Physiological Correlates of Beat-to-Beat, Ambulatory, and Day-to-Day Home Blood Pressure Variability After Transient Ischemic Attack or Minor Stroke
Background and Purpose—Visit-to-visit and day-to-day variability in systolic blood pressure (SBP) are associated with an increased risk of stroke, more strongly than variability on 24-hour ambulatory BP monitoring, but underlying physiological mechanisms are unclear. We related potentially relevant physiological characteristics to beat-to-beat, ambulatory, and day-to-day BP variability to identify underlying mechanisms and potential therapeutic targets.
Methods—BP variability (coefficient of variation [CV]) on 1-month home BP monitoring (3 sitting readings, 3× daily), on 24-hour ambulatory BP monitoring, and on 5-minute beat-to-beat monitoring was related to BP reactivity (to mental arithmetic), arterial aging (aortic stiffness: carotid-femoral pulse wave velocity; aortic pulsatility), heart rate variability (CV of normal-to-normal R-R interval), and orthostatic responses.
Results—In 223 patients within 6 weeks of a transient ischemic attack or minor stroke, beat-to-beat and home SBP-CVs were associated with response to arithmetic (beat-to-beat odds ratio per SD=1.64; P<0.0001 and home BP monitoring, 1.41; P=0.025), aortic stiffness (1.84; P<0.0001 and 1.31; P=0.04), aortic pulsatility (1.98; P<0.0001 and 1.61; P<0.0001), and heart rate variability–CV of normal-to-normal R-R interval (1.34; P=0.03 and 1.35; P=0.03), independently of age, sex, and aortic BP. Orthostatic BP changes were associated only with SBP-CV on home BP monitoring (0.62; P=0.002). In contrast, no physiological measures were associated with within-day BP variability on awake ambulatory BP monitoring except response to mental arithmetic (1.40; P=0.01).
Conclusions—Beat-to-beat and day-to-day SBP variability, but not variability on ambulatory BP monitoring, had similar physiological correlates, suggesting common underlying mechanisms and identifying potentially treatable targets that may be responsible for the relationship between SBP variability and stroke risk.
Patients with episodic hypertension in clinic after a previous transient ischemic attack or stroke have a high risk of recurrent stroke,1,2 residual visit-to-visit variability in blood pressure (BP) on antihypertensive treatment has a poor prognosis despite good control of mean BP,1,2 and benefits of some antihypertensive drugs in the prevention of stroke may partly result from reduced variability in systolic BP (SBP).3,4 Home day-to-day BP variability (home BP monitoring [HBPM])5 is also associated with an increased stroke risk,6 particularly for BP variability in the morning.7 In contrast, short-term BP variability on awake ambulatory BP monitoring (ABPM) is only weakly predictive.1,8 However, the physiological mechanisms underlying the relationship between BP variability and stroke risk are unclear. Baroreceptor gain is reduced after stroke,9 associated with an increase in all-cause mortality,10,11 possibly because of changes in cerebral hemodynamics,12 but impaired orthostatic responses also increase cardiovascular risk.13 Reduced vascular compliance, manifest as increased arterial stiffness, is associated with both visit-to-visit BP variability14 and cardiovascular risk,15 whereas BP reactivity in early life increases late life hypertension16 and cardiovascular risk.17–19
To understand apparent differences in predictive value between forms of BP variability and to identify treatable mechanisms, we aimed to identify important physiological correlates. However, the few such studies20,21 are mostly in uncomplicated hypertension, assessing only a single facet of cardiovascular physiology. Therefore, in a cohort of patients with recent transient ischemic attack or minor stroke, we aimed to assess the validity of short-term beat-to-beat BP variability as a marker of ambulatory and day-to-day BP variability and measure whether 5-minute beat-to-beat, 1-month home, and 24-hour ambulatory BP variability are dependent on endogenous cardiovascular rhythms, cardiovascular responses to external and internal stimuli, or adaptive control mechanisms.
Consecutive patients were recruited between April 2008 and December 2012 from the Oxford Vascular Study (OXVASC)22 transient ischemic attack and minor stroke clinic.23 The OXVASC population consists of 92 728 individuals registered with 100 primary-care physicians in Oxfordshire, United Kingdom.22 All consenting patients underwent a standardized medical history and examination, ECG, blood tests, and a stroke protocol MRI brain and contrast-enhanced MR angiography (or computed tomography-brain and carotid Doppler ultrasound or computed tomography angiogram), an echocardiogram, and 5-day cardiac monitoring. All patients were reviewed by a study physician, the diagnosis verified by the senior neurologist, etiology determined by a panel of stroke neurologists and were followed up face to face at 1, 3, 6, and 12 months. Participants were excluded if they were <18 years, cognitively impaired (mini-mental state examination <23), pregnant, had cancer, autonomic failure, a recent myocardial infarction, unstable angina, heart failure (New York Heart Association, 3–4 or ejection fraction, <40%), or untreated bilateral carotid stenosis (>70%) and were excluded from this analysis if they had atrial fibrillation during testing.24 The study was approved by the Oxfordshire Research Ethics Committee.
Two-sitting clinic BPs, 5 minutes apart, were measured at ascertainment and 1 month in the nondominant arm, by trained personnel after 5 minutes of rest. From ascertainment, all patients had HBPM training and were asked to perform 3 home readings for 10 minutes, 3× daily (after waking, midmorning, and evening) with a Bluetooth-enabled, regularly calibrated, telemetric IEM Stabil-o-Graph, or A&D UA-767 BT. Patients were instructed to relax in a chair for 5 minutes before measuring BP in the nondominant arm or the higher-reading arm when the mean SBP differed by >20 mm Hg. Anonymized measures were securely transmitted via Bluetooth radio and a mobile phone to a password-protected website (t+Medical, Abingdon, United Kingdom). The day before the 1-month follow-up, ABPM was performed with an A&D TM-2430 monitor in the nondominant arm. BP was measured every 30 minutes during the day and 60 minutes at night. BP was treated according to guidelines,25 most frequently with perindopril, indapamide, or amlodipine. Hypertension was defined as a known diagnosis, use of BP-lowering medications, or a mean premorbid BP >140/90.
Patients were tested at the ascertainment or 1-month clinic in a quiet, dimly-lit, temperature-controlled room (21–23°C). Continuous 3-lead ECG and finger arterial BP were acquired at 200 Hz (Finometer MIDI) via a Powerlab 8/35 (ADInstruments). Automated calibration was performed until the recording was stable, but turned off during each test, and readings calibrated offline to the mean of 2 supine brachial readings. Orthostatic BP changes were determined after 5 minutes of rest: 2 minutes standing (sit–stand), 2 minutes supine, and then 2 minutes standing (lie–stand). Supine beat-to-beat BP and heart rate (HR) were monitored for 10 minutes after 20 minutes of rest. BP reactivity was assessed by a 2-minute serial-subtraction test (arithmetic) and by 1-minute hand immersion in iced water (0–4°C; cold pressor test). Aortic BP, pulsatility, and augmentation index (normalized to 75 bpm) were estimated by radial applanation tonometry (Sphygmocor, AtCor Medical, Sydney) and aortic pulse wave velocity by carotid-femoral applanation tonometry, averaging 2 acceptable measures (SD<2).26
HBPM variability was derived from 7 days after ascertainment until readings were performed on <3 days/wk or 90 days had elapsed, averaging the last 2 readings of each cluster. Awake and asleep BP variability on ABPM were derived after automated and manual exclusion of artifacts.27 BP variability was derived as the coefficient of variation (CV=SD/mean), the residual CV (rCV) about a 9-day moving average on HBPM, or by variation independent of the mean.1 Percentage BP responses to stressors were estimated after 60 seconds (orthostatic), as the maximum increase over baseline (arithmetic) or as the maximum increase above minimum during cold pressor test, to adjust for vasoconstriction-induced artifactual BP fall.
Beat-to-beat BP and HR variability (HRV) were derived as CV of normal-to-normal R-R intervals (SBP-CV and HRV-CVNN) from 5 minutes of supine monitoring, after linear interpolation of ectopic beats, detected automatically, and by visual review. Baroreceptor gain was calculated in low (0.04–0.15 Hz) and high (0.15–0.4 Hz) frequency ranges (Welch method), after 2-Hz interpolation of the time-series with a cubic spline, by the transfer function between pulse interval and SBP where mean-squared coherence >0.4.5,6
Relationships among beat-to-beat, home, awake, and asleep BP variability were determined by linear regression; by receiver operating characteristic curve analysis for the top quartile of BP variability; and by κ statistics for quartiles between measurement methods. Associations with increasing quartiles of BP variability were determined by ordinal regression (per SD of the population) and by general linear models, with and without adjustment for age, sex, and mean aortic BP. Stepwise general linear models were used to analyze 5 imputed datasets, accounting for missing data in each physiological predictor. Principal component analysis (PCA; varimax rotation) extracted components explaining the greatest proportion of variance in physiological parameters and BP variability.
A total of 223 (90%) of 248 eligible patients had adequate beat-to-beat monitoring: 15 (6%) had atrial fibrillation during testing, 7 (2.8%) had inadequate peripheral circulation for beat-to-beat monitoring, and 3 patients (2%) received excluded diagnoses (multiple system atrophy, metastatic carcinoma). A total of 206 (92%) had ABPM and 194 (87%) had HBPM (2.9 readings per cluster for median 29 days). Aortic pulse wave velocity could not be performed in 6 (2%) patients, 8 (3%) patients refused arithmetic, 27 (12%) patients withdrew their hand during cold pressor test, and 23 (10%) and 37 (17%) patients had musculoskeletal problems preventing sit–stand or lie–stand maneuvers (Table 1).
Beat-to-beat SBP-CV was associated with home SBP-rCV and variation independent of the mean (Tables I–III in the online-only Data Supplement), on all readings, at each time of day and within-day (Table IV in the online-only Data Supplement), but not with awake ABPM-CV. Agreement was weak for classification of beat-to-beat and home SBP variability into quartiles (κ=0.13; 0.04–0.22), or the top quartile (κ=0.26; 0.12–0.41) and beat-to-beat SBP-CV only modestly discriminated patients in the top quartile of HBPM SBP-rCV (area under the curve=0.68; 0.59–0.77; P<0.0001), with a 66% sensitivity and 60% specificity for beat-to-beat SBP-CV of 5%.
Beat-to-beat SBP-CV and diastolic BP (DBP)-CV were associated with arithmetic and cold pressor test (Table 2), home SBP-rCV and awake SBP-CV were only associated with arithmetic (Table 2), whereas home DBP-rCV and awake or asleep ABPM DBP-CV were not associated with either (Tables I–VIII in the online-only Data Supplement). Response to arithmetic was associated more strongly with day-to-day home SBP variability in the middle of the morning than after waking, in the evening, or within-day (Table VI in the online-only Data Supplement). Orthostatic BP responses were associated with home SBP-rCV but not beat-to-beat SBP-CV (Table 2).
Aortic-pulse pressure was the measure most strongly associated with beat-to-beat and home SBP-CV (Table 2), because of high SBP and low DBP. Aortic pulse wave velocity was associated with beat-to-beat SBP-CV and home SBP-rCV, before adjustment (Table 2). Associations were also present for beat-to-beat DBP-CV but not for either home DBP-rCV or ABPM variability. HRV-CVNN was correlated positively with beat-to-beat and home SBP variability (Tables 2 and 3), particularly in the low frequency range (P=0.013), but baroreceptor gain was not associated significantly with BP variability, as it adjusts HR variability for BP variability. Associations were unchanged for variation independent of the mean (Table II in the online-only Data Supplement). Differences in the strength of association of BP variability with physiological parameters between monitoring methods were not confounded by differences in strength of association with mean SBP, which was similar for different methods (Table IX in the online-only Data Supplement).
In combined models, responses to arithmetic, aortic pulsatility, age, and HRV-CVNN were independent predictors of both beat-to-beat and home SBP-rCV (Table 3), whereas orthostatic responses were only associated with home SBP-rCV. Similarly, the first principal component analysis component, explaining 23% of total variance, was related most strongly to beat-to-beat and home SBP variability as well as age, aortic pulsatility, response to mental arithmetic, and HRV-CVNN. A total of 16% of total variance was explained by the second component, related to mental arithmetic, home, and awake SBP variability, but not beat-to-beat SBP variability, with no other component being related to measures of SBP variability. Mental arithmetic and HRV-CVNN were not associated (Tables X and XI in the online-only Data Supplement), but aortic-pulse pressure was associated with both mental arithmetic and HRV-CVNN. Similarly, age and female sex were consistent predictors of aortic-pulse pressure and HRV-CVNN, but they were unrelated to mental arithmetic.
Beat-to-beat SBP-CV was correlated with HBPM SBP-CV, although too weakly to estimate HBPM SBP-CV from beat-to-beat SBP-CV, but not with variability on awake or asleep ABPM. Day-to-day and beat-to-beat SBP-CV were associated with similar physiological indices, including reactivity to arithmetic, aortic stiffness, aortic pulsatility, and HRV, whereas home SBP-rCV was also associated with response to orthostatic challenges. However, neither awake nor asleep SBP variability on ABPM was associated with any physiological parameter except response to arithmetic, possibly explaining the limited prognostic value of BP variability on ABPM.1,8
No previous study has addressed the multi-factorial physiological basis of day-to-day BP variability on home monitoring and few studies have investigated beat-to-beat, visit-to-visit, or ambulatory BP variability, demonstrating only isolated associations with arterial stiffness14,15,21,22 and autonomic dysfunction.28,29 Ours is the first study demonstrating the independence of associations between these mechanisms and different forms of BP variability and demonstrating the similarity between correlates of beat-to-beat and home SBP variability, suggesting common underlying mechanisms. Furthermore, although beat-to-beat BP is inadequate to estimate home BP variability, it may itself be prognostically significant.
The paucity of physiological correlates of BP variability on ABPM may partly explain its weak prognostic value8 and demonstrates that ABPM is inadequate for investigation of the prognostic significance of BP variability in this population. This may reflect different physiological mechanisms underlying BP variability on ABPM, but it is likely that ABPM is more prone to artifacts, despite intensive data cleaning, particularly in aging patients with cerebrovascular disease.8 This noise may result from a greater alerting response,30 physical activity during monitoring, or other unidentified mechanisms. Alternatively, the physiological measures in this article may only reflect supine and sitting BP control, as seen on beat-to-beat and HBPM, respectively, but may be less strongly associated with standing BP and changes in BP with posture, as reflected by ABPM. Furthermore, nocturnal mean SBP and day:night difference in BP are independently prognostic31 and could be associated with the physiological measures assessed in this article. Therefore, further research is warranted to address the complex relationships between these BP indices on ABPM and the physiological measures in this article.
Endogenous autonomic rhythms were associated with BP variability at low frequencies, dependent on sympathetic and parasympathetic function, and at parasympathetically determined high frequencies,32 indicating normal BP variability associated with an intact autonomic nervous system.33 Therefore, the association between baroreceptor gain and mortality after stroke6 may result from cardiac events, rather than BP variability–related stroke. Alternatively, the correlation between HBPM SBP-rCV and orthostatic control may indicate that failure of postural BP control could predispose to stroke risk and therefore partly explain the association between visit-to-visit BP variability and stroke. BP variability was associated with other adaptive mechanisms, including aortic stiffness and pulsatility, potentially increasing stroke risk through either reduced vascular compliance and reduced damping of acute BP changes or as a marker of age-related arteriopathy.26 In addition, BP variability was correlated with cardiovascular reactivity, consistent with studies demonstrating an association between early-life BP reactivity and the development of hypertension,16,17 potentially explaining the association between environmental stressors and stroke risk.18,19
This study has limitations. First, the strength of associations seems relatively weak, with r2 values of 0.1 to 0.2. However, the intraclass correlation coefficients of BP variability are only 0.18 to 0.32.5 As the strength of a linear association cannot significantly exceed the reproducibility of its parameters, the maximum attainable r2 in this study would not exceed 0.3, and therefore these associations are proportionately strong. Furthermore, r2 values of only 0.12 to 0.16 are often found with causative physiological correlations, such as those between arterial stiffness and age or arterial pulsatility.34 Second, this study’s generalizability may be limited as increased BP variability and associated physiological changes may result from pre-existing cerebrovascular disease.1,2 Additional studies are therefore required in primary prevention of early and established hypertension. Third, most patients were treated with antihypertensives. However, medication adherence did not explain the associations with variability in home SBP (Table III in the online-only Data Supplement) and could not explain the associations with beat-to-beat SBP variability. Finally, HRV and baroreceptor sensitivity were measured supine. As such, associations with beat-to-beat and home BP variability are more likely to reflect underlying rhythmic variations in autonomic tone, rather than dynamic autonomic responses to perturbations in BP.
This study identified similar physiological correlates for beat-to-beat and HBPM BP variability, suggesting common underlying mechanisms, including autonomic nervous system oscillations, cardiovascular reactivity, and impaired adaptive mechanisms. Therefore, in patients with abnormalities in these mechanisms, measurement of BP variability on HBPM and variability-directed treatment3,4 may be warranted. However, further research is required to determine the independent prognostic significance of these physiological correlates and beat-to-beat BP variability, whether they are present in patients without cerebrovascular disease, are they affected by treatment with calcium channel blockers, and how they relate to the occurrence of cardiovascular events associated with increased BP variability.
We acknowledge the invaluable support from Paul Leeson and Barbara Casadei in establishing this study and the facilities provided by the Cardiovascular Clinical Research Facility.
Sources of Funding
Oxford Vascular Study has been funded by the Wellcome Trust, Wolfson Foundation, UK Stroke Association, British Heart Foundation, Dunhill Medical Trust, National Institute of Health Research (NIHR), Medical Research Council, and the NIHR Oxford Biomedical Research Centre. A.J.S. Webb is in receipt of a Medical Research Council Fellowship.
We also acknowledge the use of the facilities of the Acute Vascular Imaging Centre, Oxford.
Guest Editor for this article was Tatjana Rundek, MD, PhD.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.113.003321/-/DC1.
- Received August 26, 2013.
- Revision received November 19, 2013.
- Accepted November 20, 2013.
- © 2014 American Heart Association, Inc.
- Cuffe RL,
- Howard SC,
- Algra A,
- Warlow CP,
- Rothwell PM
- Hansen TW,
- Thijs L,
- Li Y,
- Boggia J,
- Kikuya M,
- Björklund-Bodegård K,
- et al
- Robinson TG,
- James M,
- Youde J,
- Panerai R,
- Potter J
- Robinson TG,
- Dawson SL,
- Eames PJ,
- Panerai RB,
- Potter JF
- Sykora M,
- Diedler J,
- Rupp A,
- Turcani P,
- Steiner T
- Davis BR,
- Langford HG,
- Blaufox MD,
- Curb JD,
- Polk BF,
- Shulman NB
- 24.↵National Institute for Health and Clinical Excellence. The clinical management of primary hypertension in adults: clinical guideline 127. London: NICE; 2011.
- Webb AJS,
- Rothwell PM
- Webb AJ,
- Simoni M,
- Mazzucco S,
- Kuker W,
- Schulz U,
- Rothwell PM
- James MA,
- Potter JF
- Kabutoya T,
- Hoshide S,
- Ishikawa J,
- Eguchi K,
- Shimada K,
- Kario K
- Catalano M,
- Scandale G,
- Cinquini M,
- Minola M,
- Dimitrov G,
- et al