OUP user menu

The relationship between high resting heart rate and ventricular arrhythmogenesis in patients referred to ambulatory 24 h electrocardiographic recording

Elsayed Z. Soliman, Mostafa Abd Elsalam, Yabing Li
DOI: http://dx.doi.org/10.1093/europace/eup344 261-265 First published online: 3 November 2009

Abstract

Aims High resting heart rate (HR) has been associated with sudden cardiac death (SCD). This association is not fully explained by the reported association between HR with coronary heart disease (CHD) or left ventricular systolic dysfunction, the major pathological substrates for SCD. Ventricular arrhythmia is the most common antecedent event before SCD. Examining associations between resting HR and ventricular arrhythmogenesis may enhance our understanding of the association between high resting HR and SCD.

Methods and results This study included 867 patients (age 54 ± 5, 57% females) who underwent 24 h ambulatory electrocardiographic (ECG) recording (Holter) in the period from 1998 to 2000. We examined the unadjusted and multivariable-adjusted associations between resting HR with factors involved in ventricular arrhythmogenesis [ventricular late potentials (LPs) detected by signal-averaged ECG, heart rate variability (HRV), and premature ventricular complexes (PVCs)]. Linear regression models were used for continuous outcomes and logistic regression analysis was used for categorical outcomes. The multivariable models included first age and sex, then history of hypertension, diabetes, hypercholesterolaemia, CHD, heart failure, left ventricular ejection fraction (LVEF), smoking, body mass index, the use of anti-arrhythmic drugs, and ST-depression in the 24 h ambulatory ECG recording (Holter) were included in the final models. In the unadjusted and multivariable-adjusted analysis, high resting HR was significantly associated with positive ventricular LPs, depressed HRV indices, and increased prevalence of PVCs/24 h independently from demographic and clinical variables including LVEF, history of CHD, and the presence of ST-depression in Holter (P-value <0.05 in all comparisons and models).

Conclusion High resting HR is independently associated with ventricular arrhythmogenesis, the major cause of SCD. These findings could partially explain the reported association between increased HR and SCD.

  • Resting heart rate
  • Sudden cardiac death
  • Signal-averaged electrocardiogram
  • Holter
  • Heart rate variability

Introduction

Approximately 400 000 sudden cardiac death (SCD) cases occur each year in the USA.1,2 It has been reported that coronary heart disease (CHD) and depressed left ventricular systolic function are the major pathological substrates of SCD, and ventricular arrhythmia is by far the most common antecedent event.3 Most of these deaths occur outside the hospitals. Therefore, using simple clinical markers to identify those at risk of SCD or its pathological basis, CHD and/or ventricular arrhythmia, is important from the prevention perspective. In this context, a high resting heart rate (HR) has been associated with both SCD and CHD.410 The mechanistic relationship between HR and SCD, however, is not clear. Specifically, it is not known whether the association between HR and SCD is based on the association between HR with CHD and depressed left ventricular function or related to an independent association between HR and ventricular arrhythmogenesis.

In this study, we examined associations between high resting HR and ventricular arrhythmogenesis detected by positive ventricular late potentials (LPs) by signal-averaged ECG (SA-ECG), depressed hear rate variability (HRV), and frequency of premature ventricular complexes (PVCs) in a large sample of patients referred to ambulatory 24 h ECG recording (Holter). Better understanding of the association between HR and arrhythmogenesis could enhance our ability to choose a specific SCD preventive strategy in patients at high risk of SCD whether this could be anti-arrhythmic drugs/devices vs. detection and correction of myocardial ischaemia/dysfunction.

Methods

Study population

This study included all patients who had been referred to our Holter unit for 24 h ambulatory ECG recording in the period from 1998 to 2000. Patients with poor-quality resting 12-lead ECG, patients with Holter recording <18 h, or patients with missing left ventricular ejection fraction (LVEF) data were excluded from this study (total excluded = 140) leaving 867 patients for this analysis.

Electrocardiography

Standard 12-lead resting electrocardiogram

Heart rate was determined from a 10 s long strip of lead II in a 12-lead resting ECG that was recorded the same day of Holter recording. All patients were supine and were resting at least 5 min before ECG recording. On the basis of the 10 s recording, the average RR interval (seconds) was calculated: HR = (60 bpm/average RR interval). When a premature ectopic beat was included in the 10 s recording, it was included in the estimate of HR.

Twenty-four-hour ambulatory electrocardiographic recording (Holter)

All patients included in this analysis underwent 24 h ambulatory ECG recording (Holter). This was performed using Oxford Medilog Prima Holter management system (Oxford Medical Instruments, Old Woking, Surrey, UK) including an FD3 solid-state, three-channel recorder and prima Holter analysis software version 7.1 running under Microsoft Windows. The system has capabilities for arrhythmia, ST, SA-ECG, and HRV analyses. All recordings were analysed by a trained cardiologist.

Heart rate variability

Heart rate variability (HRV) indices were obtained from 24 h ambulatory ECG recordings (Holter). Non-sinus-originated beats (supraventricular and ventricular ectopic beats, AV blocks, and periods of atrial fibrillation) and artefacts were initially detected by the software and then checked and confirmed by a cardiologist. Patients with insufficient recordings (recording duration <18 h and sinus beats <50%) were excluded. Five HRV indices were calculated:11 the standard deviation of all filtered RR intervals over the length of the recording (SDNN); mean of the standard deviations of all filtered RR intervals for all 5 min segments of the recording (SDNNi); the standard deviation of the means of all filtered RR intervals for all 5 min segments of all recording (SDANNi); the root mean square of the difference of successive RRs (RMSSD); and the percentage of RR interval that are greater than adjacent RR by 50 ms (PNN50).

Signal-averaged electrocardiogram (SA-ECG)

Time domain SA-ECG was used to detect ventricular LPs. QRS complexes were recorded and analysed by Oxford Medilog system (Oxford Medical Instruments). The system is designed to acquire ECG data from the 24 h ECG recording. QRS complexes in each of the three channels in a predetermined period (5 min) were averaged and filtered. The 40 Hz high pass filter was used in this study. This filter can attenuate only the low-frequency component of the sine wave <40 Hz, whereas higher frequencies are passed. The maximum noise level allowed was 0.7 µV. Ventricular LPs were considered present (positive LPs) if any two of three criteria were met:12 (i) filtered QRS duration (QRSd) >114 ms, (ii) root-mean-square voltage of the last 40 ms of the QRS complex (RMS-40) <20 μV, and (iii) duration of low amplitude (<40 µV) signal of the terminal portion of the QRS (LAS) >38 ms.

Premature ventricular complexes

Premature ventricular complexes were visually confirmed and counted per 24 h recording. Fusion beats were counted as PVCs. Non-sustained ventricular tachycardia (NSVT) defined as three or more consecutive PVCs that last no more than 30 s and terminate spontaneously was found in five patients. The number of PVCs in each of the NSVT episodes was counted towards the total number of PVCs. None of the patients had sustained ventricular tachycardia. To have a uniform comparison of the number of PVCs between groups with different ECG recording durations, we used the PVCs/24 h for comparison. This was obtained by using the formula: [(number of PVCs/duration of recording) × 24].

ST-depression

Significant ST-depression was defined as flat or downward sloping ST-segment shift >0.1 mV in magnitude at the J point that persisted for >1 min. Where there was pre-existing ST-depression, 0.2 mV of additional ST-depression was regarded as a significant change from baseline. Changes in T-wave vector were ignored unless accompanied by the ST-segment changes described.

Other variables

Demographics and history of hypertension, diabetes, CHD, heart failure, the use of anti-arrhythmic drugs, LVEF, body mass index (BMI), and smoking status were obtained from the patients' health records.

Statistical analysis

Resting HR was categorized into quartiles. Analysis of variance and χ2 square for trend were used to compare the results of the continuous variables (mean ± SD) and categorical variables (%) across quartiles of HR. We used logistic regression analysis (for categorical outcomes as the presence of ventricular LPs and ST abnormalities) and linear regression analysis (for all continuous outcomes) to estimate the multivariable-adjusted associations between 1 SD increase in HR with SA-ECG, HRV, PVCs, and ST-depression. Heart rate variability indices and the number of PVCs/24 h were logarithmically transformed in the linear regression analysis, since they showed skewed distribution. However, in the univariate analysis, we used the mean and SD of these values to allow for realistic presentation and comparison with other studies. The first set of the multivariable models was adjusted for demographic variables (age and sex). The final models included the demographic variables in the first set of models plus history of hypertension, diabetes, hypercholesterolaemia, CHD (angina, myocardial infarction, and coronary revascularization), heart failure, LVEF, smoking, the use of anti-arrhythmic drugs, BMI, and ST-depression in the 24 h ECG recording (Holter).

Results

This analysis included 867 patients aged 54.15 ± 5.3 years of whom 57% were females. Patients underwent 24 h ambulatory ECG recording (Holter) with an average recording duration of 22.7 h (SD ± 1.9) The prevalence of hypertension, diabetes, hypercholesterolaemia, CHD, and heart failure were 27, 17, 14, 12, and 4%, respectively. Table 1 shows the characteristics of the study population at the time of the 24 h ambulatory ECG recording (Holter).

View this table:
Table 1

Characteristics of the study population (n = 867)a

Age (years)54.2 ± 5.3
Females57%
History of hypertension234 (27%)
History of diabetes147 (17%)
History of hypercholesterolaemia requiring medications121 (14%)
History of coronary heart disease107 (12%)
History of heart failure39 (4%)
Use of anti-arrhythmic drugs including beta-blockers130 (15%)
Smoking
 Current78 (9%)
 Previous295 (34%)
Body mass index31.6 ± 5.24
Left ventricular ejection fraction52.1 ± 11.4
Mean heart rate in the 12-lead ECG65.6 ± 9.9
  • aVariables are presented as either mean ± SD or n (%).

Table 2 shows the results of the HRV indices, SA-ECG measures, frequency of PVCs/24 h, and ST-depression in patients across quartiles of resting HR. As shown, all of these measures and indices were getting worse with increasing levels of the HR from the first quartile to the fourth quartile (P < 0.001 for trend), i.e. HRV indices were becoming lower, whereas SA-ECG measures, the prevalence of ventricular LPs, the number of PVCs/24 h, and the prevalence of ST-depression were becoming higher with the increase in the levels of resting HR.

View this table:
Table 2

Heart rate variability indices, SA-ECG measures, frequency of PVCs/24 h, and ST-depression across quartiles of resting HR

Heart rate quartiles
<67, n= 21767–77, n= 23278–93, n= 215>93, n= 203
Heart rate variability indices
 SDNN (ms)119.8 ± 28.6115.9 ± 27.4107.1 ± 25.499.4 ± 23.8
 SDNNi (ms)58.4 ± 17.054.2 ± 16.852.9 ± 15.149.3 ± 13.9
 SDANNi (ms)99.9 ± 30.395.5 ± 29.590.7 ± 27.882.7 ± 23.6
 RMSSD (ms)25.2 ± 12.621.7 ± 11.919.4 ± 11.114.9 ± 12.1
 PNN50 (%)4.9 ± 5.94.2 ± 5.23.7 ± 3.92.1 ± 1.9
Signal-averaged ECG measures
 QRSd100.1 ± 10.7103.9 ± 10.8107.5 ± 11.3114.3 ± 11.0
 LAS30.9 ± 8.133.5 ± 8.935.8 ± 9.140.7 ± 13.5
 RMS-4028.5 ± 17.926.1 ± 15.623.7 ± 13.421.8 ± 17.1
 Positive ventricular LPs20 (9%)25 (11%)32 (15%)42 (21%)
Premature ventricular complexes/24 h14 ± 1316 ± 1327 ± 1464 ± 28
ST-depression (%)3 (1%)7 (3%)8 (4%)14 (7%)
  • P < 0.001 for analysis of variance and χ2 trend tests for all continuous and categorical variables.

In an age- and sex-adjusted linear regression model, 1 SD (10 bpm) increase in resting HR was significantly associated with a lower values of all HRV indices and increased number of PVCs/24 h, separately (Table 3). These statistically significant associations persisted after further adjusting for history of hypertension, diabetes, CHD, and heart failure, and also adjusting for the use of anti-arrhythmic drugs, smoking status, LVEF, BMI, and the presence of ST-depression in the 24 h ambulatory ECG recording (Holter) (P-value <0.05 in all models).

View this table:
Table 3

Associations between resting heart (1 SD increasea) and HRV indices in a linear regression analysis

β coefficient and 95% confidence interval
Model 1: age and sexModel 2: Model 1 + history of hypertension, diabetes, coronary heart disease, heart failure, use of anti-arrhythmic drugs, body mass index, left ventricular ejection fraction, and smoking status
Heart rate variability indices
 Log SDNN (ms)−0.19 (−0.10, −0.28)−0.12 (−0.07, −0.17)
 Log SDNNi (ms)−0.17 (−0.09, −0.25)−0.09 (−0.04, −0.14)
 Log SDANNi (ms)−0.21 (−0.13–0.29)−0.13 (−0.07, −0.19)
 Log RMSSD (ms)−0.06 (−0.02, −0.10)−0.03 (−0.01, −0.05)
 Log PNN50−0.03 (−0.01, −0.05)−0.02 (−0.01, −0.04)
Log PVCs/24 h0.41 (0.29, 0.53)0.32 (0.19, 0.55)
  • a1 SD = 10 bpm.

Table 4 shows the odds ratios of positive ventricular LPs and ST-depression in multivariable logistic regression models adjusted first for age and sex (Model 1), and then further adjusting for history of hypertension, diabetes, CHD, heart failure, the use of anti-arrhythmic drugs, smoking status LVEF, and BMI. In the full model of ventricular LPs, ST-depression in the 24 h ambulatory ECG recording (Holter) was also included. One SD increase in resting HR was significantly associated with ventricular LPs [OR (95% CI): 1.09 (1.05, 1.14) and 1.05 (1.02, 1.08) in Models 1 and 2, respectively] and ST-depression [OR (95% CI): 1.25 (1.14, 1.39) and 1.13 (1.05, 1.21) in Models 1 and 2, respectively]. There was no statistically significant interaction between ST-depression and resting HR in models where positive ventricular LP was the outcome.

View this table:
Table 4

Associations between resting HR (1 SD increasea) with ST-depression and positive ventricular LPs in 24 h Holter recording in multivariable logistic regression analysis

Odds ratio and 95% confidence interval
Model 1: age and sexModel 2: Model 1 + history of hypertension, diabetes, coronary heart disease, heart failure, use of anti-arrhythmic drugs, body mass index, left ventricular ejection fraction, and smoking status
Positive ventricular late potentials1.09 (1.05, 1.14)1.05 (1.02, 1.08)
ST-depression1.25 (1.14, 1.39)1.13 (1.05, 1.21)
  • a1 SD = 10 bpm.

Discussion

The key finding of this study is that a high resting HR is associated with the arrhythmogenic substrates (expressed as positive ventricular LPs), the triggers (expressed as increased PVCs), and the predisposing electrophysiological environment (expressed as depressed HRV) of ventricular arrhythmia. In other words, high resting HR is associated with ventricular arrhythmogenesis. Such an association is independent from the association between HR with demographic and clinical variables including CHD or LVEF. From these findings, it could be suggested that the reported association between high resting HR and SCD410 could be partially explained by the independent association of the earlier with ventricular arrhythmias, the major antecedent event before SCD. These findings might also explain the reported protective effect of regular exercise13,14 and HR slowing medications in the primary and/or secondary prevention of cardiovascular disease.1519 The reported associations between life expectancy and HR20 along with our results raise the provocative idea that various means of slowing HR might be useful in prolonging life, or at least preventing dying from SCD in patients with or without CHD.

In this study, because 24 h ambulatory ECG recording (Holter) has a limited value in detecting all events of ventricular arrhythmia which could be too sporadic to be captured on only 24 h recording, we used the association between HR with the factors involved in ventricular arrhythmogenesis (ventricular LPs, HRV, and PVCs) as surrogate markers for the association between the earlier and ventricular arrhythmia. Ventricular LPs result from slowing down of the electrical transmission and losing its homogeneity across the damaged myocardial area.2124 The presence of ventricular LPs represents an anatomical substrate for repeated ventricular arrhythmia.25 Similarly, depressed HRV indices provide the electrophysiological environment that enhances ventricular arrhythmogenesis.11 The past two decades have witnessed growing evidence of a significant relationship between abnormalities in the cardiac autonomic neuropathy, expressed as depressed HRV indices, and SCD and non-SCD.2630 Finally, PVCs provide the triggers of ventricular arrhythmia, which in the presence of appropriate arrhythmogenic substrate and electrophysiological instability can trigger repeated ventricular arrhythmia. Premature ventricular complexes have been always associated with increased mortality either SCD or non-SCD.4

The association between high resting HR with depressed ST segment as a marker of CHD (as shown in the results) could also partially explain the association between HR and SCD. These findings emphasize the importance of detection and treatment of CHD in patients at high risk of SCD. However, based on the results of this study, electrophysiological studies may be warranted in those without documented CHD and still at risk of SCD. The last notion, however, requires further examination of the association between ventricular arrhythmogenesis and high resting HR in a prospective study.

The results of this study should be read in the context of some unavoidable limitations that warrant highlighting. The number of patients with heart failure seems to be very low in this analysis. Therefore, although we adjusted for LVEF in the multivariate analysis, the results have to be interpreted with caution. Because of the inability of this study (and similar cross-sectional studies) to adjust for unrecognized confounders, it is possible that high resting HR is a surrogate for one or more unmeasured variables that may be causally linked to SCD. We tried as much as we could to adjust for all of the plausible potential confounders available to us. Further, this analysis was conducted on patients referred to the Holter unit who might be different from general population. However, those patients referred for ambulatory ECG recording (Holter) represent a group of patients who are at a higher risk for arrhythmic events or unrecognized myocardial ischaemia, in whom risk stratification for SCD and any subsequent preventive strategies would be cost-effective at least compared with general population. Another limitation that may affect the generalizability of our results is that females constituted the majority of this study population (57%). It is well known that female gender shows a higher HR in the general population. We avoided the confounding effect of the gender by adjusting for it in the multivariable. The last limitation is related to the differential ability of ambulatory ECG recording (Holter) to detect arrhythmia and ischaemia which might have resulted in less detection of ST-depression. This might have affected our conclusion that the association between HR and ventricular arrhythmogenesis is independent from CHD. To overcome such a limitation, we included history of CHD and CHD common risk factors in the models to compensate for any possible underestimation of ST-depression.

In conclusion, the results of this study show that there is a significant association between high resting HR and ventricular arrhythmogenesis, an association that is independent from the association between HR and demographic and clinical variables including CHD and LVEF. These findings could partially explain the reported association between high resting HR and SCD which may call for further investigations on the effect of lowering HR as a preventive measure for SCD.

Conflict of interest: none declared.

References

View Abstract