QT dispersion and RR variations on 12-lead ECGs in patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy

European Heart Journal (1996) 17, 258-263 QT dispersion and RR variations on 12-lead ECGs in patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy L. Fei, J. H. Goldman, K. Prasad, P. J. Keeling, K. Reardon, A. J. Camm and W. J. McKenna Department of Cardiological Sciences, St George's Hospital Medical School, Cranmer Terrace, London SW17 ORE, U.K. Increased QT dispersion, which has been proposed as a marker of ventricular repolarization inhomogeneity, may predispose to ventricular arrhythmias. Data on QT dispersion in patients with congestive heart failure are scarce. In this study, conventional 12-lead ECGs were recorded in 135 consecutive patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy. Seventy-five patients were excluded from QT interval assessments due to one or more of the following reasons: (1) low amplitude of the T wave (n = 3), (2) atrial fibrillation (n=26) and (3) bundle branch block (n=46). QT dispersion was calculated as (1) QT-range: the difference between the maximum and minimum QT intervals on any of the 12 leads and (2) QT-SD: the standard deviation of the QT interval in all the 12 leads. RR intervals were measured in leads II, avl, V 2 and V 5. QT-SD (20-85 ± 500 ms) was significantly (r=0-8997, P<0001) related to QT-range (65-65 ± 15-77 ms), but not to the QT interval. Neither QT-range nor QT-SD was significantly related to age, left ventricular dimensions, left ventricular end diastolic pressure, left ventricular ejection fraction or left ventricular wall thickness. There was no significant difference in QT dispersion between survivors and those who died (n = 8) or were transplanted (n = 9) during 34 ± 23 month follow-up. No significant difference in QT dispersion was observed between patients with and without ventricular tachycardia (;> three consecutive beats) detected on 24-h Holter ECGs. RR interval variation was significantly lower in patients who died compared with survivors (standard deviation: 10-37 ±361 vs 3602 ± 3503 ms, /><0001; coefficient of variance: 1-87 ±0-7% vs 4-50 ±4-9%, />=0001). This was also true in patients with bundle branch block. These observations suggest that QT dispersion in idiopathic dilated cardiomyopathy is not significantly related to either QT interval or cardiac size and function and does not predict death. The application of QT dispersion assessment is limited by the commonly encountered atrial fibrillation and bundle branch block in this patient population. However, reduced RR variation on standard 12-lead ECGs has important prognostic implications in these patients. (Eur Heart J 1996; 17: 258-263) Key Words: Congestive heart failure, heart rate variability, QT interval. Introduction The QT interval has long been known to vary significantly between the individual leads of the 12-lead electrocardiogram (ECG) 1 ' 1, but it was not until recently that a potential application of this interlead difference was proposed by Day et alp\ It is currently thought that the interlead difference in QT interval may provide a measure of repolarization inhomogeneity 131, which may represent an electrophysiological substrate for ventricular arrhythmias 1 '. Increased non-uniform recovery time may result either from inhomogeneity of Revision submitted 16 May 1995, and accepted 22 June 1995. Correspondence: Dr Lu Fei, Krannert Institute of Cardiology, 1111 West 10th Street, Indianapolis, IN 46202^800, U.S.A. repolarization duration or from localized delay in activation. QT dispersion has been shown to be increased in a number of cardiac diseases, and increased QT dispersion may predispose to arrhythmic events 131. Several studies on the influence of antiarrhythmic drugs on QT dispersion suggest that the reduction in the inhomogeneity of ventricular repolarization by some class III antiarrhythmic agents may be involved in the antiarrhythmic mechanism of these drugs' 7 ' 81. Conversely, increased QT dispersion caused by drugs in susceptible individuals may be associated with their proarrhythmic effects' 91. However, the predictive value of QT dispersion has been questioned by several investigators' 10 '. Patients with congestive heart failure are at risk of sudden cardiac death, but identification of those at risk of sudden cardiac death remains problematic. 0195-668X/96/020258+06 $18.00/0 19% The European Society of Cardiology

QT dispersion in heart failure 259 Table 1 Clinical characteristics of patients bundle branch block (n=60) Age Male/Female Patients with symptoms Chest pain Palpitation Syncope and/or presyncope NYHA functional classes Ventricular ectopic frequency Left atrial dimension Left ventricular end diastolic dimension Left ventricular end systolic dimension Maximum left ventricular wall thickness Left ventricular ejection fraction Pulmonary artery wedge pressure Left ventricular end diastolic pressure Patients on amiodarone 40 ± 14 years 47/13 NYHA = New York Heart Association classification. 11 17 9 11 = 29, 111=21, IV=10 2197 ±3316 beats. day"' 43 ±9 mm 67 ± 10 mm 56 ± 12 mm 10 ±2 mm 28 ± 10% 20 ± lommhg 19 ± 11 mmhg 8 Available postinfarction risk stratification techniques are not helpful in these patients. Decreased 24 h heart rate variability has been shown to be associated with increased risk of mortality in heart failure patients waiting for heart transplantation 111], but fails to predict those at risk of sudden cardiac death in congestive heart failure 1 ' 21. A recent report demonstrated that increased QT dispersion might identify patients at risk of sudden cardiac death in congestive heart failure 1 ' 31. In this study, QT dispersion on 12 lead ECGs was assessed in patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy in order to evaluate the pathophysiological correlates of QT dispersion and the predictive value of QT dispersion in these patients. The predictive value of RR variation on 12 lead ECGs was also assessed. Methods Patients without Measurement of RR and QT intervals All measurements were made on conventional 12 lead ECGs which were recorded at a speed of 25 mm. s ~ ' using a Hewlett Packard electrocardiographic recorder. The QT interval (defined as the interval between the beginning of QRS complex and the end of the T wave) was measured in all 12 leads by a physician (K.P.) who was unaware of clinical conditions at the time of QT measurements. The end of the T wave was defined as the intersection of the isoelectric line and the tangent of the maximal slope on the downward limb of the T wave 1 ' 61. QT dispersion was calculated as (1) QT-range: the difference between the maximum and minimum QT intervals in any of the 12 leads and (2) QT-SD: the standard deviation of the QT interval in all of the 12 leads. One sinus RR interval immediately preceding the QT interval was measured in leads II, avl, V 2 and V 5, respectively. Heart rate was derived from the mean of the RR intervals. RR variation was calculated as the standard deviation of the RR intervals (RRSD) and the coefficient of variance, i.e. standard deviation divided by mean RR interval and expressed in percentage (RRCV). The beat chosen for QT/RR measurements was not immediately after ectopic beats. Follow-up All patients with congestive heart failure were followed up in a heart failure clinic at least every 3 months. The cause of patient death was determined by direct communication with the patient families, or their general practitioners, or from hospital records. Every effort was made to discriminate sudden death from pump failure death. As in the Cardiac Arrhythmia Pilot Study (CAPS)' 171, sudden cardiac death in this study was defined as death within 1 h of the onset of new symptoms. The definition also included instantaneous death, death during sleep as well as unwitnessed death that occurred within 1 h of the patients last being seen alive. The study patient population consisted of 135 consecutive patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy who presented to St George's Hospital. The clinical diagnosis of dilated cardiomyopathy was made according to criteria as recommended by the World Health Organization and the National Heart, Lung and Blood Institute 1 ' 4-151. Clinical investigations included 12 lead electrocardiography, chest radiography, two-dimensional Doppler echocardiography and 24 h Holter electrocardiography. Left ventricular ejection fraction was determined by radionuclide ventriculography. Patients with bundle branch block (n=46), atrial fibrillation (n = 26) and low amplitude of the T wave (n = 3) were excluded from the assessment of ventricular repolarization using the QT interval. The clinical characteristics of the remaining 60 patients are summarized in Table 1. Statistical analysis All data were expressed as mean ± standard deviation. Student's t-test, Chi-square test and multiple linear correlation and regression analyses were used where appropriate. A two-tailed P value <0-05 was considered statistically significant. Results In the 60 patients without bundle branch block, the mean RR interval was 74450 ± 228-88 ms and the mean QT interval was 380-38 ± 5540 ms. Using Bazett's formula, the correct QT interval for mean RR intervals was 447-20 ± 3100 ms" 2. QT-range and QT-SD were

260 L. Fei et al. 65-65 ± 15-77 and 20-48 ± 5-00 ms, respectively. RRSD and RRCV were 28-76 ± 31-83 ms and 3-75 ±4-34%, respectively. Correlation of QT dispersion with other clinical variables QT-SD was significantly (r = 0-8997, P<0001) related to QT-range, but not to mean, minimal and maximal values of QT intervals. Neither QT-range nor QT-SD was significantly related to age, left atrial dimension, left ventricular dimensions, left ventricular end-diastolic pressure, pulmonary artery wedge pressure, left ventricular ejection fraction, left ventricular wall thickness or ventricular premature complex frequency (ln[vpc+l]). The QT interval was not significantly related to the above variables except pulmonary artery wedge pressure (mean QT: r= - 0-59, minimal QT: r = 0-60; maximal QT: r= - 0-61; n = 28, />=0-001 for all the correlation coefficients between the QT interval and pulmonary artery wedge pressure). No significant relationship between QT dispersion and RR intervals was found. There was no significant difference in QT dispersion between patients with and without non-sustained ventricular tachycardia (QT-range: 62-19 ±11-97 vs 7000 ±17-54 ms, />=0135; QT-SD: 19-39 ±4-27 vs 22-08 ± 5-60 ms, P=0119). There is a non-significant increase in QT-SD in patients treated with amiodarone (n = 8) compared with those not treated with amiodarone (23-75 ±4-58 vs 19-97±4-90, / > =O058), but not QTrange (70-00 ± 11-95 vs 64-98 ± 16-26 ms, />=0-316). Predictive value of QT dispersion During 34 ± 23 months of follow-up in the 60 patients with congestive heart failure, eight died (including two from sudden cardiac death and six from pump failure death) and nine underwent heart transplantation (progressive heart failure). Death at 1, 1-5 and 2 years accounted for 47%, 76% and 94% of all-cause mortality, respectively. There was no significant difference in QT intervals, QT dispersion and left ventricular ejection fraction between survivors and those who were dead and/or transplanted (Table 2). Neither QT-range (6000 ± 14-22 ms) nor QT-SD (20-34 ± 400 ms) in the two patients who died suddenly was significantly different from that in survivors or from that in those who died from pump failure. Heart rate and RR variations In the absence of bundle branch block, RR interval variation on the 12 lead ECGs, in contrast to QT dispersion, was significantly lower in patients who died compared with survivors (RRSD: 10-37 ±3-61 vs 3602 ± 3503 ms, /><0-001; RRCV: 1-87 ± 0-70% vs 4-50 ±4-90%, / > =0001) (Fig. 1). Figure 2 shows Kaplan-Meier cumulative survival curves for the RR Table 2 QT intervals and QT dispersion in patients with idiopathic dilated cardiomyopathy No. of patients Age (years) LA (mm) LVDD (mm) LVSD (mm) LVWT (mm) WP (mm) LVEDP (mmhg) LVEF (%) FU length (months) Ln(VPC/day+l) Mean RR (ms) Mean QT (ms) QTc (ms" 2 ) QT-SD (ms) QT-range (ms) 43 39 ± 15 41-84 ±8-59 64-90 ±9-84 54-02 ± 11-60 10-34 ±2-58 16-80 ±8-53 15-35 ± 814 28-80 ±9-58 41 ±21 5±3 819±219 397 ± 50 443 ±33 20-85 ±5-28 67-09 ± 16-25 All cause death 17 43 ± 14 48-80 ±9-41 72-14 ±8-87 62-93 ± 12-46 9-86 ±0 38 26-88 ± 9-20 2711 ± 1218 27-67 ±14-68 I4± 12 7±4 556 ±120 339 ±46 456 ± 27 19-52±4-17 62-00 ± 14-22 P value 0-363 0054 0017 0029 0 337 0020 0022 0-863 0-000 0-516 0-000 0-000 0128 0-312 0-239 FU = follow-up; LA = left atrial cavity dimension; LVDD = left ventricular diastolic dimension; LVEDP=left ventricular end diastolic pressure; LVEF=left ventricular ejection fraction; LVSD = left ventricular systolic dimension; LVWT = left ventricular wall thickness; QT-SD = standard deviation of all 12-lead QT intervals; VPC=ventricular premature complex; WP=mean pulmonary artery wedge pressure. variation on the 12 lead ECGs using all-cause death as the end point. The predictive values of RR variations also exited in the presence of bundle branch block (Table 3). In the 46 patients with bundle branch block, most of them had left bundle branch block morphology (n=40), but there was no significant difference in allcause mortality between patients with left and right bundle branch block morphology (P= 1 00). The 2-year mortality rate, however, was significantly higher in patients with bundle branch block compared with those without (50% vs 27%, P=0-019), suggesting that the presence of this condition is associated with increased mortality in these patients. We noted that heart rate was significantly higher in patients who died than survivors, but its predictive value was significantly lower compared with RR variations (Table 4). Furthermore, heart rate on 12 lead ECGs was significantly related to that derived from 24 h Holter recordings (r = 0-78, / > <0001). The correlations for RRSD and RRCV, however, were much weaker (RRSD: r=0-39, />=0-046; RRCV:r=014, i>=0-479). Discussion QT dispersion in normal subjects and in patients with congestive heart failure Data in most recent studies suggest that QT dispersion is usually between 20 and 50 ms in normal subjects' 41 and between 60 and 80 ms in patients following myocardial infarction' 31. The QT dispersion measured by multiple Eur Heart J. Vol 17. February 1996

QT dispersion in heart failure 261 100 1000 p< 0.001 p = 0.001 100-10- Q as as 10-1 - 0.1 Death Death Figure 1 Scattergram of the standard deviation (RRSD) and coefficient of variance (RRCV) of RR intervals as measured on standard 12-lead ECGs in patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy (n=60). Death is all cause death (n = 17). Horizontal bar indicates the mean values. Table 3 RR variations and other clinical variables in patients with bundle branch block 1.0 RRSD > 15 ms 10 20 30 40 Follow-up (months) 50 60 Figure 2 Kaplan-Meier survival curve for the standard deviation of RR intervals (RRSD) measured on the 12lead ECGs in 60 patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy. All cause death was used as the end point (n = 17). Dichotomized at 15 ms, RRSD strongly predicts all cause mortality with sensitivity of 94%, specificity of 77% and positive predictive accuracy of 62%. site mapping tends to be bigger (34 to 88 ms) than that on standard 12 lead ECGt' ). We have demonstrated that measurement of QT interval duration and QT dispersion in healthy volunteers are reproducible within a short time period although the reproducibility is relatively poor1'61. QT-range in the present study was 66 ± 16 ms. This is higher than the normal values (from 20 to 50 ms). Pye et a/.'191 demonstrated increased QT dispersion in patients with arrhythmias and left ventricular dysfunction compared with patients with arrhythmias and good left ventricular function. However, a significant correlation between QT dispersion and frequency of ventricular extrasystole or the presence of nonsustained ventricular tachycardia on Holter monitoring was neither demonstrated by us in this study nor by All cause death P value 18 46 ± 17 43-78 ± 9-20 68-40 ±10-93 59-20 ± 13 54 11-25 ±3-58 18-00± 1111 23-43 ±11-31 25-60 ±4-72 50 ±21 716±152 20-76 ±12-61 2-86 ± 1-37 28* 49 ± 7 46-20 ± 11 06 74-55 ± 14-89 65-75 ±17 04 1017 ±2-62 22-22 ± 12-39 21-12 ±7-84 20-92 ± 8-68 18 ±20 682±141 1209 ± 7 0 2 1-73 ±0-96 0-607 0-587 0-258 0-299 0-444 0-539 0-577 0-267 0000 0-526 0011 0007 Including one patient who died suddenly. Abbreviations are the same as those in Table 2. Table 4 Predictive values of heart rate (HR) and RR variations for all cause death Sensitivity Specificity PPA RRSD (dichotomized at 15 ms) HR (dichotomized at 80 beats, min" 1 ) 77-78% 73-58% 71-43% 47-83% 19-35% 56-90% PPA = positive predictive accuracy; RRSD = standard deviation of the RR intervals measured on conventional 12 lead ECGs. Davey el al.[20]. Increased QTc dispersion has been reported in patients with left ventricular failure who later died suddenly but not in those dying from progressive pump failure compared with survivors'20'. We were unable to reach a conclusion regarding the difference in RRSD < 15 ms No. of patients Age (years) LA (mm) LVDD (mm) LVSD (mm) LVWT (mm) WP (mm) LVEDP (mmhg) LVEF (%) FU length (months) Mean RR (ms) RRSD (ms) RRCV (%)

262 L. Fei et al. QT dispersion between patients who died suddenly and survivors due to a small number (n = 2) of patients who died suddenly in this study. Pathophysiological correlates In this study, there was no significant relationship between mean QT intervals and QT dispersion. This is consistent with the observation that they do not always occur together, such as the class III antiarrhythmic agent induced changes in QT intervals and QT dispersion. They may increase QT interval duration without increasing QT dispersion 17-2 ' 1. A significant but relatively poor correlation (r=0-56, P<0-01) between QT dispersion and left ventricular ejection fraction has been demonstrated in patients with ventricular arrhythmias 1 ' 9 ', but we failed to demonstrated a significant relationship between QT dispersion and other clinical variables, including cardiac size and function. The main determinant of QT dispersion in congestive heart failure remains to be identified. Considering the influence of complex pathophysiological changes in congestive heart failure, our data did not preclude a possible relationship between QT dispersion and other clinical correlates in patients without significant left ventricular dysfunction. We noted that there was a significant relationship between the QT interval and pulmonary artery wedge pressure. The physiological significance of this relationship remains unclear. Methodological considerations QT dispersion has been initially defined as the difference between the maximum and minimum QT intervals measured from each of the 12 standard ECG leads (QT-range). However, the standard deviation of QT intervals is obviously more appropriate than QT-range from a statistical viewpoint, since it incorporates values from all leads instead of only two. It is not surprising to find a good correlation between QT-range and QT-SD in our study. An important issue which needs to be emphasized is whether it is necessary to correct QT dispersion for heart rate. Most of the published studies on QT dispersion are performed on conventional 12 lead ECGs, on which QT intervals cannot be measured from simultaneous recordings of all 12 leads. QTc dispersion has been used in an attempt to adjust for the use of non-simultaneous 12 lead recordings by taking into account the beat to beat variation in RR interval. It is now well accepted that correction of the QT interval for heart rate is potentially misleading under certain circumstances' 221. Hysteresis in the QT interval adaption to the change of heart rate within a period of a few beats makes the correction for heart rate less important, particularly compared with the errors in measurement of QT dispersion. This viewpoint is supported by the study of Zabel et a/.' 231 who demonstrated that dispersion of repolarization does not increase to the same extent with longer cycle length as does the duration of repolarization. We have recently demonstrated that there was no significant beat to beat variations in QT dispersion measured on simultaneous recorded 12 lead ECGs' 241. In our study, there was no significant relationship between QT dispersion and RR intervals. This is also true in simultaneous 12 lead electrocardiographic recordings' 24 '. In contrast, the RR interval on the 12 lead ECGs in this study was found to vary considerably compared with QT interval variation and to contain significant prognostic information on its own. This further suggests the limitations of correction of the QT interval and QT dispersion for heart rate. The significant difference in RR variability between survivors and death may significantly contribute to the magnitude of difference in QTc dispersion observed in the study of Barr et a/.' 13 '. Hence, we did not correct the QT interval for heart rate in our study. It has been suggested that JT interval dispersion was a better predictor of sudden cardiac death in patients with remote myocardial infarction 1 ' 0 ', but we did not assess it in this study since evaluation of the JT interval has not been generally accepted in daily practice. RR variations in congestive heart failure Decreased heart rate variability has been shown to be a powerful and independent post-infarction risk factor for all-cause death and arrhythmic events (including sudden death and sustained ventricular tachycardia)' 251, and depressed heart rate variability has been reported to be associated with increased mortality in patients with congestive heart failure 1 " 1. However, heart rate variability, like other conventional risk factors, such as left ventricular ejection fraction, does not help in identifying those at risk of sudden cardiac death' 12 '. The predictive value of heart rate variability is primarily based on data derived from 24 h Holter recordings, which requires laborious manual editing and complex computing software. In this study we demonstrated that RR variation on standard 12 lead ECGs also had significant predictive value despite the poor correlation between heart rate variability computed from 24 h Holter recordings and that derived from a short ECG strip. Its predictive value was significantly higher than the simple measure of heart rate. This suggests the value of heart rate variability analysis in daily practice. Further studies are warranted to validate this observation in other patient populations, such as those following acute myocardial infarction. Conclusions Our data suggest that QT dispersion in patients idiopathic dilated cardiomyopathy is not significantly related to the QT interval, cardiac size and function, and does not differ significantly between survivors and death.

QT dispersion in heart failure 263 The clinical application of QT dispersion assessment is also limited by the commonly encountered atrial fibrillation and bundle branch block. However, RR variation on 12 lead ECGs may be useful for risk stratification in these patients. References [1] Lepeschkin E, Surawicz B. The measurement of the QT interval of the electrocardiogram. Circulation 1952; 6: 378-88. [2] Day CP, McComb JM, Campbell RW. QT dispersion: an indication of arrhythmia risk in patients with long QT intervals. Br Heart J 1990; 63: 342^1. [3] Statters DJ, Malik M, Ward DE, Camm AJ. QT dispersion: problems of methodology and clinical significance. J Cardiovasc Electrophysiol 1994; 5: 672-85. [4] Higham PD, Campbell RW. QT dispersion. Br Heart J 1994; 71: 508-10. [5] Kuo CS, Munakata K, Reddy P, Surawicz B. Characteristics and possible mechanism of ventricular arrhythmias dependent on the dispersion of action potential durations. Circulation 1983; 67: 1356-67. [6] Bonatti V, Rolli A, Botti G. Recording of monophasic action potentials of the right ventricle in long QT syndromes complicated by severe ventricular arrhythmias. Eur Heart J 1983; 4: 168-79. [7] Day CP, McComb JM, Matthews J, Campbell RW. Reduction in QT dispersion by sotalol following myocardial infarction. Eur Heart J 1991; 12: 423-7. [8] Sedgwick ML, Rasmussen HS, Cobbe SM. Effects of the class III antiarrhythmic drug dofetilide on ventricular monophasic action potential duration and QT dispersion in stable angina pectoris. Am J Cardiol 1992; 70: 1432-7. [9] Hii JT, Wyse DG, Gillis AM, Duff HJ, Solylo MA, Mitchell LB. Precordial QT interval dispersion as a marker of torsade de pointes: Disparate effects of class la antiarrhythmic drugs and amiodarone. Circulation 1992; 86: 1376-82. [10] Zareba W, Moss AJ, le Cessie S. Dispersion of ventricular repolanzation and sudden cardiac death in ischemic heart disease. Am J Cardiol 1994; 74: 550-3. [11] Binder T, Frey B, Porenta G et al. Prognostic value of heart rate variability in patients awaiting cardiac transplantation. PACE 1992; 15:2215-20. [12] Lu Fei, Keeling PJ, Gill JS el al. Heart rate variability and its relation to ventricular arrhythmias in patients with congestive heart failure. Br Heart J 1994; 71: 322-8. [13] Barr CS, Naas A, Freeman M, Lang CC, Struthers AD. QT dispersion and sudden unexpected death in chronic heart failure. Lancet 1994; 343: 327-9. [14] Brandenburg RO, Chazov E, Cherian G et al. Report of the WHO/1SFC task force on definition and classification of the cardiomyopathies. Circulation 1981; 64: 437A-438A. [15] Manolio TA, Baughman KL, Rodeheffer R et al. Prevalence and etiology of idiopathic dilated cardiomyopathy (summary of a National Heart, Lung and Blood Institute workshop). Am J Cardiol 1992; 69: 1458-66. [16] Lu Fei, Statters DJ, Camm AJ. QT interval dispersion on 12 lead electrocardiogram in normal subjects. Its reproducibility and relation to the T wave. Am Heart J 1994; 127: 1654-65. [17] Greene HL, Richardson DW, Barker AH et al. Classification of deaths after myocardial infarction as arrhythmic (the Cardiac Arrhythmia Pilot Study). Am J Cardiol 1989; 63: 1-6. [18] Mirvis DM. Spatial variation of QT intervals in normal persons and patients with acute myocardial infarction. J Am Coll Cardiol 1985; 5: 625-31. [19] Pye M, Quinn AC, Cobbe SM. QT interval dispersion: a non-invasive marker of susceptibility to arrhythmia in patients with sustained ventricular arrhythmias? Br Heart J 1994; 71: 511-14. [20] Davey PP, Bateman J, Mulligan IP, Forfar C, Barlow C, Hart G. QT interval dispersion in chronic heart failure and left ventricular hypertrophy: relation to autonomic nervous system and Holter tape abnormalities. Br Heart J 1994; 71: 268-73. [21] Cui G, Sen L, Sager P, Uppal P, Singh BN. Effects of amiodarone, sematilide, and sotalol on QT dispersion. Am J Cardiol 1994; 74: 896-900. [22] Fei L, Statters DJ, Anderson MH, Katritsis D, Camm AJ. Is there a prolonged QT interval in sudden cardiac death survivors with a 'normal' QTc? Am Heart J 1994; 128: 73-6. [23] Zabel M, Killer BS, Woosley RL, Franz MR. Frequencydependent changes in the dispersion of repolanzation and recovery time: Implications for the measurement of QT dispersion (abstract). PACE 1994; 17: 762. [24] Fei L, Sadoul N, Malik M, Camm AJ. QT dispersion on simultaneous 12-lead electrocardiographic recordings in patients with cardiac arrhythmias: Beat-to-beat variations and cycle length dependence (Abstr). PACE 1995; (in press). [25] Camm AJ, Fei L. Risk stratification after myocardial infarction. PACE 1994; 17: 401-16.