Association of Heart Rate With Blood Pressure Variability: Implications for Blood Pressure Measurement

nature publishing group original contributions Association of Heart Rate With Blood Pressure Variability: Implications for Blood Pressure Measurement Amos Cahan 1, Iddo Z. Ben-Dov 2 and Michael Bursztyn 1 Background Antihypertensive β-blocker use is associated with greater intervisit blood pressure variability (BPV) and with less favorable outcomes compared to other antihypertensive agents. A theoretical model demonstrated that accuracy and precision of BP measurement are affected by heart rate (HR) at a constant cuff deflation rate. We aimed to examine the empirical relationship between HR and BPV in a clinical setting. Methods Intratracing variability in ambulatory BP monitoring (ABPM) were analyzed in search of a link between BPV and HR. BPV was expressed as standard deviation (s.d.), coefficient of variation (CV), and variability independent of the mean (VIM). Results In a dataset of 4,693 subjects, HR was inversely associated with BPV and independently explained 1.3% of between-subject variation in s.d. of awake systolic BP (1.5% of CV and VIM). Linear regression suggested 0.5 mm Hg increase in s.d. of systolic BP per 10 beats per minute (bpm) decrease in HR. In a subset of 1,019 patients with available data on medications, HR was independently and inversely related with awake systolic BPV (P < 0.0001), more so in diuretic (P < 0.050) and renin angiotensin system antagonists-treated (P < 0.050) patients. Associations of β-blockade with increased BPV were abolished by model-adjustment for HR. In another subset of patients who were monitored twice (n = 635), HR had a mild (0.6%) but significant (P < 0.05) inverse contribution to the change in awake systolic BPV between repeated monitoring. Conclusions Ambulatory BPV is inversely related to HR and is not increased in referred patients treated with β-blockers after correction for HR. Keywords: ambulatory blood pressure monitoring; beta-adrenergic blockers; blood pressure; blood pressure measurement; blood pressure variability; bradycardia; heart rate; hypertension American Journal of Hypertension, advance online publication 8 December 2011; doi:10.1038/ajh.2011.230 Historically, hypertensive patient risk stratification for cardiovascular outcomes was based on the average blood pressure (BP) as measured in the doctor s office. In the last 2 decades, ambulatory BP monitoring (ABPM) and self-operated home devices provide additional useful information. Other characteristics of BP such as diurnal variation and BP variability (BPV) 1,2 were found to have considerable prognostic value. Recently, increased visit-to-visit BPV has been shown in large cohorts to predict all-cause mortality 3 and the risk of stroke and other cardiovascular outcomes 4 independently of the average systolic BP (SBP). In this regard, the effect of different classes of antihypertensive medications on BPV was suggested to have role in the protection provided against cardiovascular events. Decreased within-individual BPV among patients treated 1 Department of Medicine, Hadassah Hebrew University Medical Center, Jerusalem, Israel; 2 The Rockefeller University, New York, New York, USA. Correspondence: Amos Cahan (amosc@hadassah.org.il) Received 20 June 2011; first decision 27 July 2011; accepted 23 October 2011. 2012 American Journal of Hypertension, Ltd. with a dihydropyridine calcium channel blocker, compared to β-blockers, was shown to partly account for their reduced risk of stroke. 5 In a large meta-analysis of randomized clinical trials, 6 considerable differences were found in interindividual SBP variation between groups of patients treated with different antihypertensive agent classes. Interindividual variation was regarded as a surrogate to within-individual variation since the latter was reported to account for about half of the former. Whereas treatment with calcium channel blockers was associated with reduced interindividual BPV, increased BPV was found in patient groups treated with other drug classes. The authors proposed that increased BPV accounts in part for the higher rates of stroke seen in hypertensive patients treated with β-blockers compared to calcium blockers. Recently, within-subject BPV was assessed in the X-CELLENCE trial among hypertensive patients receiving candesartan, indapamide, amlodipine or placebo. 7 Whereas the finding that all drug classes reduced BPV was expected, significant differences were found among them in the extent to which BPV decreased. The lowest BPV was found among patients treated with amlodipine. AMERICAN JOURNAL OF HYPERTENSION VOLUME 25 NUMBER 3 313-318 march 2012 313

original contributions Bradycardia and BP Variability 156 True blood pressure 154 Perceived blood pressure Pressure, mm Hg Pressure, mm Hg 152 150 148 156 154 152 Cuff deflation rate: 4 mm Hg/s 2 mm Hg/s Heart rate = 80 150 140 Heart rate = 50 0 5 10 15 20 Time, s Figure 1 Theoretical relationship between deflation rate, heart rate, and systolic blood pressure measurement detection lag. The chosen heart rates, 50 (bottom) and 80 (top), represent the 5th and 95th percentiles of resting heart rate (as inferred from our sleep monitoring data), respectively. Note underestimation of true blood pressure accompanied by amplified measurement scattering as a result of faster deflation rate (compare 2 and 4 mm Hg/s) or slower heart rate (compare upper and lower panels). Although almost 60 years ago BP measurement recommendations noted that the cuff deflation rate should vary with heart rate (HR) (2 3 mm Hg per heartbeat), 8 this method has been abandoned and current guidelines recommend a constant deflation rate of 2 mm Hg/s. 9 When using a constant cuff deflation rate, the number of heartbeats (i.e., the number of Korotkoff sounds or oscillometric pulse waves) upon which SBP and diastolic BP (DBP) are determined, is dependent on the patient s HR (Figure 1). That is, a smaller sample size of heartbeats is used when measuring BP in a patient with bradycardia compared to a nonbradycardic patient. A computer simulation of different cuff deflation rate protocols using a constant BP and hrs in the range of 40 120/min was used to determine the error in BP measurement contributable to HR. 10 The error using a constant cuff deflation rate was greater at lower HRs. For example, at a constant deflation rate of 3 mm Hg/s, the maximal error in determining systolic BP was 1.5 and 4.3 mm Hg at a HR of 120 and 40/min, respectively. The maximal error in determining diastolic BP was +1.4 and +4.3 mm Hg at HRs of 120 and 40/min, respectively. In a small study by Zheng et al, 11 participants BP was measured using an oscillometric device. Then, a faster (two- to sevenfold) cuff deflation rate was simulated using the collected data. With manual data extraction, the mean measured SBP declined from 132 to 125 mm Hg when a sevenfold deflation rate was assumed. The average within-subject systolic BPV was 4.0 and 7.5 mm Hg at baseline deflation rate and seven times faster deflation rate, respectively. The mean measured DBP increased from 78 to 84 mm Hg when a sevenfold deflation rate was assumed, and diastolic BPV increased from 2.5 to 6.0 mm Hg. Whereas simulated results of average SBP and DBP using automated modeling were robust to changes in deflation rate, systolic BPV increased by 0.9 and 2.2 mm Hg and diastolic BPV increased by 1.1 and 1.3 mm Hg, depending on the model used to interpret the oscillometric pulse waves (linear vs. polynomial, respectively). Thus, in this study, variability was positively related, and precision inversely related, with cuff deflation rate, even with modeled data extraction. These data imply that, in lower HRs both accuracy and precision of BP determination may be compromised by 314 march 2012 VOLUME 25 NUMBER 3 AMERICAN JOURNAL OF HYPERTENSION

Bradycardia and BP Variability original contributions Table 1 Demographic and clinical characteristics of patients referred to ambulatory blood pressure monitoring Variable All patients, N = 4,693 Patients with Rx data, N = 1,019 Patients with repeat ABPM, N = 635 Age, years 55 ± 16 57 ± 17 55 ± 14 Sex, % women 52 54 53 Diabetes, % 9 12 8 Treated HTN, % 58 59 63 Awake BP, mm Hg 141 ± 16/ 82 ± 10 Awake BP s.d., mm Hg 14 ± 4/ 10 ± 3 Awake BP CV, % 10 ± 3/ 12 ± 4 Awake BP VIM 0.09 ± 0.03/ 3.7 ± 1.0 139 ± 14/ 81 ± 11 12 ± 4/ 9 ± 2 9 ± 2/ 11 ± 3 0.08 ± 0.02/ 3.5 ± 0.9 144 ± 16/ 83 ± 11 15 ± 5/ 10 ± 3 10 ± 3/ 12 ± 4 0.09 ± 0.03/ 3.8 ± 1.1 Awake HR, bpm 74 ± 11 74 ± 12 73 ± 11 Sleep BP, mm Hg 125 ± 18/ 69 ± 10 Sleep BP s.d., mm Hg 12 ± 5/ 9 ± 3 Sleep BP CV, % 9 ± 4/ 13 ± 5 Sleep BP VIM 0.09 ± 0.04/ 2.1 ± 0.8 123 ± 15/ 68 ± 10 10 ± 4/ 8 ± 3 8 ± 3/ 13 ± 4 0.08 ± 0.03/ 2.1 ± 0.7 128 ± 20/ 70 ± 11 13 ± 5/ 9 ± 4 10 ± 4/ 13 ± 5 0.10 ± 0.04/ 2.2 ± 0.9 Sleep HR, bpm 64 ± 9 64 ± 9 63 ± 9 ACEi/ARB, % 35 β-blockers, % 34 Ca-blockers, % 24 Diuretics, % 27 α-blockers, % 6 Session interval, m 15 (IQR 5 36) ABPM, ambulatory blood pressure monitoring; ACEi, angiotensin-converting enzyme inhibitors; ARB, angiotensin receptor blockers; BP, blood pressure; bpm, beats per minute; BPV, blood pressure variability; Ca-blockers, calcium channel blockers; CV, coefficient of variation; HR, heart rate; HTN, hypertension; IQR, inter-quartile range; Rx, medication; VIM, variability independent of mean. Table 2 Stepwise multivariable linear regression models evaluating contributors to awake systolic blood pressure variability (for diastolic and sleep variability see Supplementary Table S1 online) Measure of awake systolic Model covariable Attribute blood pressure variability s.d. (mm Hg) CV (%) VIM Mean systolic BP Partial r a 0.361 (mm Hg) P value a <0.0001 Change in R 2b 12.7% Age (years) Partial r 0.108 0.116 0.115 Change in R 2 3.3% 4.1% 4.0% Gender (0 = m, 1 = f) Partial r 0.133 0.132 0.133 Change in R 2 1.2% 1.4% 1.4% Hypertension Rx (0 = no, 1 = yes) Partial r 0.076 0.081 0.081 Change in R 2 0.5% 0.6% 0.6% s.d. of heart rate (bpm) Partial r 0.135 0.136 0.137 Change in R 2 0.8% 0.9% 0.9% Heart rate (bpm) Partial r 0.111 0.109 0.110 Change in R 2 1.3% 1.5% 1.5% Linear models were run with s.d., CV or VIM of awake systolic blood pressure as the dependent variable. Independent variables, introduced in a stepwise manner (P entry = 0.050; P removal = 0.10), included age, gender, body mass index, treatment for hypertension, treatment for diabetes, mean systolic blood pressure, awake heart rate, and variability (s.d.) of awake heart rate. BP, blood pressure; bpm, beats per minute; CV, coefficient of variation; Rx, medication; VIM, variability independent of the mean. Partial correlation coefficients and corresponding P values ( a ) are presented for the final step of each model, while the change in R 2 ( b ), analogous to the percent variation explained, is given for the corresponding step in the models. Dashes ( ) denote the respective variables did not enter the stepwise model. Awake systolic BP variability (95% Cl) 16 15 14 13 12 11 10 9 8 7 Awake heart rate decile, bpm s.d., mm Hg cv, % vim, % 6 50 60 70 80 90 100 Figure 2 Awake systolic blood pressure variability (and 95% CI) plotted vs. awake heart rate deciles (n = 4,693). BP, blood pressure; bpm, beats per minute; CV, coefficient of variation; VIM, variability independent of the mean. HR-nonresponsive cuff deflation. With lower HR, accuracy is compromised since chances of SBP underestimation and DBP overestimation are higher (Figure 1). Moreover, with lower HR precision may also be compromised as suggested by Zheng et al. 11 The theoretical result is increased BPV on repeated measurements. Higher BPV has been postulated to account for comparatively higher stroke rates among patients treated with β-blockers. However, it may be that increased variability is in part related to bradycardia induced by these agents. Underestimation of SBP associated with bradycardia and, in turn, undertreatment of hypertension, may hypothetically provide explanation for the observed poorer outcomes in patients receiving β-blockers. In the current study, we aimed to empirically determine the relationship between HR and variability of BP measurements. AMERICAN JOURNAL OF HYPERTENSION VOLUME 25 NUMBER 3 march 2012 315

original contributions Bradycardia and BP Variability Methods Data were extracted from records consecutively collected in our ABPM service database, from 1991 through 2009. Both treated and nontreated patients were included, except those <16 years old, pregnant women, and subjects with poorquality ABPM (<50 valid measurements). Patients were referred for standard clinical indications at the discretion of the referring physician (mainly by primary care practitioners, who have been shown to use ABPM for appropriate indications). 12 Twenty-four hour ABPM was executed as previously described. 13 The monitor was mounted on the nondominant arm between 8:00 and 10:00 am and removed 24 h later. Recordings were made every 20 min between 6:00 am and midnight and every 30 min between midnight and 6:00 am. A mercury sphygmomanometer was initially attached to the monitor via a Y-connector to verify agreement between the two modes of measurement (within a range of 5 mm Hg). Cuff size was selected according to measured arm circumference: 24-cm pediatric cuff, 24- to 32-cm standard adult cuff, and >32-cm large adult cuff. Sleep, including daytime naps (reported in ~30%), was logged in a diary. Daytime sleep was not included in the awake BP average. BPV was expressed using standard deviation (s.d.) and the coefficient of variation (CV). BPV was further assessed using the variability independent of mean (VIM). VIM is a measure of variability uncorrelated with mean BP, given by VIM = s.d./mean x, wherein the parameter x is estimated from the power curve of SBP s.d. plotted against mean SBP (described in ref. 14). In our cohort x = 1.022 for awake SBP; 1.001 for sleep SBP; 0.218 for awake DBP; and 0.336 for sleep DBP. All consecutive patients were studied for ambulatory BPV. Stepwise linear regression model covariables included age, gender, treatment status for hypertension and diabetes, body mass index (BMI), BP indexes, HR and HR variability (s.d.). The normality assumption was examined by inspecting histograms and normal P-P plots of the regression standardized residuals. In a subset of patients for whom the antihypertensive medication regimen was available, we analyzed the effect of medications on the relationship between HR and BPV using medication class-stratified linear regression models. Odds ratios for increased (i.e., above median) BPV with use of β-blockers were calculated using logistic regression; other medications (number) and HR were covariates in the adjusted models. In another subset of patients with repeated ABPM sessions, we evaluated determinants of the change in BPV between sessions. Determinants were detected by multivariable repeated-measures analysis of variance and further explored with stepwise linear regression with the change in BPV as the dependent variable. Data are expressed as mean ± s.d., unless otherwise specified. Two-sided nominal P < 0.05 were considered significant. PASW Statistics 17.0 (SPSS,. Chicago, IL) was used for statistical analysis. The need for informed consent was waived by the institutional ethics committee owing to the service-based nature of our study. Results Patients characteristics (n = 4,693) are shown in Table 1. HR was inversely associated with BPV (Figure 2), and independently explained 1.3% of the variation in s.d. of awake SBP between subjects, as determined by multivariable linear Table 3 Association between heart rate and awake systolic blood pressure variability (s.d., CV or VIM) among patients with available data on medications (n = 1,019) Drug class Status Measure of awake systolic blood pressure variability s.d. (mm Hg) CV (%) VIM (%) a β (95% CI) P value β (95% CI) P value β (95% CI) P value ACEi/ARB No (n = 661) 0.024 ( 0.048 to 0.000) 0.046 0.017 ( 0.035 to 0.000) 0.051 0.015 ( 0.031 to 0.000) 0.051 Yes (n = 358) 0.044 ( 0.081 to 0.007) b 0.02 0.030 ( 0.055 to 0.004) b 0.03 0.026 ( 0.049 to 0.003) b 0.03 β-blocker No (n = 676) 0.034 ( 0.057 to 0.011) <0.01 0.024 ( 0.041 to 0.007) <0.01 0.021 ( 0.036 to 0.006) <0.01 Yes (n = 343) 0.041 ( 0.083 to 0.002) 0.06 0.027 ( 0.057 to 0.003) 0.07 0.024 ( 0.051 to 0.002) 0.08 Ca-blocker No (n = 779) 0.039 ( 0.061 to 0.017) 0.001 0.027 ( 0.043 to 0.011) 0.001 0.024 ( 0.038 to 0.010) 0.001 Yes (n = 240) 0.028 ( 0.076 to 0.021) 0.27 0.019 ( 0.054 to 0.016) 0.29 0.017 ( 0.047 to 0.014) 0.29 Diuretic No (n = 741) 0.018 ( 0.040 to 0.003) 0.10 0.013 ( 0.029 to 0.003) 0.11 0.011 ( 0.025 to 0.003) 0.11 Yes (n = 278) 0.076 ( 0.122 to 0.030) b 0.001 0.053 ( 0.086 to 0.020) b <0.01 0.047 ( 0.076 to 0.018) b <0.01 α-blocker No (n = 953) 0.036 ( 0.057 to 0.016) 0.001 0.025 ( 0.040 to 0.010) 0.001 0.022 ( 0.035 to 0.009) 0.001 Yes (n = 66) 0.068 ( 0.156 to 0.020) 0.14 0.049 ( 0.110 to 0.012) 0.12 0.043 ( 0.097 to 0.011) 0.12 Regression slopes (β) and 95% confidence intervals (95% CI) were derived from linear regression models with awake systolic blood pressure variability (s.d., CV or VIM) as the dependent variables and awake heart rate as the independent variable. Covariates in these models included age, gender, diabetes treatment status, number of blood pressure medications, body mass index, awake systolic blood pressure, and variability of awake heart rate (s.d.). The full models in the undivided population with available data on medication are presented in Supplementary Table S2 online. ACEi, angiotensin-converting enzyme inhibitors; ARB, angiotensin receptor blockers; Ca-blocker, calcium channel blocker; CV, coefficient of variation; VIM, variability independent of the mean. a In these models, VIM was entered as % (similar to CV) to facilitate comparison between the respective regression slopes. b P value for the interaction of medication class with awake heart rate < 0.050. 316 march 2012 VOLUME 25 NUMBER 3 AMERICAN JOURNAL OF HYPERTENSION

Bradycardia and BP Variability original contributions Table 4 Odds ratios and P values for increased blood pressure variability associated with use of β-blockers Measure of BPV SBP DBP OR (95% CI) of having high BPV P value Adjusted OR (95% CI) a of having high BPV P value Awake s.d. 2.41 (1.84 3.14) <0.0001 0.94 (0.65 1.35) 0.72 Awake CV 2.40 (1.84 3.13) <0.0001 1.14 (0.80 1.62) 0.48 Awake VIM 2.31 (1.77 3.02) <0.0001 1.12 (0.78 1.60) 0.54 Sleep s.d. 1.87 (1.43 2.45) <0.0001 1.18 (0.83 1.68) 0.36 Sleep CV 1.64 (1.26 2.15) <0.0001 1.16 (0.82 1.64) 0.40 Sleep VIM 1.64 (1.26 2.15) <0.0001 1.16 (0.82 1.64) 0.40 Awake s.d. 1.22 (0.94 1.58) 0.14 1.13 (0.79 1.60) 0.51 Awake CV 1.55 (1.19 2.01) 0.001 0.89 (0.62 1.27) 0.51 Awake VIM 1.23 (0.95 1.60) 0.12 0.94 (0.66 1.34) 0.74 Sleep s.d. 1.26 (0.97 1.64) 0.08 1.42 (1.01 2.00) 0.050 Sleep CV 1.22 (0.94 1.58) 0.14 1.18 (0.84 1.66) 0.34 Sleep VIM 1.29 (1.00 1.68) 0.050 1.38 (0.98 1.94) 0.07 Odds ratios were computed using logistic regression models with above-median BPV as the dependent variable, β-blockers as a categorical dependent variable and ( a ) other medications (number) and heart rate (awake or asleep, as appropriate) as covariates in the adjusted models. (Note that for sleep SBP, CV and VIM yield identical values this is due to the fact that CV of sleep SBP is already independent of the mean, and thus almost identical to VIM). BPV, blood pressure variability; CI, confidence interval; CV, coefficient of variation; DBP, diastolic blood pressure; OR, odds ratio; SBP, systolic blood pressure; VIM, variability independent of the mean. regression. Mean SBP (12.7%), age (3.3%), gender (1.2%), s.d. of HR (0.8%), and treatment for hypertension (0.5%) also contributed to the variance in the model (Table 2 and Supplementary Table S1 online). In a subset of 1,019 patients with available data on medication, we found that HR is inversely related to systolic BPV, in subjects treated or untreated with any of the five major antihypertensive drug classes (Table 3). This association was more pronounced in patients treated with diuretics or renin angiotensin system antagonists (P < 0.050 for the interaction between both medication classes and HR in the linear model, not shown). We next examined the specific association of β-blockers with BPV using logistic regression. β-blockade associated with increased odds of high (above median) BPV (s.d., CV, and VIM). These tendencies, however, were as a rule abolished by adjusting for HR (Table 4). According to a fully adjusted model (Supplementary Table S2 online), the slope of the association of awake systolic BPV (s.d.) with HR was 0.05 mm Hg (95% CI 0.06 to 0.03 mm Hg) per +1 beats per minute (bpm), or 0.45 mm Hg (95% CI 0.57 to 0.33 mm Hg) per +10 bpm. Lastly, we analyzed another subset of patients who had been monitored twice (n = 635) to investigate effectors of changes in BPV. With multivariate repeated-measures analysis of variance, we determined that Δ (awake HR) was a significant within-subject determinant of awake systolic BP variability (s.d., CV and VIM all P values <0.050), adjusting for age, Table 5 Two-step multivariable linear regression models evaluating contributors to within-subject change in awake systolic blood pressure variability (repeated ABPM sessions) Measure of awake systolic Model covariable Attribute BP variability s.d. (mm Hg) CV (%) VIM Step 1 Age (years) Partial r 0.053 0.062 0.062 P value 0.19 0.12 0.12 Gender Partial r 0.054 0.061 0.061 (0 = m, 1 = f) P value 0.18 0.13 0.13 Δ BMI (kg/m 2 ) Partial r 0.015 0.010 0.010 P value 0.72 0.81 0.81 Intersession Partial r 0.094 0.100 0.100 interval (months) P value 0.02 0.01 0.01 Δ Awake SBP Partial r 0.215 0.132 0.140 (mm Hg) P value <0.0001 0.001 0.001 Δ Heart rate Partial r 0.081 0.085 0.085 variability (bpm) P value 0.04 0.04 0.04 Change in R 2 6% 4.6% 4.8% Step 2 Δ Heart rate (bpm) Partial r 0.080 0.082 0.082 P value 0.046 0.04 0.04 Change in R 2 0.6% 0.6% 0.6% Linear models were run with s.d., CV or VIM of awake systolic blood pressure as the dependent variable. Independent variables, introduced in a stepwise manner (P entry = 0.050; P removal = 0.10), included age, gender, body mass index, treatment for hypertension, treatment for diabetes, mean systolic blood pressure, awake heart rate, and variability (s.d.) of awake heart rate. Partial correlation coefficients and corresponding P values are presented for the final step of each model, while the change in R 2, analogous to the percent variation explained, is given for the corresponding step in the models. ABPM, ambulatory blood pressure monitoring; BMI, body mass index; BP, blood pressure; bpm, beats per minute; CV, coefficient of variation; SBP, systolic blood pressure VIM, variability independent of the mean. gender, Δ (BMI), Δ (awake SBP), Δ (s.d. of awake HR), and intersession time interval (not shown). We thus quantified the association with linear regression, and found that the change in HR has a mild (0.6%) contribution to the change in awake systolic BP variability, which was significant (P < 0.050 for s.d., CV, and VIM) with partial r values between 0.080 and 0.082 (compare with r = 0.010 with the change in BMI and r = 0.100 with the intersession interval), Table 5. Discussion We found that lower HR associates with increased BPV during ABPM, independent of the use of specific antihypertensive medications. Correction for HR eliminated the increased systolic BPV found among patients treated with β-blockers. We therefore suggest that the association of greater BPV with worse outcome in users of β-blockers may be in part confounded by bradycardia. These findings were established in a AMERICAN JOURNAL OF HYPERTENSION VOLUME 25 NUMBER 3 march 2012 317

original contributions Bradycardia and BP Variability fairly large number of patients, are independent of drug treatment (within the limitations of our available data on 1,019 patients), and are also evident in analysis of repeated measurements. Thus, they represent a robust observation. It has been shown for visit-to-visit BPV that about 50% of group s.d. of SBP at any follow-up is a result of within-individual visit-to-visit variability. Thus, studies reporting increased interindividual variability of BP in association with lower HR may indirectly support our argument. Such studies include observations on elderly population (PAMELA, n = 248), 15 in whom the group variability (s.d.) of night-time BP was higher than that of daytime BP; 13.3/8.1 mm Hg vs. 12.3/7.6 mm Hg, respectively (for CV the calculated values would be 11.7/12.6% vs. 9.6/9.9%), while night-time HR and HR variability were lower than daytime; 64.2 ± 7.6 vs. 79.0 ± 8.4, respectively. Likewise, baseline data from a European hypertension drug study (n = 1,663) showed reduction in night-time HR and HR variability accompanied by increased BPV. 16 In a previous report from our laboratory, the CV of systolic BP in untreated patients rose from 9.2% to 13% and 14% during the siesta and night sleep (a rise of 41% and 52% respectively, accompanied by a sleep-related drop in HR). 17 Regardless of variability, underestimation of true SBP and subsequent undertreatment may be a true contributor to worse outcome in hypertensive patients receiving β-blockers. Our findings add empirical quantification to the thoughtful, though not widely appreciated, theoretical model of Yong and Geddes. 9 Our work also adds another dimension to the solid observations and monumental analyses of Rothwell and coworkers. 2,4 6 Despite the fact that BP was properly measured with well-validated instruments in these studies, we suggest that inherent systematic measurement error of currently used BP monitors may have been present. For perspective, our empirical finding of +1.5 mm Hg increase in s.d. of systolic BP associated with 30 bpm lower HR, may explain a large portion of the BPV difference observed in clinical studies note for example the 2.4 mm Hg difference in intervisit s.d. of SBP in atenolol (13.4 mm Hg) vs. amlodipine (11.0 mm Hg) treated Anglo-Scandinavian Cardiac Outcomes Trial Blood Pressure Lowering Arm (ASCOT-BPLA) patients. 5 Our analyses have limitations, among them the referred nature of our patients and lack of full data on medication, smoking, and complicating medical conditions that may affect HR. Yet these limitations would have tended to mask the evident effect we found, and thus it is in all likelihood an underestimation of the true effect of bradycardia on BPV. HR is a significant determinant of BPV, and the shift from manual mercury sphygmomanometry (whereby deflation rate might have been intuitively adjusted by the experienced observer in face of bradycardia), to automated instruments with constant deflation rate may augment this link. Reconsideration of prior clinical studies that evaluated nondihydropyridine calcium channel blocker, β-blockers, and central sympathomimetics, which are known to be negative chronotropics, may be interesting in context of the relationship between HR and BPV. If supported by future investigations, these observations may substantiate reviving BP measurement guidelines from the 1950 s, which advocated HR-responsive cuff deflation, 8 and algorithms modification for automated BP monitors to include deflation rate adjustment by HR. Supplementary material is linked to the online version of the paper at http://www.nature.com/ajh Acknowledgments: I.Z.B. was supported by the Stavros Niarchos Foundation through a grant to the Rockefeller University. Disclosure: The authors declared no conflict of interest. 1. Carlberg B, Lindholm LH. Stroke and blood-pressure variation: new permutations on an old theme. Lancet 2010; 375:867 869. 2. Rothwell PM. Limitations of the usual blood-pressure hypothesis and importance of variability, instability, and episodic hypertension. Lancet 2010; 375:938 948. 3. Muntner P, Shimbo D, Tonelli M, Reynolds K, Arnett DK, Oparil S. The relationship between visit-to-visit variability in systolic blood pressure and all-cause mortality in the general population: findings from NHANES III, 1988 to 1994. Hypertension 2011; 57:160 166. 4. Rothwell PM, Howard SC, Dolan E, O Brien E, Dobson JE, Dahlöf B, Sever PS, Poulter NR. Prognostic significance of visit-to-visit variability, maximum systolic blood pressure, and episodic hypertension. Lancet 2010; 375: 895 905. 5. Rothwell PM, Howard SC, Dolan E, O Brien E, Dobson JE, Dahlöf B, Poulter NR, Sever PS. Effects of beta blockers and calcium-channel blockers on within-individual variability in blood pressure and risk of stroke. Lancet Neurol 2010; 9:469 480. 6. Webb AJ, Fischer U, Mehta Z, Rothwell PM. Effects of antihypertensive-drug class on interindividual variation in blood pressure and risk of stroke: a systematic review and meta-analysis. Lancet 2010; 375:906 915. 7. Zhang Y, Agnoletti D, Safar ME, Blacher J. Effect of antihypertensive agents on blood pressure variability: the Natrilix SR versus candesartan and amlodipine in the reduction of systolic blood pressure in hypertensive patients (X-CELLENT) study. Hypertension 2011; 58:155 160. 8. Bordley J 3rd, Connor CA, Hamilton WF, Kerr WJ, Wiggers CJ. Recommendations for human blood pressure determinations by sphygmomanometers. Circulation 1951; 4:503 509. 9. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL Jr, Jones DW, Materson BJ, Oparil S, Wright JT Jr, Roccella EJ. Seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure. Hypertension 2003; 42:1206 1252. 10. Yong PG, Geddes LA. The effect of cuff pressure deflation rate on accuracy in indirect measurement of blood pressure with the auscultatory method. J Clin Monit 1987; 3:155 159. 11. Zheng D, Amoore JN, Mieke S, Murray A. How important is the recommended slow cuff pressure deflation rate for blood pressure measurement? Ann Biomed Eng 2011; 39:2584 2591. 12. Grin JM, McCabe EJ, White WB. Management of hypertension after ambulatory blood pressure monitoring. Ann Intern Med 1993; 118:833 837. 13. Cahan A, Ben-Dov IZ, Mekler J, Bursztyn M. The role of blood pressure variability in misdiagnosed clinic hypertension. Hypertens Res 2011; 34:187 192. 14. Dolan E, O Brien E. Blood pressure variability: clarity for clinical practice. Hypertension 2010; 56:179 181. 15. Sega R, Cesana G, Milesi C, Grassi G, Zanchetti A, Mancia G. Ambulatory and home blood pressure normality in the elderly: data from the PAMELA population. Hypertension 1997; 30:1 6. 16. Mancia G, Parati G, Hennig M, Flatau B, Omboni S, Glavina F, Costa B, Scherz R, Bond G, Zanchetti A. Relation between blood pressure variability and carotid artery damage in hypertension: baseline data from the European Lacidipine Study on Atherosclerosis (ELSA). J Hypertens 2001; 19:1981 1989. 17. Bursztyn M, Mekler J, Ben-Ishay D. The siesta and ambulatory blood pressure: is waking up the same in the morning and afternoon? J Hum Hypertens 1996; 10:287 292. 318 march 2012 VOLUME 25 NUMBER 3 AMERICAN JOURNAL OF HYPERTENSION