- Email: [email protected]
A 5-year follow-up of ambulatory blood pressure in healthy older adults
AJH
2003; 16:640 – 645
A 5-Year Follow-Up of Ambulatory Blood Pressure in Healthy Older Adults Iris B. Goldstein, David Shapiro, and Donald Guthrie Background: This study assessed 5-year changes in ambulatory blood pressure (ABP) in healthy, older individuals and determined the extent to which it could be predicted from earlier BP measures and other cardiovascular risk factors. Methods: A total of 162 men and women, aged 55 to 79 years, with no prior medical disorders, completed a medical examination and two 24-h ABP sessions. The procedures were repeated 5 years later in 80% (130) of these subjects. A modified hierarchical regression analysis was used to determine whether initial ABP and casual blood pressure (CBP) measures and demographic and physical examination data could predict ABP in 5 years. Results: The CBP and most ABP levels during waking and sleep increased after 5 years. However, CBP remained in the normotensive range for 73% of the subjects. The
T
ABP variability tended to decrease over time. The ABP and CBP measures accounted for at least 50% of the variance in the prediction of ABP level after 5 years. In comparison, the predictability of ABP variability was quite low, particularly during sleep (⬍30% of the variance accounted for). Conclusions: The ABP and CBP were good predictors of future ABP level in healthy older subjects, but ABP variability was more difficult to predict. Except for age, none of the standard cardiovascular risk factors contributed significantly to the prediction of ABP level or variability. Am J Hypertens 2003;16:640 – 645 © 2003 American Journal of Hypertension, Ltd. Key Words: Ambulatory blood pressure, blood pressure variability, older subjects.
he superiority of 24-h ambulatory blood pressure (ABP) monitoring over casual blood pressure (CBP) has been repeatedly demonstrated. Not only is it less likely to give biased values, but ABP appears to be a more sensitive measure of peripheral vessel end organ damage and a better predictor of cardiovascular morbidity and mortality.1 Studies comparing ABP measures over the short term (days or months),2 and up to 2 years,3 concluded that ABP mean levels exhibited reproducibility over time that was not only adequate, but generally exceeded that of measurements of office blood pressure (BP). Reproducibility of ABP variability, on the other hand, was poor.2,4 However, little is known about what happens to ABP level and variability over several years. Longitudinal studies of CBP suggest that over intervals of 5 to 10 years BP gradually increases with age, with the likelihood of a plateau or even a decrease in diastolic BP (DBP) in the fifth or sixth decade.5,6 The strongest predictor of CBP among middle-age men was initial CBP measurement.7,8 In addition, the 24-h mean of intra-arterial ABP best
predicted noninvasive ABP after 10 years in middle-aged men with varying BP levels.9 To our knowledge, no information is available on long-term progression of ABP in healthy, older men and women. The primary goal of this study was to compare initial ABP with assessments repeated after 5 years. We wanted to follow the progression of ABP over time in older men and women who had never had major chronic illnesses, nor taken medication that affected the cardiovascular or central nervous system. Moreover, we wanted to determine whether ABP could be predicted from initial ABP and CBP assessments, and demographic and physical examination data.
Received December 9, 2002. First decision March 4, 2003. Accepted March 25, 2003. From the Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles, California. This study was supported by Research Grant AG-11595 from the
National Institute of Aging.
0895-7061/03/$30.00 doi:10.1016/S0895-7061(03)00906-3
Methods Phase 1 Subjects The initial sample (phase 1) included 162 subjects (94 women, 68 men), aged 55 to 80 years, living in the community. Average education level was 15.5 ⫾ 5.8 (mean ⫾ SD) years. Exclusions involved subjects’ reports
Address correspondence and reprint requests to Dr. Iris B. Goldstein, UCLA Department of Psychiatry, 760 Westwood Plaza, Los Angeles, CA 90095-1759; e-mail: [email protected] © 2003 by the American Journal of Hypertension, Ltd. Published by Elsevier Inc.
AJH–August 2003–VOL. 16, NO. 8
AMBULATORY BP IN OLDER ADULTS
641
Table 1. Characteristics of men (n ⫽ 47) and women (n ⫽ 69) during phase 1 and phase 2 Men Characteristics 2
BMI, kg/m Exercise, hours Alcohol, drinks/week Coffee, cups/day Glucose, mg/dL Total chol, mg/dL LDL chol, mg/dL HDL chol, mg/dL Triglycerides, mg/dL BP level, mm Hg SBP casual DBP casual SBP wake DBP wake SBP sleep DBP sleep BP variability, mm Hg SBP wake DBP wake SBP sleep DBP sleep
Phase 1
Women Phase 2
Phase 1
Phase 2
Significance
25.5 8.9 3.8 2.2 109.3 223.0 142.6 48.8 161.0
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
2.7 6.1 4.8 2.2 27.0 33.6 28.4 12.1 99.2
25.6 9.9 5.8 2.3 131.8 201.6 128.0 48.0 126.0
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
2.8 8.1 5.3 2.0 33.7 32.8 28.0 10.2 69.0
23.8 11.1 2.4 1.7 116.2 233.6 133.3 70.6 139.0
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
2.8 8.9 3.5 2.0 29.5 36.5 38.1 20.0 68.6
24.0 11.5 3.9 2.0 142.9 213.4 121.6 67.9 119.4
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
3.3 9.2 4.2 1.1 45.5 25.3 24.7 14.3 60.6
122.5 75.0 127.1 74.5 110.8 63.6
⫾ ⫾ ⫾ ⫾ ⫾ ⫾
13.4 9.7 10.7 6.8 11.3 7.5
129.2 76.4 133.2 74.8 116.0 64.4
⫾ ⫾ ⫾ ⫾ ⫾ ⫾
14.3 9.7 11.8 7.6 12.4 7.6
114.4 69.0 123.6 72.6 106.1 59.7
⫾ ⫾ ⫾ ⫾ ⫾ ⫾
12.4 7.7 11.0 6.6 11.3 6.2
123.8 72.1 129.4 72.0 113.7 61.5
⫾ ⫾ ⫾ ⫾ ⫾ ⫾
15.6 8.9 12.9 7.4 13.0 7.2
G†, P‡ G†, P‡ P‡
13.4 9.6 9.6 7.5
⫾ ⫾ ⫾ ⫾
2.2 2.1 3.0 2.5
12.4 8.1 9.6 6.9
⫾ ⫾ ⫾ ⫾
2.8 2.2 4.3 2.6
14.2 11.7 9.7 7.5
⫾ ⫾ ⫾ ⫾
3.3 3.3 3.1 2.7
12.8 9.0 9.3 6.7
⫾ ⫾ ⫾ ⫾
3.3 2.3 3.7 2.6
P‡ G‡, P‡, G⫻P*
G† P‡ P‡ G*, P‡ P‡ G‡ P‡
P‡ G†, P*
P*
Values are derived from ANCOVA with gender as a group variable and phase as a repeated measure. BMI ⫽ body mass index; G ⫽ Gender; P ⫽ Phase; chol ⫽ cholesterol; BP ⫽ blood pressure; SBP ⫽ systolic BP; DBP ⫽ diastolic BP. * P ⬍ .05; † P ⬍ .01; ‡ P ⬍ .001. Table does not include 14 subjects taking medications that influence BP.
of any serious current or prior illness or medications influencing either the cardiovascular or central nervous system, history of hypertension, drug or alcohol abuse, head injuries, or obesity (body mass index [BMI] ⬎30 kg/m2). Information was confirmed by medical examination (see Procedures) and prior medical records. Abnormal findings provided further basis for exclusion.10 Psychological testing and a mental status examination excluded subjects with cognitive abnormalities or psychiatric disorders.10 Although not previously diagnosed with hypertension, 14 subjects had CBP within hypertensive levels.11
Procedures
Phase 2 Subjects
Ambulatory Monitoring
Of the 162 subjects who participated in phase 1, 130 (80%) completed phase 2 after 4.9 ⫾ 0.63 years (range, 4 to 6 years). Three subjects could not be reached, 3 moved out of the area, 2 died (1 suicide, 1 lung cancer), 1 had a stroke, 6 had incomplete ABP data, and 17 declined for personal reasons. Data presented here are based on 75 women and 55 men who participated in both phases of the study (Table 1 shows sample characteristics). Ethnic composition was as follows: 99 white, 20 Asians, 9 African Americans, 1 Latino, and 1 Native American. When comparisons were made between subjects who completed the follow-up phase and those who did not, there were no significant differences (t test or 2) for BP, demographic, or physical examination data. Subjects gave informed consent, approved by the UCLA Institutional Review Board.
A laboratory assistant obtained three standard CBP readings11 after 5 min of sitting and just before attachment of the ABP monitor. Three additional CBP readings were taken 1 week later before the second ABP session. The ABP was recorded by the Accutracker II (Suntech Medical Instruments, Raleigh, NC), which has been used widely in clinical and research studies with established reliability and validity.12 The ambulatory recorder was programmed to operate three times an hour on a variable schedule during waking hours and once an hour during sleep (based on subjects’ estimates of times of sleep and awakening). An activity monitor (Mini-Motionlogger, Ambulatory Monitoring Inc., Ardsley, NY) was used to confirm the differentiation of sleep from awake readings and to account for effects of activity on BP.13 Ambulatory data were edited for artifacts based on
Identical procedures were performed during phases 1 and 2. Subjects were given physical and mental status examinations by the physician, including a complete health history, 12-lead electrocardiogram, urinalysis, and chemical panel. Laboratory tests were done by standard techniques (SmithKline Beecham Clinical Laboratories, Van Nuys, CA). Changes in health during the 5-year period were noted; medical records were obtained from subjects’ physicians.
642
AMBULATORY BP IN OLDER ADULTS
Accutracker error codes. Editing was done by set rules.14 Classification of each reading as wake or sleep was based on diary entries and postsession reports. Only nighttime sleep values were included in the sleep category and daytime nonsleep values in the wake category. Means were obtained for six CBPs and for two sessions of ambulatory systolic BP (SBP) and DBP (wake level, sleep level, wake variability, sleep variability). For a given subject, variability was based on the SD of the wake and of the sleep period for a given day. The coefficient of variation was also used, but it did not alter results. Only findings with the SD are included. Data Analyses The BP, demographic, and blood chemistry data were looked at separately for men and women during each phase of the study in a group (gender ⫽ men/women) ⫻ repeated measures (phase ⫽ 1/2) ANOVA. Table 1 is based only on 116 subjects not taking medications with antihypertensive effects. Findings for the entire sample (n ⫽ 130) were the same as for the smaller sample (n ⫽ 116). Separate analyses indicated that activity had no appreciable effect on ABP values.13 A modified hierarchical regression analysis was constructed to determine whether ABP after 5 years could be predicted by its initial value or by other variables. Data for men and women were combined. For each measure, four sets of predictor variables were specified: initial value of the measure, initial values of other BP measures (ambulatory wake and sleep level, ambulatory wake and sleep variability, and CBP), demographic variables (age, gender, education, BMI, exercise, alcoholic drinks, caffeine), and blood chemistry measures (total cholesterol, HDL and LDL cholesterol, triglycerides, 2-h glucose after 75-g glucose load). (Substituting 5-year change for initial values of demographic and blood chemistry measures did not alter findings in any models.) Based on prior findings indicating that measures of CBP and ABP level were highly intercorrelated but were not related to variability,15 we hypothesized that measures of BP level would predict future BP level and variability measures would be predictive of future ABP variability. Therefore, variability measures were not used to predict BP level measures, nor were level measures used to predict variability. The first step of each regression forced in the initial value of the target variable. Second, we considered the other BP measures and entered them individually according to significance, based on the P value (⬍.05) within the step after preceding variables were included. Following the same procedure, the final step involved entering the remaining demographic and blood chemistry variables. This process avoids loss of predictive information due to colinearity within the steps, while still identifying variables that predict the target variable.
AJH–August 2003–VOL. 16, NO. 8
Results Health and BP Status of Subjects During the 5-year period between the two phases, 11 people were diagnosed and treated for hypertension, 3 were taking other medications with antihypertensive effects (1 heart disease, 2 benign prostatic hypertrophy), 1 had a stroke, and 3 had transient ischemic attacks. There were only 4 smokers during both phases of the study. The gender ⫻ phase ANOVA (Table 1) indicated that from phase 1 to phase 2 the following non-BP variables did not change: BMI, hours of exercise, coffee intake, HDL cholesterol. However, alcohol intake and glucose level increased and total cholesterol, LDL cholesterol, and triglycerides decreased. Because statin use increased over time, we compared lipids during phase 2 in individuals receiving statins with those who were not. We found that medication use was associated with a greater decrease in LDL and total cholesterol at follow-up (t tests, P ⬍ .05). Triglycerides and HDL cholesterol were unaffected. Also, independent of phase, men had higher BMI and total cholesterol and lower HDL cholesterol than women. Higher HDL cholesterol in women was not the result of hormone replacement therapy (HRT), because once we corrected for age, ANOVA indicated no differences in lipids between 32 women on and 43 women off HRT. ANOVA for BP showed phase effects with increases from phase 1 to phase 2 for casual SBP and DBP, SBP wake, and SBP and DBP sleep (Table 1). The SBP and DBP wake variability, and DBP sleep variability decreased over time. With regard to gender effects, compared to women, men had higher casual SBP and DBP and DBP sleep and lower DBP wake variability. There was a gender ⫻ phase interaction for DBP wake variability, with women showing a decrease of 2.7 mm Hg from phase 1 to phase 2 compared to a 1.5 mm Hg decrease in men. Shifts in BP Over Time Subjects who completed both phases of the study were classified in terms of standard definitions11 of CBP levels during each phase as follows: normal, SBP ⬍130 mm Hg and DBP ⬍85 mm Hg; high normal, SBP ⫽ 130 to 139 mm Hg or DBP ⫽ 85 to 89 mm Hg; hypertension, SBP ⬎139 mm Hg or DBP ⬎89 mm Hg or currently taking antihypertensive medication (11 subjects). (Three subjects who were taking BP-lowering drugs for nonhypertensive problems were excluded, leaving a total of 127 subjects.) A 2 analysis of change in BP classification from phase 1 to phase 2 indicated no differences between men and women. Almost 75% of subjects (n ⫽ 71) maintained normal BP status from phase 1 to phase 2. Of those who were high normal, 62% became hypertensive. Almost all subjects (13 of 14) who were hypertensive during phase 1 remained hypertensive (Table 2), although only 4 of them were later treated with drugs for hypertension.
AJH–August 2003–VOL. 16, NO. 8
643
AMBULATORY BP IN OLDER ADULTS
Table 2. Subjects’ casual blood pressure (CBP) during phase 2 (P2), relative to phase 1 (P1) blood pressure Normal P2 Normal P1 High normal P1 Hypertensive P1 Total N
High Normal P2
Hypertensive P2
n
%
n
%
n
%
Total n
71 1 1 73
73.2 6.3 7.1
16 5 0 21
16.5 31.2 0.0
10 10 13 33
10.3 62.5 92.9
97 16 14 127
Percent (%) values are based on row totals. Table does not include three subjects who were taking blood pressure-lowering drugs but were not diagnosed with hypertension.
Prediction of ABP Correlations between phase 1 and phase 2 measures of BP level (CBP and ABP) were moderately high (0.62 to 0.71) with lower correlations for ABP variability (0.28 to 0.48) (Table 3). With the modified hierarchical regression, phase 1 ABP accounted for at least 50% of the variance in ABP (SBP and DBP wake and sleep) during phase 2 (Table 4). In the models for SBP and DBP wake and SBP sleep approximately 2% to 5% more was added by CBP. In general, variability was much less predictable from phase 1 measures, particularly during sleep. Age accounted for an additional 4% to 5% of the variance for DBP wake and SBP and DBP wake variability. No other demographic or blood chemistry variables added significantly to the variance.
Discussion Health and BP Status of Subjects During the 5-year interval between phases there were no significant changes in BMI, exercise, or caffeine intake. Subjects drank more alcohol and had lower triglycerides, total cholesterol, and LDL cholesterol. Improvement in LDL and total cholesterol appears to have been associated with increased use of statins. Higher HDL cholesterol
Table 3. Correlations between phase 1 and phase 2 blood pressure Blood Pressure
Correlation Coefficient
SBP casual DBP casual SBP wake DBP wake SBP sleep DBP sleep SBP wake variability DBP wake variability SBP sleep variability DBP sleep variability Abbreviations as in Table 1. Values are in mm Hg. n ⫽ 116.
0.71 0.69 0.70 0.64 0.64 0.62 0.48 0.47 0.32 0.28
P .0001 .0001 .0001 .0001 .0001 .0001 .0001 .0001 .0005 .003
(reflected in higher total cholesterol) in women than in men, could not be explained by HRT, which is sometimes associated with elevations in HDL.16 The decline in total and LDL cholesterol levels with age corresponds to the literature. Ferrara et al17 found the largest total cholesterol drop after 8 years among the oldest men and women (65 to 79 years at baseline) in their study. However, findings regarding HDL cholesterol levels were more equivocal.17,18 Improvements in lipid levels in our subjects could have been influenced by lifestyle changes. Most subjects claimed that their participation was due to a general interest in health, reflected in exercise rate (approximately 10 h/w) and in the small number of smokers. Mean CBP levels for men and women combined increased approximately 8/2 mm Hg from phase 1 to phase 2. Larger increases in SBP than DBP among older individuals are consistent with prior findings.19 Although both SBP and DBP increase with age, DBP plateaus or even declines in later years.5,6 Compared to women, men in the current study had consistently higher CBP. In contrast, Cornoni-Huntley et al.20 reported that after age 55, prevalence rates for hypertension in women exceeded men. However, Pearson et al19 suggested that gender cross-over in BP was less likely to occur in healthy, carefully screened individuals and showed that both longitudinally and cross-sectionally men’s BP levels exceeded those of women at all ages, with a narrowing of gender differences at age 70. The ABP levels also increased over time, with only DBP wake showing no changes. Moreover, during both phases, DBP sleep was the only ambulatory level measure that was higher in men than in women. The SBP and DBP wake variability and DBP sleep variability decreased from phase 1 to 2, an unexpected result in view of previous findings of greater variability in older than in younger adults.21,22 These differences may be due to discrepancies between cross-sectional and prospective studies. The current results also showed that, compared to men, women exhibited greater DBP wake variability, which also dropped more from phase 1 to phase 2. A gender difference in ABP variability was found previously in older but not in younger adults.22
644
AMBULATORY BP IN OLDER ADULTS
AJH–August 2003–VOL. 16, NO. 8
Table 4. Modified hierarchical regression to predict phase 2 ambulatory blood pressure (ABP) Follow-up BP Variables SBP wake DBP wake
SBP sleep DBP sleep
SBP wake variability DBP wake variability SBP sleep variability DBP sleep variability
Predictors
Regression Coefficient
P
Cumulative % R2
SBP wake SBP casual DBP wake DBP sleep DBP casual Age SBP sleep SBP wake DBP sleep DBP wake SBP wake DBP casual SBP wake variability Age DBP wake variability Age SBP sleep variability SBP wake variability DBP sleep variability SBP wake variability
0.545 0.241 0.295 0.251 0.303 ⫺0.214 0.464 0.296 0.482 0.296 ⫺0.189 0.212 0.419 0.236 0.435 0.198 0.214 0.282 0.184 0.272
.0001 .015 .003 .002 .001 .001 .0001 .004 .0001 .019 .054 .023 .0001 .005 .0001 .018 .022 .003 .051 .004
53.1 55.5 47.3 50.6 55.9 60.3 47.6 51.3 45.2 51.0 52.8 55.0 23.4 28.6 22.1 25.9 10.1 16.9 8.0 14.4
Other abbreviations as in Tables 1 and 3. ABP was predicted from phase 1 ambulatory and casual blood pressure and from demographic (age, gender, education, body mass index, exercise, alcohol, caffeine), and blood chemistry variables (total cholesterol, HDL, LDL, triglycerides, glucose). n ⫽ 116.
In contrast to our findings, CBP has generally been found to be higher than waking ABP,23 probably as a consequence of the white coat effect and the limited number of measurements taken in stressful clinic settings.24 However, increasing the number of office measurements can lead to lower CBP.25 Lower CBP compared with ABP in our subjects could have been due to CBP being recorded during the second and third visits (not the first) in a stress-free, nonmedical setting by research assistants. Furthermore, there is evidence that at least in older, white men, white coat hypertension is rare.25 Shifts in BP Over Time Individuals who start life with low BP are more likely to exhibit smaller BP increases with age.26 The fact that the CBP of 73% of our subjects remained in the normotensive range after 5 years may have been due to their initial good health and relatively low BP. Of the 16 subjects with high normal BP during phase 1, 62% became hypertensive during phase 2. Although casual BP differed between subjects who became hypertensive and those who did not (t test, P ⫽ .007), neither ABP level nor variability differed significantly between the two groups. Our results are consistent with findings that older individuals with high normal CBP are most likely to develop hypertension, even after just 4 years.27 Prediction of ABP Just as current CBP predicted CBP 5 years later,7 wake and sleep ABP predicted themselves fairly well after 5
years. These results are in accord with a 10-year follow-up study, using 24-h BP mean in place of waking BP.9 We also found that multiple measures of ABP level during wake and sleep as well as CBP added significantly to the total variance predicted for ABP wake and sleep. The relationship between different measures of BP level is in keeping with their high intercorrelations and their occurrence on a single BP factor.15 Age was the only non-BP variable contributing significantly to the total predicted variance (4.4% for DBP wake). Although at least 50% of the variance was explained, the remaining variance may be due to other factors that are more difficult to assess, such as changes in experience and lifestyle. In comparison to ABP level, predictability of variability was low. Correlations between phase 1 and phase 2 BP indicated higher associations between levels than between variability measures, with decreasing values during sleep. The ABP variability has been found to show poor reproducibility,4 which Ward et al2 suggested may be due to frequent behavior changes and the responsiveness of BP to “external demands and internal homeostatic requirements.” The prediction of ABP was not influenced by any of the following standard risk factors: gender, education, BMI, hours of exercise, alcohol intake, caffeine intake, lipid level, or glucose level. Only age added to the variance for some ABP measures. A power analysis of our data showed that, with 116 subjects, if any of these other variables had been involved in predicting BP, they would have showed up in the analyses. Moreover, a 10-year longitudinal study
AJH–August 2003–VOL. 16, NO. 8
indicated that, other than initial BP, baseline characteristics did not significantly predict either future CBP or 24-h BP in middle-aged men.9 Summary and Conclusions Increases in subjects’ systolic ABP level over time are consistent with our CBP findings and those of other investigations of older subjects.5,6,19 In addition, individuals who started out with lower ABP were more likely to maintain low levels after 5 years. Our results also showed that women did not differ very much from men in the progression of BP to higher levels. With regard to the prediction of ABP, prior BP measures accounted for at least 50% of the variance. Variability was more difficult to predict. Finally, except for age, no other standard cardiovascular risk factor contributed to the prediction of ABP. Because of the strong association of ABP level and variability with end organ damage and cardiovascular morbidity and mortality,1,21 it is vital that future investigations focus on long-term studies of ABP, particularly in healthy, older populations where limited information is available.
References 1.
Meredith PA, Perloff D, Mancia G, Pickering T: Blood pressure variability and its implications for antihypertensive therapy. Blood Press 1995;4:5–11. 2. Ward MM, Turner JR, Johnston DW: Temporal stability of ambulatory cardiovascular monitoring. Ann Behavioral Med 1994;16:3– 11. 3. Gosse P, Bougaleb M, Clementy J: Long-term reproducibility of ambulatory blood pressure monitoring. The´rapie 1996;51:5–9. 4. Palatini P, Mormino P, Canali C, Santonastaso M, De Venuto G, Zanata G, Pessina A: Factors affecting ambulatory blood pressure reproducibility. Results of the HARVEST trial. Hypertension 1994; 23:211–216. 5. Kannel WB: Blood pressure as a cardiovascular risk factor. Prevention and treatment. JAMA 1996;275:1571–15776. 6. Landahl S, Bengtsson C, Sigurdsson JA, Svanborg A, Sva¨rdsudd K: Age-related changes in blood pressure. Hypertension 1986;8:1044 – 1049. 7. Carroll D, Smith GD, Sheffield D, Shipley MJ, Marmot MG: Pressor reactions to psychological stress and prediction of future blood pressure: data from the Whitehall II study. Br Med J 1995;310:771– 775. 8. Carroll D, Smith GD, Shipley MJ, Steptoe A, Brunner EJ, Marmot MG: Blood pressure reactions to acute psychological stress and future blood pressure status: a 10-year follow up of men in the Whitehall II Study. Psychosom Med 2001;63:737–743. 9. Jokiniitty JM, Majahalme SK, Ka¨ho¨nen MAP, Tuomisto MT, Turjanmaa MH: Prediction of blood pressure level and need for antihypertensive medication: 10 years of follow-up. J Hypertens 2001; 19:1193–1201. 10. Goldstein IB, Bartzokis G, Hance D, Shapiro D: Relationship between blood pressure and subcortical lesions in healthy elderly people. Stroke 1998;29:765–772.
AMBULATORY BP IN OLDER ADULTS
645
11. Joint National Committee on Prevention: The Sixth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Arch Intern Med 1997;157: 2413–2446. 12. Jyothinagaram SG, Watson D, Padfield P: Suntech Accutracker ambulatory blood pressure monitor— clinical validation. J Ambulatory Monitor 1990;3:63–67. 13. Shapiro D, Goldstein IB: Wrist actigraph measures of physical activity level and ambulatory blood pressure in healthy elderly persons. Psychophysiology 1998;35:305–312. 14. Goldstein IB, Jamner LD, Shapiro D: Ambulatory blood pressure and heart rate in healthy male paramedics during a workday and a nonworkday. Health Psychol 1992;11:48 –54. 15. Goldstein IB, Shapiro D, La Rue A, Guthrie D: Relationship between 24-hour ambulatory blood pressure and cognitive function in healthy elderly people. Aging, Neuropsychol & Cognition 1998;5: 215–224. 16. Rijpkema AHM, van der Sanden AA, Ruijs AHC: Effects of postmenopausal oestrogen–progestogen replacement therapy on serum lipids and lipoproteins: a review. Maturitas 1990;12:259 –285. 17. Ferrara A, Barrett-Connor E, Shan J: Total, LDL, and HDL cholesterol decrease with age in older men and women. The Rancho Bernardo Study 1984 –1994. Circulation 1997;96:37–43. 18. Abbott RD, Sharp DS, Burchfiel CM, Curb JD, Rodriguez BL, Hakim AM, Yano K: Cross-sectional and longitudinal changes in total and high-density-lipoprotein cholesterol levels over a 20-year period in elderly men: The Honolulu Heart Program. Ann Epidemiol 1997;7:417–424. 19. Pearson JD, Morrell CH, Brant LJ, Landis PK, Fleg JL: Ageassociated changes in blood pressure in a longitudinal study of healthy men and women. J Gerontol Med Sci 1997;52A:M177– M183. 20. Cornoni-Huntley J, LaCroix AZ, Havlik RJ: Race and sex differentials in the impact of hypertension in the United States. The National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study. Arch Intern Med 1989;149:780 –788. 21. Mancia G, Di Rienzo M, Grassi G, Parati G: Blood pressure variability and reflex control in the elderly. Aging 1995;7:3–9. 22. Jaquet F, Goldstein IB, Shapiro D: Effects of age and gender on ambulatory blood pressure and heart rate. J Hum Hypertens 1998; 12:253–257. 23. Mancia G, O’Brien E, Imai Y, Redon J: Task Force II: ambulatory blood pressure monitoring in population studies. Blood Press Monit 1999;4:295–301. 24. Stergiou GS, Baibas NM, Gantzarou A, Skeva II, Kalkana CB, Roussias LG, Mountokalakis TD: Reproducibility of home, ambulatory, and clinic blood pressure: implications for the design of trials for the assessment of antihypertensive drug efficacy. Am J Hypertens 2002;15:101–104. 25. Pearce KA, Grimm RH, Rao S, Svendsen K, Liebson PR, Neaton JD, Ensrud K: Population-derived comparisons of ambulatory and office blood pressures. Implications for the determination of usual blood pressure and the concept of white coat hypertension. Arch Intern Med 1992;152:750 –756. 26. Whelton PK: Blood pressure in adults and the elderly, in Bulpitt CJ (ed): Epidemiology of Hypertension. Amsterdam, Elsevier, 1985, pp 51–67. 27. Vasan RS, Larson MG, Leip EP, Kannel WB, Levy D: Assessment of frequency of progression to hypertension in nonhypertensive participants in the Framingham Heart Study: A cohort study. J Lancet 2002;358:1682–1686.
2003; 16:640 – 645
A 5-Year Follow-Up of Ambulatory Blood Pressure in Healthy Older Adults Iris B. Goldstein, David Shapiro, and Donald Guthrie Background: This study assessed 5-year changes in ambulatory blood pressure (ABP) in healthy, older individuals and determined the extent to which it could be predicted from earlier BP measures and other cardiovascular risk factors. Methods: A total of 162 men and women, aged 55 to 79 years, with no prior medical disorders, completed a medical examination and two 24-h ABP sessions. The procedures were repeated 5 years later in 80% (130) of these subjects. A modified hierarchical regression analysis was used to determine whether initial ABP and casual blood pressure (CBP) measures and demographic and physical examination data could predict ABP in 5 years. Results: The CBP and most ABP levels during waking and sleep increased after 5 years. However, CBP remained in the normotensive range for 73% of the subjects. The
T
ABP variability tended to decrease over time. The ABP and CBP measures accounted for at least 50% of the variance in the prediction of ABP level after 5 years. In comparison, the predictability of ABP variability was quite low, particularly during sleep (⬍30% of the variance accounted for). Conclusions: The ABP and CBP were good predictors of future ABP level in healthy older subjects, but ABP variability was more difficult to predict. Except for age, none of the standard cardiovascular risk factors contributed significantly to the prediction of ABP level or variability. Am J Hypertens 2003;16:640 – 645 © 2003 American Journal of Hypertension, Ltd. Key Words: Ambulatory blood pressure, blood pressure variability, older subjects.
he superiority of 24-h ambulatory blood pressure (ABP) monitoring over casual blood pressure (CBP) has been repeatedly demonstrated. Not only is it less likely to give biased values, but ABP appears to be a more sensitive measure of peripheral vessel end organ damage and a better predictor of cardiovascular morbidity and mortality.1 Studies comparing ABP measures over the short term (days or months),2 and up to 2 years,3 concluded that ABP mean levels exhibited reproducibility over time that was not only adequate, but generally exceeded that of measurements of office blood pressure (BP). Reproducibility of ABP variability, on the other hand, was poor.2,4 However, little is known about what happens to ABP level and variability over several years. Longitudinal studies of CBP suggest that over intervals of 5 to 10 years BP gradually increases with age, with the likelihood of a plateau or even a decrease in diastolic BP (DBP) in the fifth or sixth decade.5,6 The strongest predictor of CBP among middle-age men was initial CBP measurement.7,8 In addition, the 24-h mean of intra-arterial ABP best
predicted noninvasive ABP after 10 years in middle-aged men with varying BP levels.9 To our knowledge, no information is available on long-term progression of ABP in healthy, older men and women. The primary goal of this study was to compare initial ABP with assessments repeated after 5 years. We wanted to follow the progression of ABP over time in older men and women who had never had major chronic illnesses, nor taken medication that affected the cardiovascular or central nervous system. Moreover, we wanted to determine whether ABP could be predicted from initial ABP and CBP assessments, and demographic and physical examination data.
Received December 9, 2002. First decision March 4, 2003. Accepted March 25, 2003. From the Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles, California. This study was supported by Research Grant AG-11595 from the
National Institute of Aging.
0895-7061/03/$30.00 doi:10.1016/S0895-7061(03)00906-3
Methods Phase 1 Subjects The initial sample (phase 1) included 162 subjects (94 women, 68 men), aged 55 to 80 years, living in the community. Average education level was 15.5 ⫾ 5.8 (mean ⫾ SD) years. Exclusions involved subjects’ reports
Address correspondence and reprint requests to Dr. Iris B. Goldstein, UCLA Department of Psychiatry, 760 Westwood Plaza, Los Angeles, CA 90095-1759; e-mail: [email protected] © 2003 by the American Journal of Hypertension, Ltd. Published by Elsevier Inc.
AJH–August 2003–VOL. 16, NO. 8
AMBULATORY BP IN OLDER ADULTS
641
Table 1. Characteristics of men (n ⫽ 47) and women (n ⫽ 69) during phase 1 and phase 2 Men Characteristics 2
BMI, kg/m Exercise, hours Alcohol, drinks/week Coffee, cups/day Glucose, mg/dL Total chol, mg/dL LDL chol, mg/dL HDL chol, mg/dL Triglycerides, mg/dL BP level, mm Hg SBP casual DBP casual SBP wake DBP wake SBP sleep DBP sleep BP variability, mm Hg SBP wake DBP wake SBP sleep DBP sleep
Phase 1
Women Phase 2
Phase 1
Phase 2
Significance
25.5 8.9 3.8 2.2 109.3 223.0 142.6 48.8 161.0
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
2.7 6.1 4.8 2.2 27.0 33.6 28.4 12.1 99.2
25.6 9.9 5.8 2.3 131.8 201.6 128.0 48.0 126.0
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
2.8 8.1 5.3 2.0 33.7 32.8 28.0 10.2 69.0
23.8 11.1 2.4 1.7 116.2 233.6 133.3 70.6 139.0
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
2.8 8.9 3.5 2.0 29.5 36.5 38.1 20.0 68.6
24.0 11.5 3.9 2.0 142.9 213.4 121.6 67.9 119.4
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
3.3 9.2 4.2 1.1 45.5 25.3 24.7 14.3 60.6
122.5 75.0 127.1 74.5 110.8 63.6
⫾ ⫾ ⫾ ⫾ ⫾ ⫾
13.4 9.7 10.7 6.8 11.3 7.5
129.2 76.4 133.2 74.8 116.0 64.4
⫾ ⫾ ⫾ ⫾ ⫾ ⫾
14.3 9.7 11.8 7.6 12.4 7.6
114.4 69.0 123.6 72.6 106.1 59.7
⫾ ⫾ ⫾ ⫾ ⫾ ⫾
12.4 7.7 11.0 6.6 11.3 6.2
123.8 72.1 129.4 72.0 113.7 61.5
⫾ ⫾ ⫾ ⫾ ⫾ ⫾
15.6 8.9 12.9 7.4 13.0 7.2
G†, P‡ G†, P‡ P‡
13.4 9.6 9.6 7.5
⫾ ⫾ ⫾ ⫾
2.2 2.1 3.0 2.5
12.4 8.1 9.6 6.9
⫾ ⫾ ⫾ ⫾
2.8 2.2 4.3 2.6
14.2 11.7 9.7 7.5
⫾ ⫾ ⫾ ⫾
3.3 3.3 3.1 2.7
12.8 9.0 9.3 6.7
⫾ ⫾ ⫾ ⫾
3.3 2.3 3.7 2.6
P‡ G‡, P‡, G⫻P*
G† P‡ P‡ G*, P‡ P‡ G‡ P‡
P‡ G†, P*
P*
Values are derived from ANCOVA with gender as a group variable and phase as a repeated measure. BMI ⫽ body mass index; G ⫽ Gender; P ⫽ Phase; chol ⫽ cholesterol; BP ⫽ blood pressure; SBP ⫽ systolic BP; DBP ⫽ diastolic BP. * P ⬍ .05; † P ⬍ .01; ‡ P ⬍ .001. Table does not include 14 subjects taking medications that influence BP.
of any serious current or prior illness or medications influencing either the cardiovascular or central nervous system, history of hypertension, drug or alcohol abuse, head injuries, or obesity (body mass index [BMI] ⬎30 kg/m2). Information was confirmed by medical examination (see Procedures) and prior medical records. Abnormal findings provided further basis for exclusion.10 Psychological testing and a mental status examination excluded subjects with cognitive abnormalities or psychiatric disorders.10 Although not previously diagnosed with hypertension, 14 subjects had CBP within hypertensive levels.11
Procedures
Phase 2 Subjects
Ambulatory Monitoring
Of the 162 subjects who participated in phase 1, 130 (80%) completed phase 2 after 4.9 ⫾ 0.63 years (range, 4 to 6 years). Three subjects could not be reached, 3 moved out of the area, 2 died (1 suicide, 1 lung cancer), 1 had a stroke, 6 had incomplete ABP data, and 17 declined for personal reasons. Data presented here are based on 75 women and 55 men who participated in both phases of the study (Table 1 shows sample characteristics). Ethnic composition was as follows: 99 white, 20 Asians, 9 African Americans, 1 Latino, and 1 Native American. When comparisons were made between subjects who completed the follow-up phase and those who did not, there were no significant differences (t test or 2) for BP, demographic, or physical examination data. Subjects gave informed consent, approved by the UCLA Institutional Review Board.
A laboratory assistant obtained three standard CBP readings11 after 5 min of sitting and just before attachment of the ABP monitor. Three additional CBP readings were taken 1 week later before the second ABP session. The ABP was recorded by the Accutracker II (Suntech Medical Instruments, Raleigh, NC), which has been used widely in clinical and research studies with established reliability and validity.12 The ambulatory recorder was programmed to operate three times an hour on a variable schedule during waking hours and once an hour during sleep (based on subjects’ estimates of times of sleep and awakening). An activity monitor (Mini-Motionlogger, Ambulatory Monitoring Inc., Ardsley, NY) was used to confirm the differentiation of sleep from awake readings and to account for effects of activity on BP.13 Ambulatory data were edited for artifacts based on
Identical procedures were performed during phases 1 and 2. Subjects were given physical and mental status examinations by the physician, including a complete health history, 12-lead electrocardiogram, urinalysis, and chemical panel. Laboratory tests were done by standard techniques (SmithKline Beecham Clinical Laboratories, Van Nuys, CA). Changes in health during the 5-year period were noted; medical records were obtained from subjects’ physicians.
642
AMBULATORY BP IN OLDER ADULTS
Accutracker error codes. Editing was done by set rules.14 Classification of each reading as wake or sleep was based on diary entries and postsession reports. Only nighttime sleep values were included in the sleep category and daytime nonsleep values in the wake category. Means were obtained for six CBPs and for two sessions of ambulatory systolic BP (SBP) and DBP (wake level, sleep level, wake variability, sleep variability). For a given subject, variability was based on the SD of the wake and of the sleep period for a given day. The coefficient of variation was also used, but it did not alter results. Only findings with the SD are included. Data Analyses The BP, demographic, and blood chemistry data were looked at separately for men and women during each phase of the study in a group (gender ⫽ men/women) ⫻ repeated measures (phase ⫽ 1/2) ANOVA. Table 1 is based only on 116 subjects not taking medications with antihypertensive effects. Findings for the entire sample (n ⫽ 130) were the same as for the smaller sample (n ⫽ 116). Separate analyses indicated that activity had no appreciable effect on ABP values.13 A modified hierarchical regression analysis was constructed to determine whether ABP after 5 years could be predicted by its initial value or by other variables. Data for men and women were combined. For each measure, four sets of predictor variables were specified: initial value of the measure, initial values of other BP measures (ambulatory wake and sleep level, ambulatory wake and sleep variability, and CBP), demographic variables (age, gender, education, BMI, exercise, alcoholic drinks, caffeine), and blood chemistry measures (total cholesterol, HDL and LDL cholesterol, triglycerides, 2-h glucose after 75-g glucose load). (Substituting 5-year change for initial values of demographic and blood chemistry measures did not alter findings in any models.) Based on prior findings indicating that measures of CBP and ABP level were highly intercorrelated but were not related to variability,15 we hypothesized that measures of BP level would predict future BP level and variability measures would be predictive of future ABP variability. Therefore, variability measures were not used to predict BP level measures, nor were level measures used to predict variability. The first step of each regression forced in the initial value of the target variable. Second, we considered the other BP measures and entered them individually according to significance, based on the P value (⬍.05) within the step after preceding variables were included. Following the same procedure, the final step involved entering the remaining demographic and blood chemistry variables. This process avoids loss of predictive information due to colinearity within the steps, while still identifying variables that predict the target variable.
AJH–August 2003–VOL. 16, NO. 8
Results Health and BP Status of Subjects During the 5-year period between the two phases, 11 people were diagnosed and treated for hypertension, 3 were taking other medications with antihypertensive effects (1 heart disease, 2 benign prostatic hypertrophy), 1 had a stroke, and 3 had transient ischemic attacks. There were only 4 smokers during both phases of the study. The gender ⫻ phase ANOVA (Table 1) indicated that from phase 1 to phase 2 the following non-BP variables did not change: BMI, hours of exercise, coffee intake, HDL cholesterol. However, alcohol intake and glucose level increased and total cholesterol, LDL cholesterol, and triglycerides decreased. Because statin use increased over time, we compared lipids during phase 2 in individuals receiving statins with those who were not. We found that medication use was associated with a greater decrease in LDL and total cholesterol at follow-up (t tests, P ⬍ .05). Triglycerides and HDL cholesterol were unaffected. Also, independent of phase, men had higher BMI and total cholesterol and lower HDL cholesterol than women. Higher HDL cholesterol in women was not the result of hormone replacement therapy (HRT), because once we corrected for age, ANOVA indicated no differences in lipids between 32 women on and 43 women off HRT. ANOVA for BP showed phase effects with increases from phase 1 to phase 2 for casual SBP and DBP, SBP wake, and SBP and DBP sleep (Table 1). The SBP and DBP wake variability, and DBP sleep variability decreased over time. With regard to gender effects, compared to women, men had higher casual SBP and DBP and DBP sleep and lower DBP wake variability. There was a gender ⫻ phase interaction for DBP wake variability, with women showing a decrease of 2.7 mm Hg from phase 1 to phase 2 compared to a 1.5 mm Hg decrease in men. Shifts in BP Over Time Subjects who completed both phases of the study were classified in terms of standard definitions11 of CBP levels during each phase as follows: normal, SBP ⬍130 mm Hg and DBP ⬍85 mm Hg; high normal, SBP ⫽ 130 to 139 mm Hg or DBP ⫽ 85 to 89 mm Hg; hypertension, SBP ⬎139 mm Hg or DBP ⬎89 mm Hg or currently taking antihypertensive medication (11 subjects). (Three subjects who were taking BP-lowering drugs for nonhypertensive problems were excluded, leaving a total of 127 subjects.) A 2 analysis of change in BP classification from phase 1 to phase 2 indicated no differences between men and women. Almost 75% of subjects (n ⫽ 71) maintained normal BP status from phase 1 to phase 2. Of those who were high normal, 62% became hypertensive. Almost all subjects (13 of 14) who were hypertensive during phase 1 remained hypertensive (Table 2), although only 4 of them were later treated with drugs for hypertension.
AJH–August 2003–VOL. 16, NO. 8
643
AMBULATORY BP IN OLDER ADULTS
Table 2. Subjects’ casual blood pressure (CBP) during phase 2 (P2), relative to phase 1 (P1) blood pressure Normal P2 Normal P1 High normal P1 Hypertensive P1 Total N
High Normal P2
Hypertensive P2
n
%
n
%
n
%
Total n
71 1 1 73
73.2 6.3 7.1
16 5 0 21
16.5 31.2 0.0
10 10 13 33
10.3 62.5 92.9
97 16 14 127
Percent (%) values are based on row totals. Table does not include three subjects who were taking blood pressure-lowering drugs but were not diagnosed with hypertension.
Prediction of ABP Correlations between phase 1 and phase 2 measures of BP level (CBP and ABP) were moderately high (0.62 to 0.71) with lower correlations for ABP variability (0.28 to 0.48) (Table 3). With the modified hierarchical regression, phase 1 ABP accounted for at least 50% of the variance in ABP (SBP and DBP wake and sleep) during phase 2 (Table 4). In the models for SBP and DBP wake and SBP sleep approximately 2% to 5% more was added by CBP. In general, variability was much less predictable from phase 1 measures, particularly during sleep. Age accounted for an additional 4% to 5% of the variance for DBP wake and SBP and DBP wake variability. No other demographic or blood chemistry variables added significantly to the variance.
Discussion Health and BP Status of Subjects During the 5-year interval between phases there were no significant changes in BMI, exercise, or caffeine intake. Subjects drank more alcohol and had lower triglycerides, total cholesterol, and LDL cholesterol. Improvement in LDL and total cholesterol appears to have been associated with increased use of statins. Higher HDL cholesterol
Table 3. Correlations between phase 1 and phase 2 blood pressure Blood Pressure
Correlation Coefficient
SBP casual DBP casual SBP wake DBP wake SBP sleep DBP sleep SBP wake variability DBP wake variability SBP sleep variability DBP sleep variability Abbreviations as in Table 1. Values are in mm Hg. n ⫽ 116.
0.71 0.69 0.70 0.64 0.64 0.62 0.48 0.47 0.32 0.28
P .0001 .0001 .0001 .0001 .0001 .0001 .0001 .0001 .0005 .003
(reflected in higher total cholesterol) in women than in men, could not be explained by HRT, which is sometimes associated with elevations in HDL.16 The decline in total and LDL cholesterol levels with age corresponds to the literature. Ferrara et al17 found the largest total cholesterol drop after 8 years among the oldest men and women (65 to 79 years at baseline) in their study. However, findings regarding HDL cholesterol levels were more equivocal.17,18 Improvements in lipid levels in our subjects could have been influenced by lifestyle changes. Most subjects claimed that their participation was due to a general interest in health, reflected in exercise rate (approximately 10 h/w) and in the small number of smokers. Mean CBP levels for men and women combined increased approximately 8/2 mm Hg from phase 1 to phase 2. Larger increases in SBP than DBP among older individuals are consistent with prior findings.19 Although both SBP and DBP increase with age, DBP plateaus or even declines in later years.5,6 Compared to women, men in the current study had consistently higher CBP. In contrast, Cornoni-Huntley et al.20 reported that after age 55, prevalence rates for hypertension in women exceeded men. However, Pearson et al19 suggested that gender cross-over in BP was less likely to occur in healthy, carefully screened individuals and showed that both longitudinally and cross-sectionally men’s BP levels exceeded those of women at all ages, with a narrowing of gender differences at age 70. The ABP levels also increased over time, with only DBP wake showing no changes. Moreover, during both phases, DBP sleep was the only ambulatory level measure that was higher in men than in women. The SBP and DBP wake variability and DBP sleep variability decreased from phase 1 to 2, an unexpected result in view of previous findings of greater variability in older than in younger adults.21,22 These differences may be due to discrepancies between cross-sectional and prospective studies. The current results also showed that, compared to men, women exhibited greater DBP wake variability, which also dropped more from phase 1 to phase 2. A gender difference in ABP variability was found previously in older but not in younger adults.22
644
AMBULATORY BP IN OLDER ADULTS
AJH–August 2003–VOL. 16, NO. 8
Table 4. Modified hierarchical regression to predict phase 2 ambulatory blood pressure (ABP) Follow-up BP Variables SBP wake DBP wake
SBP sleep DBP sleep
SBP wake variability DBP wake variability SBP sleep variability DBP sleep variability
Predictors
Regression Coefficient
P
Cumulative % R2
SBP wake SBP casual DBP wake DBP sleep DBP casual Age SBP sleep SBP wake DBP sleep DBP wake SBP wake DBP casual SBP wake variability Age DBP wake variability Age SBP sleep variability SBP wake variability DBP sleep variability SBP wake variability
0.545 0.241 0.295 0.251 0.303 ⫺0.214 0.464 0.296 0.482 0.296 ⫺0.189 0.212 0.419 0.236 0.435 0.198 0.214 0.282 0.184 0.272
.0001 .015 .003 .002 .001 .001 .0001 .004 .0001 .019 .054 .023 .0001 .005 .0001 .018 .022 .003 .051 .004
53.1 55.5 47.3 50.6 55.9 60.3 47.6 51.3 45.2 51.0 52.8 55.0 23.4 28.6 22.1 25.9 10.1 16.9 8.0 14.4
Other abbreviations as in Tables 1 and 3. ABP was predicted from phase 1 ambulatory and casual blood pressure and from demographic (age, gender, education, body mass index, exercise, alcohol, caffeine), and blood chemistry variables (total cholesterol, HDL, LDL, triglycerides, glucose). n ⫽ 116.
In contrast to our findings, CBP has generally been found to be higher than waking ABP,23 probably as a consequence of the white coat effect and the limited number of measurements taken in stressful clinic settings.24 However, increasing the number of office measurements can lead to lower CBP.25 Lower CBP compared with ABP in our subjects could have been due to CBP being recorded during the second and third visits (not the first) in a stress-free, nonmedical setting by research assistants. Furthermore, there is evidence that at least in older, white men, white coat hypertension is rare.25 Shifts in BP Over Time Individuals who start life with low BP are more likely to exhibit smaller BP increases with age.26 The fact that the CBP of 73% of our subjects remained in the normotensive range after 5 years may have been due to their initial good health and relatively low BP. Of the 16 subjects with high normal BP during phase 1, 62% became hypertensive during phase 2. Although casual BP differed between subjects who became hypertensive and those who did not (t test, P ⫽ .007), neither ABP level nor variability differed significantly between the two groups. Our results are consistent with findings that older individuals with high normal CBP are most likely to develop hypertension, even after just 4 years.27 Prediction of ABP Just as current CBP predicted CBP 5 years later,7 wake and sleep ABP predicted themselves fairly well after 5
years. These results are in accord with a 10-year follow-up study, using 24-h BP mean in place of waking BP.9 We also found that multiple measures of ABP level during wake and sleep as well as CBP added significantly to the total variance predicted for ABP wake and sleep. The relationship between different measures of BP level is in keeping with their high intercorrelations and their occurrence on a single BP factor.15 Age was the only non-BP variable contributing significantly to the total predicted variance (4.4% for DBP wake). Although at least 50% of the variance was explained, the remaining variance may be due to other factors that are more difficult to assess, such as changes in experience and lifestyle. In comparison to ABP level, predictability of variability was low. Correlations between phase 1 and phase 2 BP indicated higher associations between levels than between variability measures, with decreasing values during sleep. The ABP variability has been found to show poor reproducibility,4 which Ward et al2 suggested may be due to frequent behavior changes and the responsiveness of BP to “external demands and internal homeostatic requirements.” The prediction of ABP was not influenced by any of the following standard risk factors: gender, education, BMI, hours of exercise, alcohol intake, caffeine intake, lipid level, or glucose level. Only age added to the variance for some ABP measures. A power analysis of our data showed that, with 116 subjects, if any of these other variables had been involved in predicting BP, they would have showed up in the analyses. Moreover, a 10-year longitudinal study
AJH–August 2003–VOL. 16, NO. 8
indicated that, other than initial BP, baseline characteristics did not significantly predict either future CBP or 24-h BP in middle-aged men.9 Summary and Conclusions Increases in subjects’ systolic ABP level over time are consistent with our CBP findings and those of other investigations of older subjects.5,6,19 In addition, individuals who started out with lower ABP were more likely to maintain low levels after 5 years. Our results also showed that women did not differ very much from men in the progression of BP to higher levels. With regard to the prediction of ABP, prior BP measures accounted for at least 50% of the variance. Variability was more difficult to predict. Finally, except for age, no other standard cardiovascular risk factor contributed to the prediction of ABP. Because of the strong association of ABP level and variability with end organ damage and cardiovascular morbidity and mortality,1,21 it is vital that future investigations focus on long-term studies of ABP, particularly in healthy, older populations where limited information is available.
References 1.
Meredith PA, Perloff D, Mancia G, Pickering T: Blood pressure variability and its implications for antihypertensive therapy. Blood Press 1995;4:5–11. 2. Ward MM, Turner JR, Johnston DW: Temporal stability of ambulatory cardiovascular monitoring. Ann Behavioral Med 1994;16:3– 11. 3. Gosse P, Bougaleb M, Clementy J: Long-term reproducibility of ambulatory blood pressure monitoring. The´rapie 1996;51:5–9. 4. Palatini P, Mormino P, Canali C, Santonastaso M, De Venuto G, Zanata G, Pessina A: Factors affecting ambulatory blood pressure reproducibility. Results of the HARVEST trial. Hypertension 1994; 23:211–216. 5. Kannel WB: Blood pressure as a cardiovascular risk factor. Prevention and treatment. JAMA 1996;275:1571–15776. 6. Landahl S, Bengtsson C, Sigurdsson JA, Svanborg A, Sva¨rdsudd K: Age-related changes in blood pressure. Hypertension 1986;8:1044 – 1049. 7. Carroll D, Smith GD, Sheffield D, Shipley MJ, Marmot MG: Pressor reactions to psychological stress and prediction of future blood pressure: data from the Whitehall II study. Br Med J 1995;310:771– 775. 8. Carroll D, Smith GD, Shipley MJ, Steptoe A, Brunner EJ, Marmot MG: Blood pressure reactions to acute psychological stress and future blood pressure status: a 10-year follow up of men in the Whitehall II Study. Psychosom Med 2001;63:737–743. 9. Jokiniitty JM, Majahalme SK, Ka¨ho¨nen MAP, Tuomisto MT, Turjanmaa MH: Prediction of blood pressure level and need for antihypertensive medication: 10 years of follow-up. J Hypertens 2001; 19:1193–1201. 10. Goldstein IB, Bartzokis G, Hance D, Shapiro D: Relationship between blood pressure and subcortical lesions in healthy elderly people. Stroke 1998;29:765–772.
AMBULATORY BP IN OLDER ADULTS
645
11. Joint National Committee on Prevention: The Sixth Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Arch Intern Med 1997;157: 2413–2446. 12. Jyothinagaram SG, Watson D, Padfield P: Suntech Accutracker ambulatory blood pressure monitor— clinical validation. J Ambulatory Monitor 1990;3:63–67. 13. Shapiro D, Goldstein IB: Wrist actigraph measures of physical activity level and ambulatory blood pressure in healthy elderly persons. Psychophysiology 1998;35:305–312. 14. Goldstein IB, Jamner LD, Shapiro D: Ambulatory blood pressure and heart rate in healthy male paramedics during a workday and a nonworkday. Health Psychol 1992;11:48 –54. 15. Goldstein IB, Shapiro D, La Rue A, Guthrie D: Relationship between 24-hour ambulatory blood pressure and cognitive function in healthy elderly people. Aging, Neuropsychol & Cognition 1998;5: 215–224. 16. Rijpkema AHM, van der Sanden AA, Ruijs AHC: Effects of postmenopausal oestrogen–progestogen replacement therapy on serum lipids and lipoproteins: a review. Maturitas 1990;12:259 –285. 17. Ferrara A, Barrett-Connor E, Shan J: Total, LDL, and HDL cholesterol decrease with age in older men and women. The Rancho Bernardo Study 1984 –1994. Circulation 1997;96:37–43. 18. Abbott RD, Sharp DS, Burchfiel CM, Curb JD, Rodriguez BL, Hakim AM, Yano K: Cross-sectional and longitudinal changes in total and high-density-lipoprotein cholesterol levels over a 20-year period in elderly men: The Honolulu Heart Program. Ann Epidemiol 1997;7:417–424. 19. Pearson JD, Morrell CH, Brant LJ, Landis PK, Fleg JL: Ageassociated changes in blood pressure in a longitudinal study of healthy men and women. J Gerontol Med Sci 1997;52A:M177– M183. 20. Cornoni-Huntley J, LaCroix AZ, Havlik RJ: Race and sex differentials in the impact of hypertension in the United States. The National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study. Arch Intern Med 1989;149:780 –788. 21. Mancia G, Di Rienzo M, Grassi G, Parati G: Blood pressure variability and reflex control in the elderly. Aging 1995;7:3–9. 22. Jaquet F, Goldstein IB, Shapiro D: Effects of age and gender on ambulatory blood pressure and heart rate. J Hum Hypertens 1998; 12:253–257. 23. Mancia G, O’Brien E, Imai Y, Redon J: Task Force II: ambulatory blood pressure monitoring in population studies. Blood Press Monit 1999;4:295–301. 24. Stergiou GS, Baibas NM, Gantzarou A, Skeva II, Kalkana CB, Roussias LG, Mountokalakis TD: Reproducibility of home, ambulatory, and clinic blood pressure: implications for the design of trials for the assessment of antihypertensive drug efficacy. Am J Hypertens 2002;15:101–104. 25. Pearce KA, Grimm RH, Rao S, Svendsen K, Liebson PR, Neaton JD, Ensrud K: Population-derived comparisons of ambulatory and office blood pressures. Implications for the determination of usual blood pressure and the concept of white coat hypertension. Arch Intern Med 1992;152:750 –756. 26. Whelton PK: Blood pressure in adults and the elderly, in Bulpitt CJ (ed): Epidemiology of Hypertension. Amsterdam, Elsevier, 1985, pp 51–67. 27. Vasan RS, Larson MG, Leip EP, Kannel WB, Levy D: Assessment of frequency of progression to hypertension in nonhypertensive participants in the Framingham Heart Study: A cohort study. J Lancet 2002;358:1682–1686.