Abstract
Background
Heart rate variability (HRV) analysis is uncommonly undertaken in patients with atrial fibrillation (AF) due to an assumption that ventricular response is random. We sought to determine the effects of head‐up tilt (HUT), a stimulus known to elicit an autonomic response, on HRV in patients with AF; we contrasted the findings with those of patients in sinus rhythm (SR).
Methods
Consecutive, clinically indicated tilt tests were examined for 207 patients: 176 in SR, 31 in AF. Patients in AF were compared to an age‐matched SR cohort (n = 69). Five minute windows immediately before and after tilting were analyzed using time‐domain, frequency‐domain and nonlinear HRV parameters. Continuous, noninvasive assessment of blood pressure, heart rate and stroke volume were available in the majority of patients.
Results
There were significant differences at baseline in all HRV parameters between AF and age matched SR. HUT produced significant hemodynamic changes, regardless of cardiac rhythm. Coincident with these hemodynamic changes, patients in AF had a significant increase in median [quartile 1, 2] DFA‐α2 (+0.14 [−0.03, 0.32], p < .005) and a decrease in sample entropy (−0.17 [−0.50, −0.01], p < .005).
Conclusion
In the SR cohort, increasing age was associated with fewer HRV changes on tilting. Patients with AF had blunted HRV responses to tilting, mirroring those seen in an age matched SR group. It is feasible to measure HRV in patients with AF and the changes observed on HUT are comparable to those seen in patients in sinus rhythm.
Keywords: atrial fibrillation, ECG signal processing, head‐up tilt, heart rate variability
1. INTRODUCTION
Heart rate variability (HRV) is a surrogate marker for the function of the autonomic nervous system (ANS) and the technique is widely available (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996). There are a variety of methods for the derivation of HRV, through the application of different mathematical functions to consecutive RR‐intervals. These mathematical functions fall broadly into three groups: time domain, frequency domain, and nonlinear analysis (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996).
A relationship between reduced HRV and prognosis has been shown in health (Hillebrand et al., 2013) and in numerous conditions, including after myocardial infarction and in patients with heart failure (Bilchick et al., 2002).
HRV techniques are generally not applied to patients in atrial fibrillation (AF; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996). This is an important limitation, as AF is not only very prevalent but it is present in 30%–50% of the heart failure population, a condition in which HRV has been shown to be useful in predicting outcomes (Bilchick et al., 2002). Recently, bridging this gap in knowledge has become even more pertinent due to the introduction of ablative interventions or device implantation (e.g. renal denervation, baroreceptor stimulators, vagal nerve stimulators, spinal cord stimulators) which modulate the ANS as a potential treatment strategy in heart failure and other diseases (Ardell et al., 2014; Patel et al., 2013).
The argument against the use of HRV techniques in patients with AF is based upon the assertion that the RR intervals in AF are truly less dependent on physiological mechanisms measurable with HRV. Though in clinical examination AF is characterized crudely by an irregularly irregular pulse, generally considered random, there is a growing body of evidence that supports a different view (Carrara et al., 2015; Cygankiewicz et al., 2015; Hayano, Sakata, Okada, Mukai, & Fujinami, 1998; Hayano et al., 1997; Rawles & Rowland, 1986). Rawles and Rowland (1986) demonstrated in 74 patients in AF, using an auto‐correlation technique, that at rest approximately a third of patients had a nonrandom ventricular rhythm. While the effect of the ANS on the sinus node is a major determinant of HRV in sinus rhythm (SR), the ANS is equally important in AF, through its effects on the refractory period and conductivity of the AV node, the frequency and irregularity of atrial impulses and the degree of concealed conduction (Bollmann et al., 2006; Hayano et al., 1998; Lim et al., 2011).
The purpose of this study was to determine the validity of measuring HRV in patients with AF. To achieve this we used head‐up tilt testing (HUT) as an intervention that predictably activates the sympathetic nervous system (SNS) and leads to withdrawal of the parasympathetic nervous system (PNS; Mehlsen, Kaijer, & Mehlsen, 2008). We contrasted the effects of HUT on HRV in a cohort of individuals with AF and a group in SR.
The aging process is an important consideration in studies of autonomic physiology (Petersen, Williams, Gordon, Chamberlain‐Webber, & Sutton, 2000). Not only is increasing age a risk factor for AF but it has also been shown to reduce HRV in cross‐sectional studies (Laitinen, Niskanen, Geelen, Lansimies, & Hartikainen, 2004; Sosnowski, Macfarlane, & Tendera, 2011; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996). It is vital that we match for age and interpret our findings in the context of a more elderly population. To aid this interpretation we carried out additional analyses on a cohort in SR to establish the effect of aging on HRV responses to tilt in our patients.
2. METHODS
2.1. Study patients
We obtained data retrospectively on consecutive patients with permanent AF who underwent clinically indicated tilt testing at two hospitals (over a cumulative 9 years). All patients in SR from one of the hospitals also had their data analyzed to provide the control population. Patients were excluded from this analysis if they experienced syncope or presyncope in the tilt phase or had a paced rhythm. Data were available for 176 patients in SR and 31 in AF. National Health Service (UK) management permission for use of anonymized patient data for ethical research was obtained.
2.2. Tilt‐test protocol
The tilt table examination was performed in a dedicated room. Patients were fasted for two hours prior to the HUT and did not have medications stopped. A motorized bed with footplate support was used to achieve tilt angles of 60–80°. Each patient had a 10 min supine baseline period after which they were subjected to 20 min of tilt.
2.3. Data acquisition and preprocessing
Continuous, noninvasive, high resolution, beat‐to‐beat heart rate (1,000 Hz sampling frequency), and blood pressure monitoring (500 Hz sampling frequency) was performed at both sites using either the Task Force Monitor (CNS SystemsMedizintechnik AG, Graz, Austria) or Nexfin (BMEYE B.V, Amsterdam, The Netherlands). The Task Force Monitor also estimates cardiac output and total peripheral resistance using impedance cardiography.
Time series for heart rate (beat to beat NN intervals) were extracted and automatically filtered to exclude artifacts and ectopics using a validated and freely available programme Kubios HRV (http://kubios.uef.fi).
2.4. Heart rate variability
We standardized our analysis windows to five minutes to minimize bias as it is known that the total variance of HRV increases in proportion to the length of recording, in line with international recommendations (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996). Windows immediately before and during the first five minutes of HUT were analyzed. Time domain, frequency domain and nonlinear methods for determining HRV were applied to the data (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996).
Time domain analysis involves application of simple statistical techniques straight to the successive RR intervals. We elected to study only SDRR (standard deviation of successive RR intervals) and RMSSD (root of the mean squared differences in successive RR intervals; Tarvainen, Niskanen, Lipponen, Ranta‐Aho, & Karjalainen, 2014) as both of these parameters can be used in 5 min recordings of RR intervals and the other time domain parameters are either derived from them or are highly correlated with them (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996).
Frequency domain analysis required the RR interval time series to be converted to an equidistantly sampled series and this was performed using the cubic spline interpolation method (Tarvainen et al., 2014). The power spectral density was estimated using parametric autoregressive modeling (order number 16 without factorization) and absolute power in the low frequency (LF: 0.04–0.15 Hz) and high frequency (HF: 0.15–0.4 Hz) bands calculated (Tarvainen et al., 2014). These powers can be normalized to minimize the effects of changes in total power on this parameters, e.g. normalized LF (LFnu) is calculated as: LF/(Total power−very low frequency power).
Finally we also applied the following nonlinear methods of HRV analysis: Poincaré plots, detrended fluctuation analysis and entropy. Poincaré plots are a graphical representation of the correlation between successive RR intervals. It can be assessed qualitatively by looking at the shape of the plot and quantitatively by fitting an ellipse to the plot and calculating the standard deviation of the points perpendicular to the line of identity (SD1) and along the line of identity (SD2; Tarvainen et al., 2014).
Detrended fluctuation analysis measures correlation with the signal for different time scales. A series of RR intervals are integrated and are divided into a series of regular intervals. For each interval the fluctuation of the data from a straight line of linear interpolation is calculated. DFA‐α1 corresponds to short‐term fluctuations within an interval range of 4–16 whereas DFA‐α2 characterizes longer term fluctuations within the interval range 16–64 (Tarvainen et al., 2014).
Sample entropy, which refers to the degree of irregularity or randomness with a series and are estimates for the negative natural logarithm of the condition probability that a length of data having repeated itself within a tolerance r for m points, will also repeat itself for m + 1 points. We used the default value of m = 2 and r = .2 SDRR (Tarvainen et al., 2014).
2.5. Statistics
Some of the HRV parameters were not normally distributed and so we adopted nonparametric statistical analysis throughout for consistency. Continuous variables are summarized as median (quartile 1, quartile 3) and compared using the Mann–Whitney U test (independent samples), the Kruskal–Wallis test (more than two independent samples) and the Wilcoxon signed‐rank test (dependent samples, i.e. comparing parameters before and after HUT in the same cohort). Categorical variables are presented as counts or proportions (%) and analyzed using Fisher's exact text. Strength of correlation between variables are presented using the Pearson's product‐moment correlation (r). A p ≤ 0.05 was considered statistically significant for analysis of baseline clinical features and hemodynamics of patients. This level of significance was made more stringent to p ≤ 0.005 when analyzing the HRV parameters using the Bonferroni method to correct for multiple testing. A concern with the Bonferroni method is that it can elevate the type II error rate (accepting the null hypothesis when the alternative is correct) and for that reason we have also provided complete p values or at least made a summative distinction between a parameter that changed at p < .05 and one at p < .005. All analyses were performed using SPSS (Version 22, IBM).
3. RESULTS
Data were available in total for 176 patients in SR and 31 patients in AF. Of these, all SR patients and 19 AF patients were from the same institution and had a full data set including noninvasive beat to beat heart rate, blood pressure, stroke volume, and peripheral resistance. Data for the remaining 12 patients in AF were obtained from another institution for whom noninvasive stroke volume or peripheral resistance measurements were not available. Correlations between HRV variables at rest.
3.1. The effect of HUT in AF and SR
The demographics of the 31 patients in AF and 69 age‐matched patients in SR are summarized in Table 1. Patients with AF were significantly more likely to have hypertension and be on more medications (angiotensin converting enzyme inhibitor, angiotensin receptor blocker, beta‐blocker, calcium channel blocker, and digoxin). Only seven (22.6%) patients with AF were not on any of the six classes of medication detailed in Table 1, compared with 54 (78.3%) in the SR cohort.
Table 1.
Demographics, past medical and medication history of the patients with atrial fibrillation and age matched sinus rhythm
| Atrial fibrillation (n = 31) | Sinus rhythm (n = 69) | p | |
|---|---|---|---|
| Age | 74.3 (68.9, 83.8) | 70.3 (62.9, 77.7) | .056 |
| Male | 20 (64.5%) | 46 (66.7%) | 1.000 |
| Diabetes | 5 (16.1%) | 5 (7.2%) | .277 |
| Hypertension | 23 (74.2%) | 11 (15.9%) | <.001 |
| Heart failure | 2 (6.5%) | 0 (0%) | .094 |
| Medications | |||
| ACEi/ARB | 15 (48.4%) | 12 (17.4%) | .002 |
| Beta‐blockers | 10 (32.3%) | 3 (4.3%) | <.001 |
| CCB | 8 (25.8%) | 3 (4.3%) | .003 |
| Digoxin | 7 (22.6%) | 0 (0%) | <.001 |
| Diuretics | 7 (22.6%) | 6 (8.7%) | .103 |
| Median | 2 (1, 2) | 0 (0, 0) | <.001 |
Data are presented as median (quartile 1, quartile 3) or count (%).
ACEi, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; CCB, calcium channel blocker.
Head‐up tilt causes a decrease in stroke volume, which is coupled with an increase in blood pressure (diastolic), heart rate and total peripheral resistance (Table 2). The magnitude and direction of change, though similar for both cohorts, were statistically more convincing in patients with SR.
Table 2.
Baseline and change with HUT of hemodynamic and HRV data
| Cohort | Baseline | Change from baseline after HUT | ||||
|---|---|---|---|---|---|---|
| AF (n = 31) | SR (n = 81) | p (AF vs. SR) | AF | SR | p (AF vs. SR) | |
| Hemodynamics | ||||||
| SBP (mm Hg) | 126.5 (112.0, 139.1) | 123.1 (112.7, 134.9) | .469 | +5.2 (−2.5, 12.1) | +6.6 (−1.7, 16.6)** | .431 |
| DBP (mm Hg) | 79.2 (68.9, 83.5) | 77.8 (70.6, 84.5) | .871 | +4.8 (−0.6, 13.1)* | +10.4 (1.5, 16.1)** | .100 |
| HR (beats/min) | 74.9 (67.6, 87.5) | 70.7 (60.9, 78.2) | .018 | +3.8 (1.4, 7.1)** | +4.6 (1.7, 8.2)** | .776 |
| Stroke index (ml/m2) | 36.8 (25.9, 42.2) | 34.7 (29.9, 41.5) | .729 | −5.2 (−8.1, 1.0)* | −5.7 (−10.0, −1.1)** | .417 |
| Cardiac index (L/[min.m2]) | 2.41 (2.12, 2.94) | 2.52 (2.17, 2.98) | .663 | −0.13 (−0.44, 0.26) | −0.26 (−0.54, 0.06)** | .326 |
| TPR index (dyne*s*m2/cm5) | 3,464 (2,666, 3,792) | 2,930 (2,585, 3,488) | .252 | +612 (−168, 1,092)* | +628 (106, 1,028)** | .515 |
| Time | ||||||
| SDNN (ms) | 102.8 (28.3, 160.4) | 39.2 (25.3, 57.1) | .003 | +0.9 (−13.9, 18.4) | +3.4 (−6.7, 18.8)* | .396 |
| RMSSD (ms) | 139.2 (18.4, 208.1) | 26.9 (15.9, 47.2) | .001 | −5.5 (−21.4, 4.2) | −2.2 (−22.1, 21.5) | .291 |
| Frequency | ||||||
| LF (ms2) | 2,075 (135, 5,243) | 254 (156, 612) | .005 | 65.1 (−223, 320) | −21.6 (−151, 137) | .364 |
| HF (ms2) | 3,921 (106, 10,491) | 209 (75, 539) | .001 | −92.5 (−2,124, 119)* | −28.0 (−234, 36)* | .149 |
| LF/HF | 0.58 (0.46, 1.02) | 1.39 (0.80, 2.25) | <.001 | +0.10 (−0.12, 0.35)* | +0.61 (−0.32, 2.33)** | .216 |
| LFnu (%) | 36.4 (31.4, 50.4) | 58.0 (44.3, 69.2) | <.001 | +4.4 (−4.0, 9.9) | +7.1 (−4.3, 17.5)** | .341 |
| HFnu (%) | 63.1 (49.6, 68.2) | 41.8 (30.7, 55.6) | <.001 | −4.2 (−9.8, 3.8) | −7.1 (−17.2, 4.3)** | .331 |
| Nonlinear | ||||||
| SD1 (ms) | 98.5 (13.0, 147.4) | 19.0 (11.3, 33.4) | .001 | −3.9 (−15.2, 3.0) | −1.4 (−10.9, 5.8) | .393 |
| SD2 (ms) | 110.6 (34.9, 163.9) | 49.3 (33.6, 66.4) | .006 | +1.9 (−14.4, 29.3) | +4.7 (−6.4, 26.7)** | .669 |
| DFA‐α1 | 0.70 (0.61, 1.01) | 1.06 (0.84, 1.22) | <.001 | +0.01 (−0.11, 0.09) | +0.07 (−0.14, 0.45)* | .179 |
| DFA‐α2 | 0.69 (0.54, 0.86) | 0.94 (0.80, 1.11) | <.001 | +0.14 (−0.03, 0.32)** | +0.14 (−0.03, 0.36)** | .806 |
| Sample entropy | 1.78 (1.26, 2.05) | 1.26 (0.94, 1.55) | <.001 | −0.17 (−0.50, −0.01)** | −0.25 (−0.55, 0.04)** | .887 |
SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; TPR, total peripheral resistance; SDNN, standard deviation of the RR interval; RMSSD, root of the mean squared differences in successive RR intervals; LFnu, low frequency power in normalized units; HFnu, high frequency power in normalized units; SD1, minor axis on Poincaré plots; SD2, major axis on Poincaré plots, DFA, detrended fluctuation analysis.
*p ≤ .05 (within group delta from baseline, paired t‐test); **p ≤ .005 (within group delta from baseline, paired t‐test). DFA1, DFA2 and Sample Entropy are dimensionless.
All HRV parameters at rest were significantly different between the AF and SR cohorts. On HUT only two parameters (DFA‐α2 and sample entropy) changed significantly (p < .005) in both groups (Table 2). SDRR, LFnu and SD2 increased in patients in SR on HUT, whereas HF decreased in patients with AF at the uncorrected significance level of p < .05. There was no overall difference in the direction of change between either group.
3.2. The effect of aging on cardiovascular autonomic reflexes
The cohort of 176 patients in SR was divided into tertiles of age (with median ages 22, 47, and 73 years). Their demographic data are detailed in Table 3 and suggests that the three groups were balanced apart from there being proportionally more females in the youngest cohort. In particular there were no differences with respect to prescribed medications.
Table 3.
Demographics, medical and medication history of the patients in sinus rhythm across tertiles of age
| 0–30 (n = 50) | 30–60 (n = 69) | 60+ (n = 57) | p | |
|---|---|---|---|---|
| Tertiles of age | ||||
| Age | 22.0 (18.5, 24.9) | 47.1 (38.4, 52.4) | 72.7 (66.2, 79.4) | <.001 |
| Male | 20 (40.0%) | 34 (49.3%) | 38 (66.7%) | .018 |
| Diabetes | 9 (18.0%) | 5 (7.2%) | 4 (7.0%) | .131 |
| Hypertension | 10 (20.0%) | 14 (20.3%) | 9 (15.8%) | .807 |
| Heart failure | 1 (2.0%) | 0 (0%) | 0 (0%) | .284 |
| Medications | ||||
| ACEi/ARB | 6 (12.0%) | 10 (14.5%) | 10 (17.5%) | .659 |
| Beta‐blockers | 4 (8.0%) | 6 (8.7%) | 2 (3.5%) | .522 |
| CCB | 5 (10.0%) | 6 (8.7%) | 3 (5.3%) | .648 |
| Diuretics | 4 (8.0%) | 3 (4.3%) | 5 (8.8%) | .562 |
| Total | 0.0 (0.0, 0.3) | 0.0 (0.0, 0.0) | 0.0 (0.0, 0.0) | .947 |
Data are presented as median (quartile 1, quartile 3) or count (%). Abbreviations as per Table 1.
There were significant differences at rest between the three tertiles of age with respect to hemodynamic function and HRV (Table 4). Stroke volume index decreased with age, whereas resting heart rate did not change. With advancing age all of the HRV parameters except for DFA‐α1 and DFA‐α2 were significantly attenuated (Table 4). Throughout the tertiles of age, upon HUT, blood pressure, heart rate and total peripheral resistance index increased whilst stroke volume index decreased. However, the augmentation in heart rate was attenuated as was the decline in stroke volume index with increasing age. Furthermore, the HRV response to HUT became blunted with age, with 11/12 HRV parameters changing at a significance level of p < .005 in the youngest tertile, 9/12 in the middle cohort and only 2/12 in the oldest tertile.
Table 4.
Baseline and change with HUT of hemodynamic and HRV data across the tertiles of age in patients with sinus rhythm
| Age tertiles (years) | Baseline | Change from baseline after HUT | ||||||
|---|---|---|---|---|---|---|---|---|
| 0–30 (n = 50) | 30–60 (n = 69) | 60+ (n = 57) | p (between groups) | 0–30 | 30–60 | 60+ | p (between groups) | |
| Hemodynamics | ||||||||
| SBP (mm Hg) | 114.5 (108.4, 122.6) | 124.9 (112.8, 135.1) | 123.3 (113.2, 135.0) | .004 | +9.3 (3.7, 11.0)** | +6.6 (0.3, 17.0)** | +6.7 (−1.4, 14.7)** | .419 |
| DBP (mm Hg) | 73.8 (67.7, 81.5) | 83.6 (76.5, 88.6) | 75.9 (69.8, 81.6) | <.001 | +14.3 (6.2, 19.3)** | +10.6 (2.2, 10.6)** | +10.3 (2.5, 15.7)** | .102 |
| HR (beats/min) | 72.5 (64.8, 79.7) | 72.2 (61.9, 80.6) | 71.4 (61.9, 78.5) | .800 | +8.8 (5.6, 14.9)** | +6.1 (2.8, 11.2)** | +4.4 (1.1, 7.3)** | <.001 |
| Stroke index (ml/m2) | 50.2 (42.1, 57.3) | 38.2 (31.3, 44.6) | 34.8 (29.2, 40.5) | <.001 | −9.8 (−15.6, −4.5)** | −7.8 (−13.0, −1.7)** | −5.1 (−9.0, −1.5)** | .004 |
| Cardiac index (L/[min.m2]) | 3.39 (3.06, 4.08) | 2.68 (2.16, 3.15) | 2.52 (2.20, 3.05) | <.001 | −0.35 (−0.78, 0.06)** | −0.33 (−0.65, 0.02)** | −0.25 (−0.59, 0.09)** | .582 |
| TPR index (dyne*s*m2/cm5) | 1,985 (1,714, 2,479) | 2,960 (2,450, 3,675) | 2,925 (2,516, 3,305) | <.001 | +532 (162, 750)** | +661 (328, 1,224)** | +617 (7.7, 1,030)** | .218 |
| Time domain | ||||||||
| SDRR (ms) | 70.2 (48.9, 88.5) | 43.7 (29.6, 62.4) | 40.2 (26.3, 58.3) | <.001 | +0.2 (−9.5, 8.6)** | +7.0 (−7.0, 19.6)* | +1.8 (−7.8, 18.7) | .144 |
| RMSSD (ms) | 47.8 (37.0, 82.2) | 30.2 (17.4, 55.20 | 27.9 (16.7, 52.7) | <.001 | −19.3 (−49.9, −4.2)** | −4.0 (−18.5, 14.5) | +0.6 (−26.8, 19.0) | .001 |
| Frequency domain | ||||||||
| LF (ms2) | 1,092 (651, 2,381) | 492 (235, 806) | 251 (142, 600) | <.001 | −14.8 (−350, −15)** | +24.2 (−150, 430) | −52.6 (−155, 126) | .357 |
| HF (ms2) | 1,085 (456, 2,485) | 310 (108, 926) | 189 (76, 584) | <.001 | −592 (−1,651, −592)** | −71.9 (−392, 21)** | −23.6 (−227, 62) | <.001 |
| LF/HF | 1.13 (0.68, 1.85) | 1.69 (0.93, 2.37) | 1.25 (0.68, 2.13) | .037 | +1.56 (0.82, 3.15)** | +1.23 (0.40, 3.57)** | +0.33 (−0.25, 2.33)** | .005 |
| LFnu (%) | 53.0 (40.3, 64.8) | 62.7 (48.0, 70.3) | 55.6 (40.4, 67.9) | .037 | +20.3 (12.5, 26.3)** | +12.2 (3.6, 23.5)** | +6.3 (−4.8, 19.4)* | <.001 |
| HFnu (%) | 46.9 (35.1, 59.6) | 37.2 (29.6, 51.8) | 44.4 (32.0, 59.1) | .037 | −20.2 (−26.2, −12.3)** | −12.1 (−23.4, −3.58)** | −6.3 (−19.1, 4.8)* | <.001 |
| Nonlinear | ||||||||
| SD1 (ms) | 33.9 (26.2, 58.2) | 21.4 (12.3, 39.1) | 19.8 (11.8, 37.3) | <.001 | −14.5 (−24.7, −2.8)** | −3.3 (−14.0, 1.9)** | −1.4 (−10.4, 7.9) | <.001 |
| SD2 (ms) | 84.9 (62.7, 108.8) | 55.7 (38.8, 81.3) | 49.3 (34.1, 66.4) | <.001 | +5.2 (−8.5, 14.3) | +9.4 (−6.2, 28.4)** | +3.5 (−7.3, 27.2) | .406 |
| DFA‐α1 | 1.00 (0.84, 1.26) | 1.13 (0.92, 1.33) | 1.04 (0.77, 1.19) | .082 | +0.28 (0.11, 0.48)** | +0.24 (−0.06, 0.42)** | +0.06 (−0.21, 0.49) | .010 |
| DFA‐α2 | 0.84 (0.78, 0.94) | 0.87 (0.77, 1.05) | 0.94 (0.80, 1.06) | .142 | +0.13 (−0.04, 0.28)** | +0.12 (−0.03, 0.25)** | +0.18 (−0.3, 0.37)** | .260 |
| Sample entropy | 1.58 (1.36, 1.77) | 1.42 (1.19, 1.63) | 1.20 (0.86, 1.49) | <.001 | −0.45 (−0.66, −0.18)** | −0.30 (−0.60, −0.05)** | −0.20 (−0.55, 0.13)** | .070 |
Data are presented as median (quartile 1, quartile 3).
*p ≤ .05 (within group delta from baseline); **p ≤ .005 (within group delta from baseline). Abbreviations as per Table 2.
Correlations between each of the HRV parameters in the SR (N = 176) and AF (N = 31) cohorts are shown in Table 5. There were strong correlations some of the nonlinear parameters (SD1, SD2, and DFA‐α1) and linear parameters.
Table 5.
Correlations (r) between heart rate variability parameters in the sinus rhythm (SR‐ in white) population (N = 176) and atrial fibrillation (AF‐ in gray) population (N = 31)
| SR | AF | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SDNN (ms) | RMSSD (ms) | LF (ms2) | HF (ms2) | LF/HF | LFnu (%) | HFnu (%) | SD1 (ms) | SD2 (ms) | DFA‐α1 | DFA‐α2 | Sample entropy | |
| SDNN (ms) |
0.992 p < .001 |
0.926 p < .001 |
0.943 p < .001 |
−0.476 p = .007 |
−0.614 p < .001 |
0.613 p < .001 |
0.992 p < .001 |
0.996 p < .001 |
−0.572 p = .001 |
−0.737 p < .001 |
0.703 p < .001 |
|
| RMSSD (ms) |
0.892 p < .001 |
0.908 p < .001 |
0.935 p < .001 |
−0.514 p = .003 |
−0.664 p < .001 |
0.663 p < .001 |
1.000 p < .001 |
0.976 p < .001 |
−0.637 p < .001 |
−0.762 p < .001 |
0.735 p < .001 |
|
| LF (ms2) |
0.871 p < .001 |
0.740 p < .001 |
0.970 p < .001 |
−0.366 p = .043 |
−0.474 p = .007 |
0.474 p = .007 |
0.908 p < .001 |
0.930 p < .001 |
−0.461 p = .009 |
−0.676 p < .001 |
0.663 p < .001 |
|
| HF (ms2) |
0.841 p < .001 |
0.878 p < .001 |
0.765 p < .001 |
−0.399 p = .26 |
−0.530 p = .002 |
0.529 p = .002 |
0.935 p < .001 |
0.939 p < .001 |
−0.506 p = .004 |
−0.659 p < .001 |
0.696 p < .001 |
|
| LF/HF |
−0.227 p = .002 |
−0.444 p < .001 |
−0.106 p = .160 |
−0.327 p < .001 |
0.904 p < .001 |
−0.905 p < .001 |
−0.514 p = .003 |
−0.453 p = .010 |
0.886 p < .001 |
0.393 p = .029 |
−0.448 p = .012 | |
| LFnu (%) |
−0.274 p < .001 |
−0.553 p < .001 |
−0.063 p = .403 |
−0.444 p < .001 |
0.823 p < .001 |
−1.000 p < .001 |
−0.664 p < .001 |
−0.580 p = .001 |
0.956 p < .001 |
0.595 p < .001 |
−0.555 p = .001 |
|
| HFnu (%) |
0.275 p < .001 |
0.553 p < .001 |
0.065 p = .389 |
0.446 p < .001 |
−0.823 p < .001 |
−1.000 p < .001 |
0.663 p < .001 |
0.579 p = .001 |
−0.955 p < .001 |
−0.595 p < .001 |
0.555 p = .001 |
|
| SD1 (ms) |
0.892 p < .001 |
1.000 p < .001 |
0.740 p < .001 |
0.879 p < .001 |
−0.444 p < .001 |
−0.553 p < .001 |
0.553 p < .001 |
0.976 p < .001 |
−0.637 p < .001 |
−0.762 p < .001 |
0.735 p < .001 |
|
| SD2 (ms) |
0.986 p < .001 |
0.807 p < .001 |
0.880 p < .001 |
0.785 p < .001 |
−0.150 p = .047 |
−0.163 p = .031 |
0.164 p = .029 |
0.807 p < .001 |
−0.538 p = .002 |
−0.713 p < .001 |
0.678 p < .001 |
|
| DFA‐α1 |
−0.242 p = .001 |
−0.539 p < .001 |
−0.066 p = .381 |
−0.375 p < .001 |
0.731 p < .001 |
0.907 p < .001 |
−0.904 p < .001 |
−0.539 p < .001 |
−0.123 p = .103 |
0.551 p = .001 |
−0.598 p < .001 |
|
| DFA‐α2 |
−0.081 p = .287 |
−0.227 p = .002 |
−0.189 p = .012 |
−0.100 p = .188 |
0.187 p = .013 |
0.156 p = .039 |
−0.154 p = .041 |
−0.227 p = .002 |
−0.032 p = .674 |
0.231 p = .002 |
−0.698 p < .001 |
|
| Sample Entropy |
−0.024 p = .754 |
0.111 p = .141 |
0.096 p = .204 |
0.175 p = .020 |
−0.279 p < .001 |
−0.162 p = .032 |
0.164 p = .029 |
−0.111 p = .141 |
−0.057 p = .453 |
−0.178 p = .018 |
−0.189 p = .012 |
|
Abbreviations as per Table 2.
4. DISCUSSION
The main findings of this study are: (i) For patients in sinus rhythm, HRV at rest and in response to HUT attenuates with age; (ii) Patients in AF demonstrate similar changes in HRV on HUT to an age‐matched cohort in SR; (iii) The nonlinear measures of HRV appear more discriminatory in both AF compared with the conventional linear methods (time and frequency domain).
4.1. The effect of age on cardiovascular responses to HUT in SR patients
4.1.1. Hemodynamics
The process of shifting from a supine to an upright position results in an immediate reduction in venous return and up to a 20% reduction in stroke volume. In response, there is an activation of various homeostatic pathways, one of which is the ANS, which functions to maintain cerebral blood flow and prevent syncope (Mourot et al., 2007). In our cohort, we demonstrated that our HUT protocol was effective in inducing an adequate hemodynamic stress. We observed a reduction in stroke volume index, which was associated with an increase in blood pressure, heart rate and total peripheral resistance index.
In the supine position, the oldest tertile had the lowest stroke volume and compared with the youngest tertile had higher blood pressures and vascular resistance. Upon HUT, the increase in heart rate was less pronounced in the older cohort as was the decrease in stroke index. Cumulatively, these responses seek to maintain cardiac output (heart rate x stroke volume indexed for body surface area). Laitinen and colleagues described the effect of aging upon response to HUT in 63 individuals and found results different from ours (Laitinen et al., 2004). Similar to our data they showed that elderly subjects had smaller increases in heart rate, however, in contrast to our findings they showed that this cohort also had a larger decrease in stroke volume and larger increase in total peripheral resistance upon HUT. The most likely explanation is that the two studies have examined different populations. Laitinen and colleagues studied a healthy population who were not on any medication compared with our cohort who all had previously experienced syncope and at least a fifth had other significant comorbidities. Medications are known to influence cardiovascular responses but in our SR cohort there were no significant differences between the tertiles of age with respect to blood pressure lowering medications. Furthermore, there were differences between our studies with respect to when data were collected. In Laitinen's study all patients were rested supine for 3 hr before a 5 min baseline recording of heart rate was obtained (our protocol mandated a 10 min rest period) and were sampled at minutes 5–10 after HUT (our protocol mandated minutes 0–5).
4.1.2. Heart rate variability
Aging affects HRV both at rest and under dynamic testing using HUT. Consistent with the wider literature our data confirm that HRV reduces with age, both at rest and in response to HUT (Sosnowski et al., 2011). At rest, we did not find a change in either DFA‐α1 or DFA‐α2 with age. Others have reported similar results whilst some groups have shown a decrease in DFA‐α1 and an increase in DFA‐α2 with advancing age in health volunteers (Shiogai, Stefanovska, & McClintock, 2010; Voss, Schroeder, Heitmann, Peters, & Perz, 2015). The likely contributors to this discrepancy are: (i) our study population were not healthy volunteers; (ii) approximately a fifth of our population in SR were on cardiovascular medications (Corino et al., 2013); and (iii) our study numbers were modest and hence our investigation may be underpowered.
In the youngest tertile (0–30 years), seven of the nine HRV parameters changed significantly on HUT. Only SDRR and its correlate SD2 did not change.(Hoshi, Pastre, Vanderlei, & Godoy, 2013) In the 30–60 years of age cohort, RMSSD also failed to change significantly with HUT. Finally in the 60 years and older cohort, only two HRV parameters changed (at p < .005), DFA‐α2 and sample entropy, both of them nonlinear parameters. Studies examining the effect of HUT upon HRV, using time and frequency domain, mirror our findings and have concluded that the response to HUT in younger individuals reflects parasympathetic withdrawal at the cardiac level, which diminishes with aging and is associated with a concomitant increase in sympathetic tone to the periphery (Laitinen et al., 2004).
4.2. Comparison of responses in patients with AF and SR
Frequency domain analyses of HRV in AF have failed to detect changes in response to manoeuvres known to affect HRV in SR and our data lends further support to this assertion (Hayano et al., 1997; Leung et al., 2005). DFA‐α2 and sample entropy are the only two HRV parameters that changed significantly (p < .005) in patients in AF and/or age‐matched SR upon HUT. Though the direction of change was identical between the two groups, the magnitude of difference is likely to be different (though we are statistically underpowered to demonstrate the latter). Furthermore, there were three other HRV parameters that changed in patients with SR but did not in AF, when the type 1 error rate was reduced to 0.05: SDRR, SD2, LFnu (the former two have previously been shown to be well positively correlated; Hoshi et al., 2013). It is not unexpected to see differences in HRV effects, since our two populations are different and because of this we would always recommend analyzing HRV in patients with AF separately from those in SR. Nonetheless, our findings suggest that though HRV data may be less interpretable in AF, certain parameters do have discriminatory values rather than just the “random chaos” of ventricular response.
However, a feature of note is how few HRV changes were actually observed on HUT, even in the SR cohort. This highlights the importance of aging on HRV as discussed above. In both the oldest tertile in SR and the AF group, the nonlinear measures were the only parameters that changed significantly.
One might ask why there is a differential effect depending on which measure of HRV is employed. There is no gold standard technique for measuring HRV and currently no one method can be described as superior to another; rather each provides complementary information (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996). Though many groups have attempted to attribute individual HRV parameters to a particular limb of the autonomic nervous system, to do so is an oversimplification of what is a complicated network. At best, HRV allows an insight into autonomic modulation; as tone increases, modulation increases but once tone remains elevated and saturation occurs, modulation decreases.
4.2.1. Nonlinear measures of HRV
There is little doubt that heart rate and its variability are complex phenomena which arise from an intricate network of regulatory pathways. Heart rate is likely to be sensitive to initial conditions but this dependence is likely to diverge exponentially with time. In mathematics these types of systems are best described as nonlinear, which are fundamentally different from linear systems (examples of which include time and frequency domain analyses).This description of the underlying principles of nonlinear methods makes it immediately appealing as a technique for AF due to the apparent randomness of the latter. We examined three types of nonlinear analysis: Poincaré plots (Hoshi et al., 2013), DFA (Castiglioni et al., 2011) and entropy (Porta et al., 2007).
SD1 and SD2 represent the standard deviation in the minor and major axis of a fitted ellipse to a plot of RR interval against the subsequent RR interval. However, as derivation of these variables is based on simple statistics, groups have questioned whether analysis of Poincaré plots truly reflects a nonlinear technique. Hoshi and colleagues performed linear correlation amongst time domain, frequency domain, and nonlinear HRV parameters in 65 healthy individuals and 114 patients with coronary artery disease. Their data showed that SD1 is highly correlated with RMSSD (r = .99) and SD2 to SDRR (r = .95; Hoshi et al., 2013). Tulppo and colleagues showed a strong correlation between SD1 and HF (r = .94) as well as SD2 and SDRR (r = .99; Tulppo, Makikallio, Takala, Seppanen, & Huikuri, 1996). Our data similarly reproduced these correlations (Table 5).
DFA detects self‐similarity. An α value of 0.5 suggests that the signal is truly random (white‐noise) with larger values suggesting less noise (Brownian motion). When healthy volunteers were sequentially challenged with atropine, propranolol, and clonidine, it was shown that DFA values rise with vagal blockade and decrease with sympathetic blockade (Millar, Cotie, St Amand, McCartney, & Ditor, 2010).
Sample entropy measures regularity or randomness of heart rate variations. Higher values indicate greater irregularity and are commonly a feature of health. During HUT, it is expected that sample entropy decreases and this has been shown to be proportional to the angle of HUT (Porta et al., 2007).
4.3. Other data supporting the validity of using HRV in AF
Measuring HRV in AF is not implausible; however, there remains a marked under appreciation of it. The most basic search on PubMed reveals >18,000 articles containing the words “heart rate variability” but only 402 articles using the combination of “heart rate variability” and “atrial fibrillation”. A selection of the key publications are summarized below, however, from a broader perspective, further work in this field is required especially using the less validated nonlinear techniques.
Just as in SR there is a circadian rhythm of HR, a similar one is found in AF (Bollmann et al., 2006). Following on from this, the prognostic significance of HRV in large populations of SR patients has been widely published and though there is a similar trend in patients with AF, the literature is sparse (Frey et al., 1995; Platonov & Holmqvist, 2011). Yamada and colleagues showed in 107 patients with AF and predominately a preserved left ventricular ejection fraction, that HRV (nonlinear markers only) could predict mortality (Yamada et al., 2000). In the reduced ejection fraction population of MADIT‐II, in a subgroup of patients with AF (n = 68), those with a pNN20 < 87 had a higher mortality (Corino et al., 2015).
Finally in a cohort of 155 patients with heart failure and AF who were enrolled into the Muerte Subita en Insufficiencia Cardiaca (MUSIC) study, only nonlinear HRV parameters were found to be predictors of mortality, sudden cardiac death and heart failure progression.(Cygankiewicz et al., 2015).
Our focus was on whether reactive changes in HRV could be identified in patient with AF after a dynamic challenge. van den Berg et al. (1997) compared the role of vagal activity by using intravenous propanolol (SNS inhibitor) and methylatropine (PNS inhibitor) in 16 patients with chronic AF and 12 healthy men in SR. They demonstrated that though there were significant differences at baseline between the two groups in respect of HRV (SDRR, RMSDD, LF and HF), these parameters changed in patients with AF in a similar direction albeit visually different magnitudes to healthy individuals when subjected to pharmacological sequential autonomic blockade. In a subsequent blinded crossover trial in 60 patients with permanent AF, it was shown that both beta‐blockers and rate limiting calcium channel blockers lower heart rate and time domain parameters (SDRR, RMSDD), whilst beta‐blockers also increased irregularity (sample entropy; Corino et al., 2015). Nagayoshi and colleagues documented RR interval and SDRR in 23 patients (mean age 61 years) in response to tilt, Valsalva, hand grip and showed that the response in patients with AF was similar to those of a historic SR population (Nagayoshi, Janota, Hnatkova, Camm, & Malik, 1997).
5. LIMITATIONS
This is a retrospective study and is exposed to the inherent biases that are common with this design. We have tried to minimize selection bias by sampling consecutive patients. Observer bias was limited as the tilt‐time around which the analysis was performed was based upon what was recorded at the time of the HUT. Ideally we would have wanted to study more patients with AF but we were surprised to find only 31 patients in total at two centers spanning in combination, 9 years of data in total. Our findings are applicable to patients with permanent AF who are above the age of 60 and likely to be on heart rate lowering or blood pressure lowering medications. Drugs, duration of AF (often difficult accurately to ascertain if the condition is asymptomatic), and other diagnoses (hypertension, heart failure, diabetes mellitus) are all known to induce autonomic remodeling and it is likely to account for the heterogeneity in response to HUT in our study. However, due to our limited sample size of patients with AF, we are unable confidently to perform further subgroup analyses to investigate the relative contributions of each of these explanatory variables on HRV. Comparisons of HRV and response to HUT between the population in AF and SR are confounded by the significantly increased use of cardiovascular medications and prevalence of hypertension in the AF group (Table 1). A larger and prospective study, with a broad spectrum of AF patients matched by an equally broad spectrum of SR patients, might overcome a number of these problems. Others have analyzed blood pressure variability in patients with AF and found less “white‐noise” artifact when compared to spectral analysis of heart rate (Corino, Lombardi, & Mainardi, 2014). Future work may also study blood pressure variability and baroreceptor function as markers of the ANS in patients with AF.
We corrected for multiple statistical testing using a Bonferroni correction. However, an accepted weakness is that it often results in a reduction in power, i.e. a conclusion that there is no change, when one genuinely exists. To demonstrate the effects of this correction explicitly, we have also provided those results that achieved significance at the conventional critical p value of 0.05.
6. CONCLUSION
Our findings confirm the feasibility of using HRV in patients with AF. In particular, we were able to detect changes in HRV in response to HUT using nonlinear methods (DFA‐α2 and sample entropy) as compared to traditional linear methods in individuals with AF. However, this finding invites a larger multicenter validation study. These findings may also prove to be of value in assessing the effect of novel ANS‐modulating treatments such as renal denervation for diseases that predispose to AF, such as heart failure or hypertension.
CONFLICT OF INTEREST
The authors declare that they have no conflict of interest.
ACKNOWLEDGMENTS
HP, CH, ARL, and CDM are supported by the NIHR Cardiovascular Biomedical Research Unit.
Patel HC, Hayward C, Wardle AJ, et al. The effect of head‐up tilt upon markers of heart rate variability in patients with atrial fibrillation. Ann Noninvasive Electrocardiol. 2018;23:e12511 10.1111/anec.12511
REFERENCES
- Ardell, J. L. , Cardinal, R. , Beaumont, E. , Vermeulen, M. , Smith, F. M. , & Andrew Armour, J. (2014). Chronic spinal cord stimulation modifies intrinsic cardiac synaptic efficacy in the suppression of atrial fibrillation. Autonomic Neuroscience, 186, 38–44. 10.1016/j.autneu.2014.09.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bilchick, K. C. , Fetics, B. , Djoukeng, R. , Fisher, S. G. , Fletcher, R. D. , Singh, S. N. , & Berger, R. D. (2002). Prognostic value of heart rate variability in chronic congestive heart failure (Veterans Affairs’ Survival Trial of Antiarrhythmic Therapy in Congestive Heart Failure). American Journal of Cardiology, 90(1), 24–28. [DOI] [PubMed] [Google Scholar]
- Bollmann, A. , Husser, D. , Mainardi, L. , Lombardi, F. , Langley, P. , Murray, A. , & Sornmo, L. (2006). Analysis of surface electrocardiograms in atrial fibrillation: Techniques, research, and clinical applications. Europace, 8(11), 911–926. 10.1093/europace/eul113 [DOI] [PubMed] [Google Scholar]
- Carrara, M. , Carozzi, L. , Moss, T. J. , de Pasquale, M. , Cerutti, S. , Ferrario, M. , & Moorman, J. R. (2015). Heart rate dynamics distinguish among atrial fibrillation, normal sinus rhythm and sinus rhythm with frequent ectopy. Physiological Measurement, 36(9), 1873–1888. 10.1088/0967-3334/36/9/1873 [DOI] [PubMed] [Google Scholar]
- Castiglioni, P. , Parati, G. , Di Rienzo, M. , Carabalona, R. , Cividjian, A. , & Quintin, L. (2011). Scale exponents of blood pressure and heart rate during autonomic blockade as assessed by detrended fluctuation analysis. Journal of Physiology, 589(Pt 2), 355–369. 10.1113/jphysiol.2010.196428 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Corino, V. D. , Cygankiewicz, I. , Mainardi, L. T. , Stridh, M. , Vasquez, R. , Bayes de Luna, A. , & Platonov, P. G. (2013). Association between atrial fibrillatory rate and heart rate variability in patients with atrial fibrillation and congestive heart failure. Annals of Noninvasive Electrocardiology, 18(1), 41–50. 10.1111/anec.12019 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Corino, V. D. A. , Lombardi, F. , & Mainardi, L. T. (2014). Blood pressure variability in patients with atrial fibrillation. Autonomic Neuroscience, 185, 129–133. 10.1016/j.autneu.2014.08.002 [DOI] [PubMed] [Google Scholar]
- Corino, V. D. , Ulimoen, S. R. , Enger, S. , Mainardi, L. T. , Tveit, A. , & Platonov, P. G. (2015). Rate‐control drugs affect variability and irregularity measures of RR intervals in patients with permanent atrial fibrillation. Journal of Cardiovascular Electrophysiology, 26(2), 137–141. 10.1111/jce.12580 [DOI] [PubMed] [Google Scholar]
- Cygankiewicz, I. , Corino, V. , Vazquez, R. , Bayes‐Genis, A. , Mainardi, L. , Zareba, W. , & Platonov, P. G. (2015). Reduced irregularity of ventricular response during atrial fibrillation and long‐term outcome in patients with heart failure. American Journal of Cardiology, 116(7), 1071–1075. 10.1016/j.amjcard.2015.06.043 [DOI] [PubMed] [Google Scholar]
- Frey, B. , Heinz, G. , Binder, T. , Wutte, M. , Schneider, B. , Schmidinger, H. , & Pacher, R. (1995). Diurnal variation of ventricular response to atrial fibrillation in patients with advanced heart failure. American Heart Journal, 129(1), 58–65. [DOI] [PubMed] [Google Scholar]
- Hayano, J. , Sakata, S. , Okada, A. , Mukai, S. , & Fujinami, T. (1998). Circadian rhythms of atrioventricular conduction properties in chronic atrial fibrillation with and without heart failure. Journal of the American College of Cardiology, 31(1), 158–166. [DOI] [PubMed] [Google Scholar]
- Hayano, J. , Yamasaki, F. , Sakata, S. , Okada, A. , Mukai, S. , & Fujinami, T. (1997). Spectral characteristics of ventricular response to atrial fibrillation. American Journal of Physiology, 273(6 Pt 2), H2811–H2816. [DOI] [PubMed] [Google Scholar]
- Hillebrand, S. , Gast, K. B. , de Mutsert, R. , Swenne, C. A. , Jukema, J. W. , Middeldorp, S. , & Dekkers, O. M. (2013). Heart rate variability and first cardiovascular event in populations without known cardiovascular disease: Meta‐analysis and dose‐response meta‐regression. Europace, 15(5), 742–749. 10.1093/europace/eus341 [DOI] [PubMed] [Google Scholar]
- Hoshi, R. A. , Pastre, C. M. , Vanderlei, L. C. , & Godoy, M. F. (2013). Poincare plot indexes of heart rate variability: Relationships with other nonlinear variables. Autonomic Neuroscience, 177(2), 271–274. 10.1016/j.autneu.2013.05.004 [DOI] [PubMed] [Google Scholar]
- Laitinen, T. , Niskanen, L. , Geelen, G. , Lansimies, E. , & Hartikainen, J. (2004). Age dependency of cardiovascular autonomic responses to head‐up tilt in healthy subjects. Journal of Applied Physiology, 96(6), 2333–2340. 10.1152/japplphysiol.00444.2003 [DOI] [PubMed] [Google Scholar]
- Leung, R. S. , Bowman, M. E. , Diep, T. M. , Lorenzi‐Filho, G. , Floras, J. S. , & Bradley, T. D. (2005). Influence of Cheyne‐Stokes respiration on ventricular response to atrial fibrillation in heart failure. Journal of Applied Physiology, 99(5), 1689–1696. 10.1152/japplphysiol.00027.2005 [DOI] [PubMed] [Google Scholar]
- Lim, P. B. , Malcolme‐Lawes, L. C. , Stuber, T. , Koa‐Wing, M. , Wright, I. J. , Tillin, T. , & Kanagaratnam, P. (2011). Feasibility of multiple short, 40‐s, intra‐procedural ECG recordings to detect immediate changes in heart rate variability during catheter ablation for arrhythmias. Journal of Interventional Cardiac Electrophysiology, 32(2), 163–171. 10.1007/s10840-011-9580-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mehlsen, J. , Kaijer, M. N. , & Mehlsen, A. B. (2008). Autonomic and electrocardiographic changes in cardioinhibitory syncope. Europace, 10(1), 91–95. 10.1093/europace/eum237 [DOI] [PubMed] [Google Scholar]
- Millar, P. J. , Cotie, L. M. , St Amand, T. , McCartney, N. , & Ditor, D. S. (2010). Effects of autonomic blockade on nonlinear heart rate dynamics. Clinical Autonomic Research, 20(4), 241–247. 10.1007/s10286-010-0058-6 [DOI] [PubMed] [Google Scholar]
- Mourot, L. , Bouhaddi, M. , Gandelin, E. , Cappelle, S. , Nguyen, N. U. , Wolf, J.‐P. , & Regnard, J. (2007). Conditions of autonomic reciprocal interplay versus autonomic co‐activation: Effects on non‐linear heart rate dynamics. Autonomic Neuroscience, 137(1–2), 27–36. 10.1016/j.autneu.2007.06.284 [DOI] [PubMed] [Google Scholar]
- Nagayoshi, H. , Janota, T. , Hnatkova, K. , Camm, A. J. , & Malik, M. (1997). Autonomic modulation of ventricular rate in atrial fibrillation. American Journal of Physiology, 272(4 Pt 2), H1643–H1649. [DOI] [PubMed] [Google Scholar]
- Patel, H. C. , Rosen, S. D. , Lindsay, A. , Hayward, C. , Lyon, A. R. , & di Mario, C. (2013). Targeting the autonomic nervous system: Measuring autonomic function and novel devices for heart failure management. International Journal of Cardiology, 170(2), 107–117. 10.1016/j.ijcard.2013.10.058 [DOI] [PubMed] [Google Scholar]
- Petersen, M. E. , Williams, T. R. , Gordon, C. , Chamberlain‐Webber, R. , & Sutton, R. (2000). The normal response to prolonged passive head up tilt testing. Heart, 84(5), 509–514. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Platonov, P. G. , & Holmqvist, F. (2011). Atrial fibrillatory rate and irregularity of ventricular response as predictors of clinical outcome in patients with atrial fibrillation. Journal of Electrocardiology, 44(6), 673–677. 10.1016/j.jelectrocard.2011.07.024 [DOI] [PubMed] [Google Scholar]
- Porta, A. , Gnecchi‐Ruscone, T. , Tobaldini, E. , Guzzetti, S. , Furlan, R. , & Montano, N. (2007). Progressive decrease of heart period variability entropy‐based complexity during graded head‐up tilt. Journal of Applied Physiology, 103(4), 1143–1149. 10.1152/japplphysiol.00293.2007 [DOI] [PubMed] [Google Scholar]
- Rawles, J. M. , & Rowland, E. (1986). Is the pulse in atrial fibrillation irregularly irregular? British Heart Journal, 56(1), 4–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Shiogai, Y. , Stefanovska, A. , & McClintock, P. V. E. (2010). Nonlinear dynamics of cardiovascular ageing. Physics Reports, 488(2–3), 51–110. 10.1016/j.physrep.2009.12.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sosnowski, M. , Macfarlane, P. W. , & Tendera, M. (2011). Determinants of a reduced heart rate variability in chronic atrial fibrillation. Annals of Noninvasive Electrocardiology, 16(4), 321–326. 10.1111/j.1542-474X.2011.00458.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tarvainen, M. P. , Niskanen, J. P. , Lipponen, J. A. , Ranta‐Aho, P. O. , & Karjalainen, P. A. (2014). Kubios HRV–heart rate variability analysis software. Computer Methods and Programs in Biomedicine, 113(1), 210–220. 10.1016/j.cmpb.2013.07.024 [DOI] [PubMed] [Google Scholar]
- Task Force of the European Society of Cardiology, and the North American Society of Pacing and Electrophysiology (1996). Heart rate variability. Standards of measurement, physiological interpretation, and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. European Heart Journal, 17(3), 354–381. [PubMed] [Google Scholar]
- Tulppo, M. P. , Makikallio, T. H. , Takala, T. E. , Seppanen, T. , & Huikuri, H. V. (1996). Quantitative beat‐to‐beat analysis of heart rate dynamics during exercise. American Journal of Physiology, 271(1 Pt 2), H244–H252. [DOI] [PubMed] [Google Scholar]
- van den Berg, M. P. , Haaksma, J. , Brouwer, J. , Tieleman, R. G. , Mulder, G. , & Crijns, H. J. (1997). Heart rate variability in patients with atrial fibrillation is related to vagal tone. Circulation, 96(4), 1209–1216. [DOI] [PubMed] [Google Scholar]
- Voss, A. , Schroeder, R. , Heitmann, A. , Peters, A. , & Perz, S. (2015). Short‐term heart rate variability—influence of gender and age in healthy subjects. PLoS One, 10(3), e0118308 10.1371/journal.pone.0118308 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yamada, A. , Hayano, J. , Sakata, S. , Okada, A. , Mukai, S. , Ohte, N. , & Kimura, G. (2000). Reduced ventricular response irregularity is associated with increased mortality in patients with chronic atrial fibrillation. Circulation, 102(3), 300–306. [DOI] [PubMed] [Google Scholar]