Behavioral Neurocardiac Training in Hypertension
A Randomized, Controlled Trial
It is not established whether behavioral interventions add benefit to pharmacological therapy for hypertension. We hypothesized that behavioral neurocardiac training (BNT) with heart rate variability biofeedback would reduce blood pressure further by modifying vagal heart rate modulation during reactivity and recovery from standardized cognitive tasks (“mental stress”). This randomized, controlled trial enrolled 65 patients with uncomplicated hypertension to BNT or active control (autogenic relaxation), with six 1-hour sessions over 2 months with home practice. Outcomes were analyzed with linear mixed models that adjusted for antihypertensive drugs. BNT reduced daytime and 24-hour systolic blood pressures (−2.4±0.9 mm Hg, P=0.009, and −2.1±0.9 mm Hg, P=0.03, respectively) and pulse pressures (−1.7±0.6 mm Hg, P=0.004, and −1.4±0.6 mm Hg, P=0.02, respectively). No effect was observed for controls (P>0.10 for all indices). BNT also increased RR-high-frequency power (0.15 to 0.40 Hz; P=0.01) and RR interval (P<0.001) during cognitive tasks. Among controls, high-frequency power was unchanged (P=0.29), and RR interval decreased (P=0.03). Neither intervention altered spontaneous baroreflex sensitivity (P>0.10). In contrast to relaxation therapy, BNT with heart rate variability biofeedback modestly lowers ambulatory blood pressure during wakefulness, and it augments tonic vagal heart rate modulation. It is unknown whether efficacy of this treatment can be improved with biofeedback of baroreflex gain. BNT, alone or as an adjunct to drug therapy, may represent a promising new intervention for hypertension.
The hypothesis that a behavioral intervention augments the hypotensive effect of pharmacological treatment of hypertension was tested in a randomized, controlled trial first by Patel and North.1 Training in yoga-induced relaxation with electrodermal biofeedback significantly reduced systolic and diastolic blood pressures (BPs; SBP and DBP), whereas no change was observed with self-guided relaxation as the control intervention. The therapeutic model for this approach can be traced to early research by Smirk, who recognized that BP could be lowered to a basal level by an emotionally calm or “desensitized” state and that emotional and environmental factors contributed to the “supplemental” elevation that was observed during casual BP measurements.2 It was subsequently asserted that repeated exposure to a “hypometabolic” state, popularly known as the relaxation response, could lower BP by reducing environmentally driven sympathetic overactivity.1,3 Commentary at the time concerning a behavioral treatment for hypertension focused on 2 issues: the need for differentiation between the depressor effect of biofeedback-assisted relaxation and any placebo response4 and whether active training with biofeedback augmented or interfered with the passive mental attitude that was considered necessary for therapeutic relaxation.3
Concerning hypertension, clinical trial evidence since Patel and North1 has been equivocal with regard to an association between BP reduction and relaxation therapy or stress management. Meta-analysis indicates that relaxation training significantly reduces SBP and DBP in comparison with no-treatment control conditions.5 However, in comparison with an active control (AC) or behavioral placebo, only biofeedback training combined with behavioral relaxation reduced SBP and DBP6 or SBP alone.5 Importantly, the 2 above-noted issues central to the behavioral research agenda were not addressed. Consequently, it is not established whether a behavioral intervention will exert an additional hypotensive effect among subjects receiving antihypertensive medications, in comparison with an AC or behavioral placebo. In addition, it is unknown whether passive relaxation is critical to an antihypertensive behavioral strategy that aims to augment autonomic regulation of BP by countering emotional-environmental challenges.
Our primary objective was to evaluate whether, in comparison with passive relaxation training which served as an AC, behavioral neurocardiac training (BNT) with heart rate variability (HRV) biofeedback7 reduces significantly ambulatory SBP, DBP, or pulse pressure (PP) among subjects with hypertension after adjusting for medications. Our secondary objective was to assess whether BNT or AC improved cardiovagal contributions to the dynamic regulation of BP, as assessed by tonic vagal-heart rate (HR) modulation and spontaneous baroreflex sensitivity (BRS) during cognitive reactivity-recovery testing.
We used a randomized, placebo-controlled, 2-parallel group design to assess BNT versus AC before intervention (within 2 weeks of initiation) and after intervention (1 week after the 8-week protocol, which allowed for completion of home practice). Randomization was conducted by a computer algorithm that stratified for sex, age (35 to 49 versus 50 to 64 years), and hypertension severity (SBP: 140 to 159 mm Hg or DBP: 90 to 99 mm Hg versus SBP: 160 to 180 mm Hg or DBP: 100 to 110 mm Hg). The behavioral nature of the interventions permitted only blinding of subjects, but the treatment randomization code remained hidden during processing of BP and ECG data.
AC provided training in autogenic relaxation, which is not associated with BP reduction.5 Therefore, it served as a behavioral placebo with regard to primary outcomes. Approval was obtained from the research ethics board of the University Health Network, and all of the subjects provided informed written consent.
Subjects were 35 to 64 years of age and diagnosed with stage 1 or 2 hypertension (SBP: 140 to 180 mm Hg; DBP: 90 to 110 mm Hg).8 Among subjects not prescribed medication, hypertension status was confirmed by ambulatory BP monitoring: daytime SBP/DBP ≥135/85 mm Hg or 24-hour SBP/DBP ≥130/80 mm Hg.8 Subjects prescribed antihypertensive drugs were required to have a prescription unchanged for ≥4 months before enrollment. Exclusion criteria were as follows: diagnosis of cardiovascular disease, clinically significant arrhythmia, sleep apnea, major psychiatric illness (eg, psychosis), alcohol/drug dependence in the previous year, or an inability to comprehend English or French.
Assessments before and after interventions were conducted on 2 successive days, between 8:00 am and 12:00 pm. On day 1 before interventions, anthropometric measures (age, sex, height, weight, and body mass index [BMI; in kilograms per meter squared]) and medical and medication histories were obtained. Subjects were then fitted with an ambulatory BP monitor (SpaceLabs model 90207-30, SpaceLabs Medical Inc) with its inflatable cuff fitted to the nondominant arm. BP was measured every 15 minutes between 8:00 am and 10:00 pm and every 30 minutes between 10:00 pm and 8:00 am. Subjects were instructed to engage in typical daily activities, with the exception of showering or strenuous exercise.
On day 2, subjects were instructed to refrain from caffeine and smoking for ≥12 hours before presenting to our laboratory, where they were seated in a semirecumbent position. RR interval was recorded continuously from lead II of the ECG. A finger cuff was placed on the third digit of the left hand to assess continuous variations in systolic arterial pressure (Finometer, Finapres Medical Systems). ECG and SBP data were digitized at 500 Hz and later resampled at 1000 Hz for offline analysis. Spontaneous BRS was estimated for each 3-minute interval of the assessment protocol by the sequence method, as described previously.9 We used customized software (LabView 7.1, National Instruments) to identify sequences of ≥3 cardiac cycles where SBP increased or decreased by ≥1 mm Hg and RR interval increased or decreased concordantly by ≥4 ms within 2 cardiac cycles after the onset of SBP changes (lag=0, 1, or 2). BRS was quantified as the slope of the linear regression line relating RR to SBP (in milliseconds per millimeter of mercury).
Power spectral analysis of the RR interval was performed with fast Fourier transformation over each 3-minute segment of cognitive reactivity-recovery testing. Spectral power (in milliseconds squared per hertz) was quantified for conventional bandwidths, including the high-frequency (HF) bandwidth (0.15 to 0.40 Hz), which served as a marker of vagal-HR modulation. Respiration rate (breaths per minute) was sensed by the ECG electrodes as changes in transthoracic impedance.
Cognitive reactivity-recovery testing began with a 10-minute adaptation that was followed by 2 successive 3-minute resting intervals (baseline 1 and baseline 2). The initial 3-minute task was a cognitive control procedure. Subjects viewed 18 neutral slides (International Affective Picture System10 [IAPS]) on a computer screen in 10-second intervals, followed by a 3-minute recovery. Subjects next performed 2 randomly ordered tasks in 3-minute intervals, with each followed by 3-minute recovery. The Paced Auditory Serial Addition Test (PASAT)11 used audio-taped instruction to test attentional ability and memory. Three potential answers to each serial addition were presented on a computer screen, and subjects responded by pressing 1 of 3 corresponding keys on a handheld device that was synchronized to the computer. Performance demands were progressively increased by reducing the interval between serial digits from 2.5 to 2.0 seconds.7 The emotional Stroop test12 (eStroop) required subjects to press 1 of 4 colored buttons on a handheld device (red, yellow, green, and blue) that corresponded with the color in which 8 neutral words and 10 negatively valanced words (eg, murder and rape) were displayed on a computer screen. Each word was displayed for 1.5 seconds, with a response interval of 8.5 seconds. Neutral and negative words were randomly ordered.
Subjects received 4 weekly and then 2 biweekly 1-hour sessions of BNT or AC. Each session began with a 10-minute review of cognitive-behavioral guidelines for managing daily stress.7 Training in BNT and AC was supplemented by 20-minute audio-taped exercises for daily home practice, which replicated procedures used in the 1-hour treatment sessions.
AC subjects were informed that they would be trained in autogenic relaxation to buffer the effect of daily stress on BP. Each session included a 30-minute audio-taped autogenic relaxation procedure.7 The stated training goal was to evoke passive relaxation by silently repeating autogenic (self-initiated) phrases that focused attention on sensations of heaviness, calm, and warmth in major skeletal muscle groups.
BNT subjects were informed that they would learn an active countering skill with HRV biofeedback to decrease the influence of stress on BP. As described previously,7 each session presented subjects with a standardized psychological task (eg, serial 7 subtraction) and physical challenge (sitting-to-standing) to evoke mild-to-moderate psychophysiologic arousal. On completing each task, subjects were trained to cognitively disengage from negative or aroused affect and to focus attention on slowing respiration within their comfort zone to 10-second cycles (6 breaths per minute). The training goal for each countering exercise was to increase RR spectral power at ≈0.1 Hz, as guided by the biofeedback display of the RR power spectrum (0.003 to 0.500 Hz) and breaths per minute. The biofeedback display also included real-time variations in HR and BRS, which were not the focus of training in this study. These indices were initially sampled across a window of ≈120 seconds, and the computer display was updated continuously. All of the physiological data were time synchronized and digitized at 1000 Hz by a customized biofeedback program (LabView 7.1, National Instruments).
Primary efficacy outcomes included daytime (8:00 am to 10:00 pm) and 24-hour ambulatory measures for SBP, DBP and PP. Secondary efficacy outcomes included RR interval, HF power, and BRS during cognitive reactivity-recovery testing.
On the basis of the method proposed by Rochon13 and findings from our previous trial of BNT,7 it was estimated that 62 subjects were required to detect a significant difference between BNT and AC, with statistical power of 80% and a type 1 error of 5%, with respect to RR-HF power during cognitive reactivity-recovery testing and BP reduction.
Baseline characteristics of BNT and AC groups were examined by Pearson χ2 tests and ANOVA. The effects of BNT and AC on daytime and 24-hour SBP, DBP, and PP were evaluated by an intention-to-treat approach using linear mixed models (LMMs) with random intercept and adjustment for age, sex, BMI, and antihypertensive medications (β-blockers, calcium channel blockers, angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, or diuretics; 0: not prescribed, 1: prescribed, and 2: prescribed but modified during this trial). Main-effect models contrasted each ambulatory BP variable before intervention with outcomes after BNT and AC. The use of the random intercept adjusted for variability in BP between individual subjects. Absolute risk reduction and number-needed-to-treat statistics were used to explore whether BNT versus AC decreased the number of subjects with uncontrolled BP (24-hour SBP ≥130 mm Hg) in comparison with pretreatment.
Repeated-measures LMM with a random intercept evaluated whether BNT versus AC augmented cardiovagal functioning across each 3-minute segment of cognitive reactivity-recovery testing, in contrast to a resting baseline (baseline 2). The tasks included a cognitive control (neutral IAPS slides10), recovery, PASAT,11 recovery, eStroop,12 and recovery. Each analysis adjusted for resting baseline 1 for each secondary outcome variable, age, sex, BMI, antihypertensive medications, breaths per minute, and tasks. LMM analyses also adjusted for the serial correlation between repeated measures across the testing protocol. Secondary outcomes included RR interval, HF power, and BRS. A log-10 transformation was applied to HF power and BRS to control for skewness. Each LMM tested the main effect for interventions (before versus after BNT and AC), tasks (resting baseline 2 versus each reactivity and recovery interval), and the treatment-by-tasks interaction. Only those effects with P<0.1 were included in the final model. Use of the random intercept controlled for variability between individual subjects.
The association between posttreatment measures of log-HF and log-BRS during the cognitive reactivity tests (PASAT11 and eStroop12) and 24-hour SBP and PP was explored with partial correlations that adjusted for antihypertensive medications, SBP or PP before intervention, and resting baseline for log-HF or log-BRS. Results are reported as the mean±SE unless otherwise noted. Statistical significance was defined by 2-tailed tests at P<0.05. Analyses were conducted on SAS 9.1 (SAS Institute Inc) and SPSS 17.0 (SPSS Inc).
The sample included 65 subjects (BNT: n=35; AC: n=30; Figure 1). Baseline characteristics did not differ between groups (Table 1). Among subjects prescribed antihypertensive medications (n=46; 70.8%), 5 BNT and 2 AC subjects reported a change in medication dose or number (P=0.32). Withdrawals did not differ between groups (BNT: n=4; AC: n=2; P=0.51).
Unadjusted means for daytime and 24-hour SBP, DBP, and PP, before and after intervention, are presented in Table 2. Table 3 indicates that BNT was associated independently with a significant reduction in daytime and 24-hour SBP (−2.4±0.9 mm Hg, P=0.009; and −2.1±0.9 mm Hg, P=0.03, respectively) and daytime and 24-hour PP (−1.7±0.6 mm Hg, P=0.004; and −1.4±0.6 mm Hg, P=0.02, respectively) but not DBP (P>0.10 for both). Significant main effects indicated that daytime and 24-hour SBP were lower among subjects without (versus with) antihypertensive medications, subjects with prescribed medications that were not changed (versus changed), and subjects who were younger or who had a lower BMI. AC did not evoke significant change on any ambulatory BP measure (P>0.10 for all).
Table S1 (available in the online Data Supplement, please see http://hyper.ahajournals.org) presents unadjusted means for nighttime SBP, DBP, and PP, whereas Table S2 indicates that these measures were not significantly reduced after BNT or AC. The number of subjects with uncontrolled BP (24-hour SBP: ≥130 mm Hg) decreased from pretreatment after BNT (n=17 of 35 to 12 of 35) but not AC (n=15 of 30 to 15 of 30; absolute risk reduction: 0.14 [95% CI: 0.01 to 0.27]; number needed to treat: 7 [95% CI: 4 to 57]).
Vagal Modulation of HR and BP
Figure 2 displays unadjusted means during cognitive reactivity-recovery testing for respiration, RR interval, HF power, and BRS. Baseline values for these variables did not differ significantly before versus after BNT or AC (P>0.30 for each).
Table 4 presents the covariate-adjusted effects of BNT and AC on RR interval, log-HF, and log-BRS during cognitive reactivity-recovery testing. RR interval increased significantly from pretreatment after BNT (P<0.0001) and in comparison with AC (P<0.0001), whereas RR interval decreased from pretreatment after AC (P=0.01). The intervention-by-task interaction was not significant (P=0.30).
Log-HF increased significantly after BNT in comparison with both pretreatment (P=0.01) and AC (P=0.005). Log-HF did not change significantly after AC (P=0.18). The intervention-by-task interaction was not significant (P=0.51).
Log-BRS did not change significantly after versus before BNT or AC (P≥0.35 for both comparisons). The intervention-by-task interaction was not significant (P=0.93). Table S3 presents the complete LMMs for RR interval, log-HF, and log-BRS with main-effect comparisons between baseline 1 and each cognitive reactivity-recovery task (please see the online Data Supplement).
Association Between Ambulatory BP and Cardiovagal Response to Cognitive Tasks
Table 5 indicates that daytime and 24-hour SBP and PP were inversely associated with log-HF during the PASAT11 and eStroop12 after BNT (P<0.05 in each case) but not after AC (P>0.05 in each case). Neither BNT nor AC demonstrated a stable inverse association between ambulatory BP and BRS (P>0.05).
The present findings help to resolve 2 issues that are tied historically to efforts to develop a behavioral intervention for hypertension. Whether such an intervention can further lower BP once subjects receive antihypertensive medications has become an increasingly important challenge, because combination drug therapy is now recommended for most patients with hypertension.14 Recent meta-analyses have reported that the combination of biofeedback and relaxation training reduced SBP and DBP6 or SBP only5 in comparison with an AC or behavioral placebo. However, the efficacy of this complementary behavioral treatment was not evaluated directly for patients who were prescribed antihypertensive medications. Importantly, these meta-analyses included 8 trials that enrolled medicated subjects, of which 6 trials15–20 reported null findings. In the present trial, ≈71% of subjects were prescribed ≥1 antihypertensive medication on enrollment with pharmacotherapy stable for ≥4 months to increase the likelihood that therapeutic benefit had been achieved. Nevertheless, BNT further reduced ambulatory SBP and PP after adjusting statistically for antihypertensive medications, whereas no change was observed for AC.
The second issue concerns whether repeated exposure to passive relaxation3 is necessary or sufficient to modify symptoms of sympathetic overactivity associated with emotional or environmental challenges. Autonomic dysfunction in hypertension is characterized by inhibition of vagal-HR modulation, attenuated BRS, heightened sympathetic modulation of cardiac function and vasomotor tone, and increased BP variability.21,22 In the present study, passive relaxation with autogenic training did not significantly change tonic or reflex vagal modulation of the heart. In contrast, BNT7 significantly increased tonic vagal-HR modulation. This was achieved by repeatedly exposing subjects to standardized psychological and physical challenges to provide active training in countering skills that included focused attention, affect regulation, and slow-paced breathing. HRV biofeedback reinforced and guided the development of countering skills toward the goal of increasing vagal-HR modulation during the psychological and physical challenges. To our knowledge, this is the first behavioral trial with hypertensive patients to demonstrate a concomitant improvement in ambulatory SBP and PP (as a surrogate, in the present context, for stroke volume) in association with increased vagal-HR modulation, as evidenced by RR-HF power and lengthening of the RR interval.
Previous trials have reported an increase in vagal-HR modulation among subjects with coronary heart disease after BNT and HRV biofeedback7 or a combination of breathing exercises with relaxation training or meditation.23 The present trial extends these findings by demonstrating this therapeutic effect in association with reduced ambulatory SBP and PP among subjects with hypertension. This outcome was observed after only an 8-week intervention, which may underestimate a potential benefit of BNT for long-term prognosis. Clinical trial evidence has shown that improved vagal-HR modulation or BRS independently contributes to reduced risk for cardiovascular events among patients with coronary heart disease24,25 or with hypertension.26 In heart failure, where chronic β-adrenergic antagonism has been demonstrated to improve longevity, this drug class augments reflex vagal-HR modulation at an efferent site of interaction involving blockade of cardiac sympathetic prejunctional β2 adrenoceptors that facilitate norepinephrine release.27 The potential use of BNT as a complementary antihypertensive treatment is further supported by the exploratory finding, albeit in a small sample, that ambulatory 24-hour SBP was reduced below the clinical threshold of 130 mm Hg for 1 of every 7 BNT subjects, representing an absolute risk reduction of 14% in comparison with AC. This modest exploratory outcome suggests that there is merit in conducting a larger scale trial to evaluate the long-term clinical efficacy of BNT.
Findings from the present study are limited by the short (8-week) intervention period. In addition, subjects were studied 1 week after the final session, at which time the maximum impact of treatment may have dissipated. Evidence of an independent association between BNT and BP reduction over at least a 6-month follow-up is needed to confirm its potential role as a complementary treatment for hypertension.5 In addition, BNT and AC interventions were supplemented with instructions for daily home practice using 20-minute audio-taped exercises. Adherence to home practice was not objectively monitored; therefore, the dose-response association between BNT and BP reduction cannot be accurately established in this trial. Finally, spontaneous BRS did not increase significantly after BNT and HRV biofeedback. Future research is needed to determine whether BNT and BRS biofeedback can reduce BP while improving baroreflex gain among patients with hypertension.
Our present trial demonstrates that, in contrast to passive relaxation, BNT with HRV biofeedback7 modestly reduces ambulatory SBP and PP while improving tonic vagal-HR modulation, independent of antihypertensive medications. BNT is well suited to physiologically define treatment objectives, to specify mechanisms of therapeutic change, and, therefore, to replicate treatment outcomes. For this reason, we propose BNT as a novel advance in the development of behavioral interventions for hypertension.
We are grateful to Dr Adriana Mechetiuc, Dr Susan Barry-Bianchi, Nazia Siddiqui, Carley Hamilton, and Samantha Young for their assistance.
Sources of Funding
This trial was supported by an unrestricted research grant from Unilever Corporate Research Biosciences. J.S.F. holds the Canada Research Chair in Integrative Cardiovascular Biology and is a career investigator of the Heart and Stroke Foundation of Ontario.
This trial has been registered at www.clinicaltrials.gov (identifier NCT00811811).
- Received October 20, 2009.
- Revision received November 8, 2009.
- Accepted January 22, 2010.
Alam GM, Smirk FH. Casual and basal blood pressures II: in essential hypertension. Br Heart J. 1943; 5: 156–160.
Dickinson HO, Campbell F, Beyer FR, Nicolson DJ, Cook JV, Ford GA, Mason JM. Relaxation therapies for the management of primary hypertension in adults. Cochrane Database Syst Rev. 2008; (1): CD004935.
Padwal RS, Hemmelgarn BR, McAlister FA, McKay DW, Grover S, Wilson T, Penner B, Burgess E, Bolli P, Hill M, Mahon J, Myers MG, Abbott C, Schiffrin EL, Honos G, Mann K, Tremblay G, Milot A, Cloutier L, Chockalingam A, Khan NA, Rabkin SW, Dawes M, Touyz RM, Tobe SW. The 2007 Canadian Hypertension Education Program recommendations for the management of hypertension: part 1–blood pressure measurement, diagnosis and assessment of risk. Can J Cardiol. 2007; 23: 529–538.
Butler GC, Senn BL, Floras JS. Influence of atrial natriuretic factor on spontaneous baroreflex sensitivity for heart rate in humans. Hypertension. 1995; 25: 1167–1171.
Center for the Study of Emotion and Attention-National Institute of Mental Health. International Affective Picture System (IAPS): Technical Manual and Affective Ratings. Gainesville, FL: NIMH-Center for the Study of Emotion and Attention, University of Florida; 1999.
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.
Frankel BL, Patel DJ, Horwitz D, Friedewald WT, Gaarder KR. Treatment of hypertension with biofeedback and relaxation techniques. Psychosom Med. 1978; 40: 276–293.
Jacob RG, Shapiro AP, O'Hara P, Portser S, Kruger A, Gatsonis C, Ding Y. Relaxation therapy for hypertension: setting-specific effects. Psychosom Med. 1992; 54: 87–101.
Floras JS, Hassan MO, Jones JV, Osikowska BA, Sever PS, Sleight P. Factors influencing blood pressure and heart rate variability in hypertensive humans. Hypertension. 1988; 11: 273–281.
Nolan RP, Jong P, Barry-Bianchi SM, Tanaka TH, Floras JS. Effects of drug, biobehavioral and exercise therapies on heart rate variability in coronary artery disease: a systematic review. Eur J Cardiovasc Prev Rehabil. 2008; 15: 386–396.
La Rovere MT, Bersano C, Gnemmi M, Specchia G, Schwartz PJ. Exercise-induced increase in baroreflex sensitivity predicts improved prognosis after myocardial infarction. Circulation. 2002; 106: 945–949.
Collier D, Davies L, Bernardi L, Sleight P, Shiel J, Patel N, Misra S, Martin T, Moore D, Fok H, Scott B, David C, Caulfield M. Baroreceptor function changes with differing blood pressure treatment during the Anglo-Scandanavian Cardiac Outcomes Trial: principal results from the cardiac autonomic reflex assessment trial (CARAT). J Hypertens. 2006; 25: S292.