson RDS, Stallard TJ, Flinn RM, Littler WA: Factors determining direct arterial pressure and its variability in hypertensive man. Hypertension 1980;2:333–341.
35.
Tsuchihashi T, Abe I, Tsukashima A, et al: Effects of meal and physical activity on blood pressure variability in elderly patients: a preliminary study. Am J Hypertens 1990;3:943–946.
20.
Floras JS, Sleight P: The lability of blood pressure, in Amery A, Fagard R, Lijnen P, Staessen J (eds): Hypertensive Cardiovascular Disease: Pathophysiology and Treatment. Martinus Nijhof, The Hague, 1982, pp 104 – 117.
36.
Peitzman SJ, Berger SR: Postprandial blood pressure decrease in well elderly persons. Arch Intern Med 1989; 149:286 –288.
37.
Lipsitz LA, Fullerton KJ: Postprandial blood pressure reduction in healthy elderly. J Am Geriatr Soc 1986;34: 267–270.
38.
Caird FI, Andrews GR, Kennedy RD: Effects of posture on blood pressure in the elderly. Br Heart J 1973;35: 527–530.
39.
Zachariah PK, Sheps SG, Bailey KR, et al: Age-related characteristics of ambulatory blood pressure load and mean blood pressure in normotensive subjects. JAMA 1991;265:1414 –1417.
40.
Peterson HR, Rothschild M, Weinberg CR, et al: Body fat and the activity of the autonomic nervous system. N Eng J Med 1988;318:1077–1083.
41.
Rossi M, Marti G, Ricordi L, et al: Cardiac autonomic dysfunction in obese subjects. Clin Sci 1989;76:567–572.
21.
22.
23.
24.
Mancia G, Ferrari A, Gregorini L, et al: Blood pressure and heart rate variabilities in normotensive and hypertensive human beings. Circ Res 1983;53:96 –104. Pickering TG, Harshfield GA, Kleinert HD, et al: Blood pressure during normal daily activities, sleep, and exercise. Comparison of values in normal and hypertensive subjects. JAMA 1982;247:992–996. Drayer JIM, Weber MA, DeYoung JL, Wyle FA: Circadian blood pressure patterns in ambulatory hypertensive patients. Am J Med 1982;73:493– 499. Canonico V, DeCaprio L, Vigorito C, et al: Differences in blood pressure profile between young and elderly hypertensive patients. J Hum Hypertens 1990;4:405– 409.
1997;10:1281–1289
Factors That Affect Blood Pressure Variability A Community-Based Study in Ohasama, Japan Yutaka Imai, Akiko Aihara, Takayoshi Ohkubo, Kenichi Nagai, Ichiro Tsuji, Naoyoshi Minami, Hiroshi Satoh, and Shigeru Hisamichi
Factors that affect blood pressure (BP) variability, ie, standard deviation (SD) and variation coefficient (VC: SD/average ambulatory BP) of ambulatory BP, were examined in a communitybased sample in northeastern Japan. Screening and ambulatory BPs were measured in 823 subjects >20 years of age, and the effects of age and BP on the SD and the VC were examined. In bivariate regression analysis, the SD of ambulatory BP was positively correlated with age and the ambulatory BP. The VC was also correlated with age. Both the SD and the VC were strongly correlated with the magnitude of the nocturnal decline in BP. Ambulatory BP was positively correlated with age and negatively correlated with heart rate and the SD of heart rate. Multivariate analysis demonstrated that the nocturnal decline in BP showed the strongest association with the SD and the VC of 24-h BP. However, age and BP were still independently and positively associated with the SD and the VC of ambulatory BP. Furthermore, pulse pressure and BMI were independently and
positively associated with the SD and the VC of ambulatory BP. Since the SD and the VC of 24-h ambulatory BP were determined mainly by the nocturnal decline in BP, this variable appears to be an index of the circadian variation in BP and not an index of short-term BP variability. Pulse pressure, an index of arterial stiffness, was a relatively strong predictor of the SD and the VC of BP. In addition, the SD of heart rate, an index of baroreflex function, decreased with increasing age. Findings suggest that the increase in BP variability in hypertensive and elderly subjects may be explained, in part, by a disturbance of baroreflex function associated with an increase in arterial stiffness due to aging and hypertension. Am J Hypertens 1997;10:1281–1289 © 1997 American Journal of Hypertension, Ltd.
Received September 18, 1996. Accepted May 5, 1997. From the Department of Medicine (YI, AA, NM), Department of Public Health (TO, IT, SH), and Department of Environmental Health Science (HS), Tohoku University School of Medicine, Sendai, and Ohasama Hospital (KN), Iwate, Japan. This work was supported by a Research Grant from the Miyagi Prefecture Kidney Association, by a Research Grant from Takeda Medical Research Foundation, by a Research Grant for Cardiovascular Disease (No. 4C-3, 5C-2) and a Research Grant entitled ‘‘Eval-
uation of the effect of drug treatment on hypertension and other chronic disease conditions in the elderly’’ from the Ministry of Health and Welfare, Kosei-Kagaku Kenkyuhi, 1996 and 1997, and Rojin Hoken Jigyo Suishin Hojokin, 1996, of the Ministry of Health and Welfare and by a Research Grant for Scientific Research (07670746 and 07670420) from the Ministry of Education, Science, and Culture of Japan. Address correspondence and reprint requests to Yutaka Imai, MD, The Second Department of Internal Medicine, Tohoku University School of Medicine, 1-1 Seiryomachi, Aobaku, Sendai, 980, Japan.
© 1997 by the American Journal of Hypertension, Ltd. Published by Elsevier Science, Inc.
KEY WORDS:
Ambulatory blood pressure, blood pressure variability, heart rate, standard deviation, variation coefficient, age, pulse pressure, general population.
0895-7061/97/$17.00 PII S0895-7061(97)00277-X
1282
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
IMAI ET AL
H
ypertension is usually diagnosed based on measurements of blood pressure (BP) recorded during a visit to a physician, ie, ‘‘office’’ or ‘‘casual’’ BP. Casual BP measurements are of limited value because they do not reflect the circadian variation in BP, the ‘‘white-coat effect,’’ regression to the mean, observer bias, and other factors. The average of multiple BP measurements obtained during 24-h ambulatory BP monitoring provides a reliable estimate of cardiovascular risk.1–7 Ambulatory BP monitoring can reflect the variability in BP as well as provide the average BP over a specific period. Recent studies have suggested that the BP variability predicts target organ damage and is of prognostic value in patients with hypertension.3,6,8,9 However, data on factors that affect BP variability in the general population are lacking. We analyzed data from an ambulatory BP monitoring study performed in a cohort in northeastern Japan to determine the factors that affect BP variability. METHODS Study Population The present study is part of a longitudinal observational study of subjects who participated in our ambulatory BP monitoring project in Ohasama Town, Iwate Prefecture, Japan, initiated in 1986. Ohasama Town, characterized as a rural community, had a total population of 8040 in 1991. An ambulatory BP monitoring study was conducted of the Uchikawame and Kamegamori regions of Ohasama between 1986 and 1991. The socioeconomic and demographic characteristics of Uchikawame region have been reported elsewhere.10 The geographic and socioeconomic characteristics of Kamegamori are similar to those of Uchikawame. The most common cause of death among the residents of this town was cerebrovascular disease, followed by cancer and heart disease. The standardized mortality ratio (SMR) of the population of Ohasama Town for the period from 1988 to 1992 was 0.98 for all-cause mortality, 1.31 for cerebrovascular disease, 0.99 for cancer, and 0.59 for heart disease, when compared with mortality in the Japanese population at large. The study was approved by the Institutional Review Board of Tohoku University School of Medicine. Study participants had to be at least 20 years old and had to work near or stay in their homes during the daytime because public health nurses visited subjects to attach ambulatory BP monitoring devices during the daytime on work days (from Monday through Friday). We excluded 1040 of the 2728 individuals aged 20 or older because they worked away from home. Of the remaining 1688 individuals, 23 bedridden and 90 hospitalized residents were excluded. Thus, 1575 individuals were eligible for the study. This group consisted mainly of farmers, housewives,
and retirees. Of the eligible 1575 individuals, 383 declined to participate in the study for various reasons. Thus, 1192 subjects (395 men aged 61.7 6 13.4 years and 797 women aged 57.8 6 13.5 years) participated in the ambulatory BP monitoring projects, representing 75.7% of the total number of eligible individuals. There were no significant differences in the level of education or in the SMR for the period from 1988 to 1992 (P 5 .76 for all-cause mortality) between study subjects and residents who were excluded or declined to participate, suggesting that a selection bias was unlikely and that the subjects were representative of the general population in this community. Of the 1192 subjects, 823 (274 men aged 57.6 6 13.2 years and 549 women aged 53.5 6 12.6 years) were not receiving antihypertensive drugs. Information regarding antihypertensive medication was obtained from questionnaires sent to each household and from medical records at Ohasama Prefectural Hospital. The average ambulatory BP was 124.4 6 13.2/73.6 6 8.0 mm Hg in the 274 men and 118.4 6 12.5/69.3 6 7.3 mm Hg in the 549 women. Individuals with a history of stroke, coronary heart disease, diabetes mellitus, or other major complications were eligible unless they were taking antihypertensive medications or met the exclusion criteria described above. We initially analyzed ambulatory BP monitoring in these 823 untreated subjects. Subjects were classified according to age (Table 1). Data obtained in 706 subjects (217 men aged 60.2 6 12.2 years and 489 women aged 54.7 6 11.6 years) in whom ambulatory BP and screening BP data and body mass index (BMI) were available were analyzed by multiple regression analysis. Blood Pressure Measurements During November and December of both 1987 and 1989, a physician and public health nurses conducted health education classes regarding ambulatory BP monitoring in Uchikawame and Kamegamori. Of the 796 households in these regions, representatives of 637 (80%) attended the classes. Public health nurses visited the remaining households to provide information on ambulatory BP monitoring. Ambulatory BP was recorded using an ABPM 630 (Nippon Colin, Komaki, Tokyo, Japan), a fully automatic device preset to measure BP every 30 min. Although BP was measured simultaneously by the cuff-oscillometric method and the microphone method, we used only data obtained by the cuff-oscillometric method. Subjects aged 20 years or older in the Uchikawame region underwent a medical examination during July or August of either 1988 or 1989. Those in the Kamegamori region underwent an examination in July or August of 1990 or 1991. Blood pressure was measured by nurses or technicians twice consecutively with the subjects seated after at least 2 min of rest at 9 am and
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
BLOOD PRESSURE VARIABILITY
TABLE 1. SUBJECTS CLASSIFIED ACCORDING TO YEARS OF AGE Subjects
<29
30–39
40–49
50–59
60–69
>70
Men Women Total
7 21 28
23 65 88
39 105 144
63 173 236
91 133 224
51 52 103
12 noon or at 1 pm and 4 pm using an automatic BP measuring device (USM 700F, UEDA Electric Work Co. Ltd., Tokyo, Japan) based on the microphone method. We used a standard arm cuff to obtain ambulatory and screening BPs because the circumference of subjects’ arms was , 34 cm in most cases. Three subjects whose arm circumference exceeded 35 cm were excluded from the study. The BP measurement devices used have been previously validated.11,12 Analysis of Data We analyzed only the ambulatory BP data that were obtained during more than 8 h during the waking period (daytime) and more than 4 h during the time the subject was in bed (nighttime). Daytime and nighttime periods were determined from the subject’s diary. If the 24-hour ambulatory BP monitoring data were incomplete, the average 24-hour ambulatory BP was calculated as follows: 24-h average ambulatory BP 5 (daytime average 3 waking hours 1 nighttime average 3 sleep hours)/24, where the ‘‘sleep’’ hours were those spent in bed. The mean monitoring time was 23.4 6 1.6 h; we obtained 46.5 6 3.8 measurements in 823 subjects. Daytime and nighttime average ambulatory BPs were also calculated. The quality of sleep at night was disregarded because the study is based on a large population survey. Artifacts that occurred during monitoring were defined according to previously described criteria13 and were omitted from analysis. We calculated the standard deviations (SD) of 24-h, daytime, and nighttime ambulatory BP and the variation coefficients (VC) of BP (SD/average ambulatory BP). The SD and VC of BP measured every 30 min for the 24-h, daytime, and nighttime period were used as indexes of BP variability. The variability of BP during a 24-h period is usually defined as circadian BP variation and reflects mainly the day–night BP variation. In the present study we did not focus on circadian BP variation but on every 30-min BP variability (shortterm BP variability). Short-term BP variability, including sporadic and random variations as well as physiological variations, should be examined by beat-tobeat measurements of BP.14 However, for conceptual reasons, BP variability based on 30-min measurements is defined as short-term BP variability in the present study. Parameters were first analyzed by bivariate linear regression analysis and then by multivariate
1283
stepwise linear regression analysis to obtain the bestfit model using the most important independent variables influencing BP variability (SD or VC of ambulatory BP) as the dependent variables. Analysis was performed with the SAS PHREG procedure.15 Age, sex, screening BP, ambulatory BP (averaged for 24 h, daytime and nighttime), heart rate (HR, averaged for 24 h, daytime and nighttime), SD of HR, pulse pressure, magnitude of nocturnal fall in BP (daytime average ambulatory BP 2 nighttime average ambulatory BP), and difference between screening and ambulatory BP levels (the white-coat effect) were used as independent variables. Data are reported as the mean 6 SD. A P , .05 was accepted as statistically significant. RESULTS Bivariate Linear Regression Analysis The results of bivariate regression analysis of factors affecting the BP variability are shown in Table 2. Relationship Between Age and BP Variability The SDs of both the 24-h systolic BP (SBP) and 24-h diastolic BP (DBP) increased significantly with age but the increase was more marked with SBP (Figure 1A, Table 2). The VC of 24-h SBP also increased significantly with increasing age, but the age-dependent increase in the VC of DBP was less than the increase in the SD (Figure 1B, Table 2). The SD and the VC of daytime SBP and DBP increased with age (SBP: SD, r 5 0.419, P , .0001; VC, r 5 0.332, P , .0001; DBP: SD, r 5 0.297, P , .0001; VC, r 5 0.221, P , .0001). This trend was also observed for nighttime SBP and DBP (SBP: SD, r 5 0.323, P , .0001; VC, r 5 0.208, P , .0001; DBP: SD, r 5 0.226, P , .0001; VC, r 5 0.139, P , .0001). Age-Dependent Changes in Other Variables Both the 24-h SBP (r 5 0.351, P , .0001) and the 24-h DBP (r 5 0.147, P , .0001) increased with increasing age. The 24-h HR and the SD of 24-h HR decreased with increasing age (HR, r 5 20.226, P , .0001; SD of HR, r 5 20.187, P , .0001). Pulse pressure increased with increasing age (r 5 0.408, P , .0001). The magnitude of the nocturnal fall in SBP (r 5 20.073, P , .04) and DBP (r 5 20.070, P , .005) decreased with increasing age. The difference between the screening SBP and the ambulatory SBP (white-coat effect) increased with increasing age (SBP: r 5 0.093, P , .02). Relationship Between Ambulatory BP Level and BP Variability The SD of 24-h SBP increased with increase in 24-h SBP. The SD of 24-h DBP also increased with increase in 24-h DBP (Figure 2A, Table 2). However, the VCs of 24-h SBP and 24-h DBP were not related to increases in BP (Figure 2B, Table 2). These trends were also observed for the SD and VC of daytime BP (SBP, SD: r 5 0.429, P , .0001; VC: r 5 0.072, P , .04; DBP,
1284
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
IMAI ET AL
TABLE 2. BIVARIATE REGRESSION ANALYSIS OF FACTORS AFFECTING BLOOD PRESSURE VARIABILITY 24-H SBP
24-H DBP
SD (mm Hg)
Variable Nocturnal fall in BP (mm Hg) Pulse pressure (mm Hg) 24-h BP (mm Hg) Screening BP (mm Hg) Age (years) BMI (kg/m2) Sex Casual BP 2 24 h BP (mm Hg) 24-h HR (beats/min) SD of 24-h HR (beats/min)
VC (%)
SD (mm Hg)
VC (%)
r
P
r
P
r
P
r
P
0.617 0.473 0.458 0.338 0.332 0.23 20.106 0.036 20.04 0.063
.0001 .0001 .0001 .0001 .0001 .0001 .002 .3356 .03 .07
0.603 0.137 0.068 0.133 0.221 0.18 20.212 0.114 20.024 0.12
.0001 .0001 .0001 .0004 .0001 .0001 .0001 .003 .5 .0005
0.629 0.449 0.35 0.316 0.232 0.237 20.061 0.015 0.016 0.116
.0001 .0001 .0001 .0001 .0001 .0001 .08 .7 .6 .008
0.5 0.192 20.117 0.06 0.136 0.152 20.187 0.16 20.008 0.192
.0001 .0001 .0008 .107 .0001 .0001 .0001 .0001 .8 .0001
BP, blood pressure; SBP, systolic blood pressure; DBP, diastolic blood pressure; SD, standard deviation; VC, variation coefficient; HR, heart rate, BMI, body mass index. Sex: men 5 1, women 5 0.
SD: r 5 0.197, P , .0001; VC: r 5 20.172, P , .0001) and nighttime BP (SBP, SD: r 5 0.318, P , .0001; VC: r 5 20.030, P 5 .39; DBP, SD: r 5 0.221, P , .0001; VC r 5 20.126, P , .0003). To exclude the influence of age on the relationship between BP variability and ambulatory BP, we analyzed data for subjects classified according to age. The
FIGURE 1. Age-dependent changes in 24-h blood pressure variability. A: SD, standard deviation. B: VC, variation coefficient (SD/average ambulatory blood pressure).
SD of 24-h SBP increased with increases in the ambulatory SBP in each age group (Figure 3, #39 years: r 5 0.409, P , .0001; $40 and #59 years: r 5 0.395, P , .0001; $60 years: r 5 0.390, P , .0001). The DBP also demonstrated these trends (#39 years: r 5 0.297, P , .0001; $40 and #59 years: r 5 0.395, P , .0001, age $60 years: r 5 0.334, P , .0001). The VC of 24-h SBP
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
BLOOD PRESSURE VARIABILITY
1285
FIGURE 2. Blood pressure– dependent changes in 24-h blood pressure variability. A: SD, standard deviation. B: VC, variation coefficient (SD/average ambulatory blood pressure).
did not increase with increasing SBP (#39 years, r 5 0.029, P 5 .75; $40 and #59 years, r 5 20.015, P 5 .77; $60 years, r 5 0.0003, P 5 .996) and the VC of 24-h DBP decreased with increasing DBP (#39 years, r 5 20.209, P , .03; $40 and #59 years, r 5 20.156, P , .003; $60 years, r 5 20.132, P , .02). BP Dependent Changes in Other Variables Pulse pressure was significantly correlated with the 24-h SBP (r 5 0.875, P , .0001) and DBP (r 5 0.584, P , .0001). The magnitude of the nocturnal decline in BP (SBP: r 5 0.171, P , .0001; DBP: r 5 0.230, P , .0001) and the BMI (SBP: r 5 0.188, P , .0001; DBP, r 5 0.214, P , .0001) were positively correlated with the 24-h BP. The white-coat effect was negatively correlated with the 24-h BP (SBP: r 5 20.171, P , .0001; DBP, r 5 20.172, P , .0001). Daytime and nighttime BPs were also negatively correlated with the white-coat effect. Multivariate Stepwise Linear Regression Analysis Stepwise linear regression analysis showed that age, sex, BMI, pulse pressure, white-coat effect, and magnitude of the nocturnal fall in BP were significantly and independently related to the SD of 24-h SBP (Table 3). BP variability was significantly greater in women. Although the magnitude of the nocturnal decline in BP was a strong predictor of the SD of 24-h SBP, multivariate stepwise linear regression analysis
demonstrated that pulse pressure, age, sex, BMI, and white-coat effect were still independently correlated with the SD of the 24-h SBP. Similarly, multivariate regression analysis showed that the magnitude of the nocturnal decline in DBP, pulse pressure, white-coat effect, 24-h HR, age, BMI, and sex were significantly correlated with the SD of 24-h DBP (Table 4). Pulse pressure was a better predictor of the SDs of 24-h SBP and DBP than were age, sex, BMI, and white-coat effect. Pulse pressure was the strongest predictor of the SDs of daytime SBP and DBP (SD of daytime SBP: b-coefficient 5 0.246, SE 5 0.019, P , .0001, R2 5 0.27; SD of daytime DBP: b-coefficient 5 0.197, SE 5 0.015, P , .0001, R2 5 0.227). The following factors were independently correlated, in descending order of magnitude, with the SD of daytime SBP: pulse pressure, age, sex, and BMI. The following factors were independently correlated, in descending order of magnitude, with the SD of daytime DBP: pulse pressure, magnitude of nocturnal fall in BP, daytime HR, age, BMI, and daytime DBP. Age was the best predictor of the SD of nighttime SBP (b-coefficient 5 0.073, SE 5 0.01, P , .0001, R2 5 0.104). The SD of nighttime SBP was independently correlated, in descending order of magnitude, with age, nighttime average of SBP, magnitude of nocturnal decline in BP, nighttime HR, and BMI, although their predictive value was mini-
1286
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
IMAI ET AL
FIGURE 3. Blood pressure– dependent changes in 24-h blood pressure (SD, standard deviation) according to age. (A) Patients $39 years, (B) Patients $40 and #59 years, (C) Patients $60 years.
mal. The SD of nighttime DBP was independently correlated, in descending order of magnitude, with the nighttime DBP, magnitude of the nocturnal decline in BP, nighttime HR, age, BMI, and pulse pressure, although their predictive value was minimal. DISCUSSION In the present study, the SD and VC of BP measured every 30 min for 24-h, daytime, and nighttime periods
were used as indexes of short-term BP variability. Short-term BP variability, including sporadic and random variations as well as physiological variations, should be examined by beat-to-beat measurements of BP.15 DiRienzo et al15 reported that when the intervals between the intermittent measurements were 5, 10, or 15 min, the 24-h SD of BP obtained by intermittent measurements corresponded to the 24-h SD of BP obtained by continuous measurements. When the in-
TABLE 3. MULTIVARIATE STEPWISE LINEAR REGRESSION ANALYSIS OF THE SD OF 24-H SBP Variable Nocturnal fall in SBP (mm Hg) Pulse pressure (mm Hg) Age (years) Sex BMI (kg/m2) Casual SBP 2 24 h SBP (mm Hg)
Coefficient
SE
P
Partial R2
0.238 0.197 0.089 21.180 0.130 0.015
0.010 0.016 0.009 0.224 0.034 0.007
.0001 .0001 .0001 .0001 .0003 .0338
0.370 0.170 0.042 0.018 0.010 0.003 Model R2 0.612
BMI, body mass index; SE, standard error; SBP, systolic blood pressure. The independent variables entered in the model for multivariate regression analysis were cited in the text. The variables that showed P $ .05 in the multivariate relations were excluded from the table. Sex: men 5 1, women 5 0.
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
BLOOD PRESSURE VARIABILITY
1287
TABLE 4. MULTIVARIATE STEPWISE LINEAR REGRESSION ANALYSIS OF THE SD OF 24-h DBP Variables Nocturnal fall in DBP (mm Hg) Pulse pressure (mm Hg) Casual DBP 2 24 h DBP (mm Hg) 24 h HR (beats/min) Age (years) BMI (kg/m2) Sex
Coefficient
S.E.
P
Partial R2
0.218 0.127 0.032 0.043 0.035 0.084 20.353
0.011 0.011 0.007 0.009 0.006 0.022 0.142
.0001 .0001 .0001 .0001 .0001 .0001 .0130
0.326 0.153 0.022 0.012 0.016 0.012 0.004 Model R2 0.544
BMI: body mass index; S.E.: standard error; DBP: diastolic blood pressure; HR: heart rate. The independent variables entered in the model for multivariate regression analysis were cited in the text. Sex: men 5 1, women 5 0. The variables which showed P $ .05 in the multivariate relations were excluded from the table.
terval between intermittent measurements was 30 or 60 min, there was a marked difference in the 24-h SD between intermittent and continuous measurements in some subjects.15 However, indirect ambulatory BP monitoring using a device preset to measure BP every 30 min is widely used in clinical practice. Therefore, we examined the characteristics of BP variability based on every 30-min measurements. Bivariate Linear Regression Analysis It is generally accepted that BP variability is correlated with the BP level,16 –21 although Pickering et al did not confirm this relationship.22 In the present study, the SD of BP was significantly and positively correlated with the ambulatory BP. Mancia et al reported that the SD, but not the VC, of BP was greater in hypertensive subjects than in age-matched normotensive subjects.21 In the present study, we also found that the VC of SBP did not increase with increasing SBP, suggesting that an increase in BP variability is a function of the BP level. BP variability is also reportedly correlated with age.18 –24 However, the effect of age on BP variability does not appear to be independent of the concomitant increase in BP level. Brennan et al25 reported that there was no difference in BP variability between young and elderly subjects with the same BP level. However, the present results showed an age-dependent increase in the VC of SBP, an index of BP variability corrected by the ambulatory BP level. The increase in the SD of SBP associated with increasing ambulatory SBP was observed in all age groups, suggesting that the BP level and age independently affected BP variability. The increase in BP variability in hypertensive and elderly subjects may be partly explained by the diminished baroreflex sensitivity associated with increased arterial stiffness due to aging and hypertension,26 –31 although other factors (probably central in nature32) may also be involved.18,21,32 Conway et al33 found that subjects with elevated baroreflex sensitivity showed
greater HR variability. Although the effect of age on the SD of HR is controversial,10,18,25,34 in the present study, the SD of HR decreased with age, suggesting that aging was associated with diminished baroreflex function. Disturbed baroreflex function is related to exaggerated pressor responses to the mental and physical stimuli and causes orthostatic and postprandial hypotension,35–38 resulting in an increase in BP variability. Another possible explanation for the age-dependent increase in SBP variability is that any change in stroke volume in elderly individuals is likely to induce a greater change in BP because the arterial wall is less compliant in the elderly. Age- and BP-dependent increases in SBP variability were greater than the corresponding changes in DBP variability in the present study. A change in stroke volume, which is mediated mainly by a change in the activity of the sympathetic nervous system and is caused by changes in mental and physical activities, directly reflects SBP, but not DBP, variability. Zachariah et al previously found that there was a lesser degree of variability in DBP compared with SBP.39 The present findings are consistent with their results. Mental and physical activity are generally decreased in elderly subjects, resulting in a lesser degree of BP variability. However, age-dependent increases in BP variability were observed in the present study, not only for the 24-h BP and the daytime BP but also for the nighttime BP, indicating that BP variability is mediated by changes in mental and physical activity and by changes in intrinsic vasomotor activity. The increased variability in vasomotor activity in association with decreased vascular compliance in elderly subjects may result in a greater variability in BP in elderly individuals than in younger individuals.
1288
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
IMAI ET AL
Multivariate Stepwise Linear Regression Analysis To exclude the influence of circadian BP variation from 24-h BP variability and to confirm the independent contributions of age and BP to BP variability, we performed multivariate stepwise linear regression analysis. We also performed multivariate stepwise regression analysis using daytime and nighttime BP variability as dependent variables. The magnitude of the nocturnal decline in BP, which is an index of circadian BP variation but not of short-term BP variability, was the strongest predictor of the SD of 24-h BP, suggesting that the SD of 24-h BP is not an appropriate index of short-term BP variability. Previous studies examining the relationship between BP variability and prognosis in hypertensive subjects have usually used the SD of 24-h BP as an index of BP variability.6,8,9 Thus, the reported relationship between BP variability and prognosis may depend mainly on the circadian BP variation. Although the nocturnal decline in BP was the strongest predictor of the 24-h BP, pulse pressure was still a strong predictor of 24-h BP variability. The contributions of age, SBP, and DBP to BP variability were smaller than that of pulse pressure. Pulse pressure increases with increasing age, especially after the age of 60 years,10 because of the associated decrease in compliance of the large elastic arteries. Thus, an increase in pulse pressure reflects aging and other factors that contribute to stiffness. Decreased compliance of the large elastic artery may lead to disturbed baroreceptor reflex function, which in turn results in increased variability in SBP in response to changes in the stroke volume in elderly subjects. It is interesting that the SD of HR, an index of baroreflex function, was not identified as an independent variable for SBP variability when pulse pressure was included in the multivariate analysis. These findings suggest that decreased compliance of the large elastic artery was a major independent contributor to BP variability. Age was the strongest predictor of the SD of nighttime SBP, whereas nighttime DBP was the strongest predictor of the SD of nighttime DBP. Daytime SBP variability may have resulted from changes in stroke volume due to a change in sympathetic nerve activity and to decreased compliance of the large elastic artery. However, changes in stroke volume are minimal during sleep; therefore, age and the BP level per se may have a greater effect on BP variability at night. It is interesting that BMI was an independent variable for 24-h, daytime, and nighttime BP variability and for SBP and DBP. Cardiac autonomic dysfunction, especially vagal dysfunction, has been observed in obese subjects,36,37 which may explain the increased variability in BP in obese subjects in the present study. The contribution of the white-coat effect for the SD of
the 24-h SBP was minimal in the model, although the association was statistically significant (Table 3). Study Limitations There are a number of possible limitations to the present study. Our data may not be applicable to other geographic or racial groups. We excluded treated hypertensive subjects from the cohort, which could potentially have led to underestimation of BP variability. In addition, we did not assess the effects of sleep disturbances on nocturnal BP variability. Although this problem could be addressed in a future study by monitoring the subjects’ sleep, we believe that the large number of subjects in the present study should have eliminated the influence of sleep disturbance. It must be emphasized that multiple logistic regression analysis can only suggest associations between dependent and independent variables and cannot prove a cause-and-effect relationship. Finally, we would like to emphasize again that the BP variability in the present study represents 30-min BP variability and not beat-to-beat BP variability. In summary, multivariate stepwise linear regression analysis showed that pulse pressure, an index of aging and arterial stiffness, was a strong predictor of BP variability in the overall study population. Because the nocturnal fall in BP was the major determinant of 24-h ambulatory BP, the SD of 24-h BP should not be used as an index of short-term BP variability. Future studies are needed to address the relationship between BP variability and cardiovascular events. REFERENCES 1.
Sokolow M, Werdegar D, Kain HK, Hinman AT: Relationship between level of blood pressure measured casually and by portable recorders and severity of complications in essential hypertension. Circulation 1966; 34:279 –307.
2.
Julius S, Ellis CN, Pascual AV, et al: Home blood pressure determination. Value in borderline (‘‘labile’’) hypertension. JAMA 1974;229:663– 666.
3.
Perloff D, Sokolow M, Cowan R: The prognostic value of ambulatory blood pressures. JAMA 1983;249:2792– 2798.
4.
Conway J: Home blood pressure recording. Clin Exp Hypertens 1986;A8:1247–1294.
5.
Devereux RB, Pickering TG, Harshfield GA, et al: Left ventricular hypertrophy in patients with hypertension: importance of blood pressure response to regularly recurring stress. Circulation 1983;68:470 – 476.
6.
Parati G, Pomidossi G, Albini F, et al: Relationship of 24-hour blood pressure mean and variability to severity of target-organ damage in hypertension. J Hypertens 1987;5:93–98.
7.
White WB, Schulman P, McCabe EJ, Dey HM: Average daily blood pressure, not office blood pressure, determines cardiac function in patients with hypertension. JAMA 1989;261:873– 877.
8.
Frattola A, Parati G, Cuspidi C, et al: Prognostic value
AJH–NOVEMBER 1997–VOL. 10, NO. 11, PART 1
of 24-hour blood pressure variability. J Hypertens 1993; 11:1133–1137.
BLOOD PRESSURE VARIABILITY
1289
25.
Pickering TG, James GD: Ambulatory blood pressure and prognosis. J Hypertens 1994;12(suppl 8):S29 –S33.
Brennan M, O’Brien E, O’Malley K: The effect of age on blood pressure and heart rate variability in hypertension. J Hypertens 1986;4(suppl 6):S269 –S272.
26.
10.
Imai Y, Nagai K, Sakuma M, et al: Ambulatory blood pressure of adults in Ohasama, Japan. Hypertension 1993;22:900 –912.
McGarry K, Laher M, Fitzgerald D, et al: Baroreflex function in elderly hypertensives. Hypertension 1983; 5:763–766.
27.
11.
Imai Y, Abe K, Sasaki S, et al: Determination of clinical accuracy and nocturnal blood pressure pattern by new portable device for monitoring indirect ambulatory blood pressure. Am J Hypertens 1990;3:293–301.
Gribbin B, Pickering TG, Sleight P, Peto R: Effect of age and high blood pressure on baroreflex sensitivity in man. Circulation 1971;29:424 – 431.
28.
Shimada K, Kitazumi T, Ogura H, et al: Differences in age-independent effects of blood pressure on baroreflex sensitivity between normal and hypertensive subjects. Clin Sci 1986;70:489 – 494.
29.
Floras JS, Hassan MO, Jones JV, et al: Consequences of impaired baroreflexes in essential hypertension: effects on pressor responses, plasma noradrenaline and blood pressure variability. J Hypertens 1988;6:525–535.
30.
Avolio AP, Chen SG, Wang RP, et al: Effects of aging on changing arterial compliance and left ventricular load in a northern Chinese urban community. Circulation 1983;68:50 –58.
31.
O’Rourke M: Arterial stiffness, systolic blood pressure, and logical treatment of arterial hypertension. Hypertension 1990;15:339 –347.
32.
Mancia G, Parati G, Pomidossi G, et al: Arterial baroreflexes and blood pressure and heart rate variabilities in humans. Hypertension 1986;8:147–153.
9.
12.
13.
Imai Y, Abe K, Sekino H: Evaluation of automatic devices for blood pressure measurement, in Koshikawa S, Nagasama T (eds): Annual Review Nephrology 1992. Chugai-Igakusha, Tokyo, 1991, pp 55– 61 (in Japanese). Imai Y, Nihei M, Abe K, et al: A finger volume-oscillometric device for monitoring ambulatory blood pressure: laboratory and clinical evaluation. Clin Exp Hypertens 1987;A9:2001–2025.
14.
SAS Institute Inc: SAS/STAT User’s Guide, Release 6.03 Edition. SAS Institute Inc., Cary, NC, 1988.
15.
DiRienzo M, Grassi G, Pedotti A, Mancia G: Continuous vs intermittent blood pressure measurements in estimating 24-hour average blood pressure. J Hypertens 1983;5:264 –269.
16.
Clement DL: Blood pressure variability in hospitalized patients. Acta Clin Belg 1977;32:163–167.
17.
Goldberg AD, Raftery EB, Cashman MM, Stott FD: Study of untreated hypertensive subjects by means of continuous intra-arterial blood pressure recordings. Br Heart J 1978;40:656 – 664.
33.
Conway J, Boon N, Vann Jones J, Sleight P: Mechanisms concerned with blood pressure variability throughout the day. Clin Exp Hypertens 1985;A7:153– 157.
18.
Mancia G, Ferrari A, Gregorini L: Blood pressure variability in man: its relation to high blood pressure, age and baroreflex sensitivity. Clin Sci 1980;59(suppl 1): 401S– 404S.
34.
Kawano A, Shimada K, Matsubayashi K, et al: Cardiovascular regulatory function in elderly patients with hypertension. Hypertension 1989;13:401– 407.
19.
Watson RDS, Stallard TJ, Flinn RM, Littler WA: Factors determining direct arterial pressure and its variability in hypertensive man. Hypertension 1980;2:333–341.
35.
Tsuchihashi T, Abe I, Tsukashima A, et al: Effects of meal and physical activity on blood pressure variability in elderly patients: a preliminary study. Am J Hypertens 1990;3:943–946.
20.
Floras JS, Sleight P: The lability of blood pressure, in Amery A, Fagard R, Lijnen P, Staessen J (eds): Hypertensive Cardiovascular Disease: Pathophysiology and Treatment. Martinus Nijhof, The Hague, 1982, pp 104 – 117.
36.
Peitzman SJ, Berger SR: Postprandial blood pressure decrease in well elderly persons. Arch Intern Med 1989; 149:286 –288.
37.
Lipsitz LA, Fullerton KJ: Postprandial blood pressure reduction in healthy elderly. J Am Geriatr Soc 1986;34: 267–270.
38.
Caird FI, Andrews GR, Kennedy RD: Effects of posture on blood pressure in the elderly. Br Heart J 1973;35: 527–530.
39.
Zachariah PK, Sheps SG, Bailey KR, et al: Age-related characteristics of ambulatory blood pressure load and mean blood pressure in normotensive subjects. JAMA 1991;265:1414 –1417.
40.
Peterson HR, Rothschild M, Weinberg CR, et al: Body fat and the activity of the autonomic nervous system. N Eng J Med 1988;318:1077–1083.
41.
Rossi M, Marti G, Ricordi L, et al: Cardiac autonomic dysfunction in obese subjects. Clin Sci 1989;76:567–572.
21.
22.
23.
24.
Mancia G, Ferrari A, Gregorini L, et al: Blood pressure and heart rate variabilities in normotensive and hypertensive human beings. Circ Res 1983;53:96 –104. Pickering TG, Harshfield GA, Kleinert HD, et al: Blood pressure during normal daily activities, sleep, and exercise. Comparison of values in normal and hypertensive subjects. JAMA 1982;247:992–996. Drayer JIM, Weber MA, DeYoung JL, Wyle FA: Circadian blood pressure patterns in ambulatory hypertensive patients. Am J Med 1982;73:493– 499. Canonico V, DeCaprio L, Vigorito C, et al: Differences in blood pressure profile between young and elderly hypertensive patients. J Hum Hypertens 1990;4:405– 409.