The relationship between QT intervals and mortality in ambulant patients with chronic heart failure

Transcription

1 European Heart Journal (1999) 20, Article No. euhj , available online at on The relationship between QT intervals and mortality in ambulant patients with chronic heart failure The United Kingdom Heart Failure Evaluation and Assessment of Risk Trial (UK-HEART) P. Brooksby*, P. D. Batin, J. Nolan, S. J. Lindsay, R. Andrews*, M. Mullen, W. Baig, A. D. Flapan, R. J. Prescott, J. M. M. Neilson, A. J. Cowley* and K. A. A. Fox *Department of Cardiovascular Medicine, University Hospital, Nottingham; The Yorkshire Heart Centre, Leeds; Department of Cardiology and The Royal Infirmary, Doncaster; University of Edinburgh, Edinburgh, U.K. Aims Mortality in patients with heart failure remains high and is difficult to predict. QT interval parameters on a 12-lead ECG have been shown to predict arrhythmic events in patients with a variety of myocardial diseases. There is some, but not consistent, evidence that QT interval parameters may act as predictors of mortality, in particular sudden death, in patients with heart failure. In an adequately powered prospective study we have studied QT interval parameters in patients with stable chronic heart failure in order to determine whether they are predictive of all-cause mortality or mode of death. Methods and Results Five hundred and fifty-four ambulant outpatients with chronic heart failure were recruited. A 12-lead ECG, chest radiograph, echocardiogram, 24 h ambulatory electrocardiogram and serum for biochemical analysis were obtained at baseline. Patients were followed for days. QT intervals were measured in all leads blinded to patient s characteristics and outcome, were corrected for heart rate, and the maximum QT intervals, and QT dispersion (range of QT intervals) were determined. The same parameters were determined for JT intervals. The primary end-point was all-cause mortality, secondary end-points were sudden cardiac death and death due to progressive heart failure. Multivariate analysis with the Cox s proportional hazards model was used to determine which variables were independently related to outcome. Four hundred and ninety-five patients had analysable ECGs at study entry and of these 71 died during follow-up. The heart rate corrected QT dispersion and maximum QT interval were significant univariate predictors of all-cause mortality (P=0 026 and < respectively), and also of sudden death and progressive heart failure death, but were not related to outcome in the multivariate analysis. The independent predictors of all-cause mortality were cardiothoracic ratio (P=0 0003), creatinine (P=0 0009), heart rate (P=0 007), echocardiographically derived left ventricular end-diastolic dimension (P=0 007) and ventricular couplets on 24 h electrocardiographic monitoring (P=0 015). Conclusion In an adequately powered prospective study none of the QT or JT parameters were shown to be independent predictors of outcome in patients with mild to moderate congestive heart failure. These variables do not therefore add to the prognostic information which can be gained from simple radiographic, biochemical, echocardiographic and Holter data in this group of patients. (Eur Heart J 1999; 20: ) Key Words: Heart failure, congestive, electrocardiology, mortality, death, sudden, cardiac, QT intervals, QT dispersion. See page 1290 for the Editorial comment on this article Revision submitted 20 January 1999, and accepted 3 February Correspondence: Dr P. Brooksby, Department of Cardiology, Pontefract General Infirmary, Pontefract WF8 1PL, U.K X/99/ $18.00/ The European Society of Cardiology

2 1336 P. Brooksby et al. Introduction Chronic heart failure is common [1] and the mortality rate remains high despite the use of therapies proven to improve prognosis [2,3]. Although many markers of a very poor prognosis are already known [4], it remains difficult to accurately predict high and low risk groups in ambulant patients with mild to moderate symptoms and impairment of left ventricular function [5]. This remains an important task as the annual mortality rate for patients with mild to moderate chronic heart failure still exceeds 10% despite treatment with angiotensin converting enzyme inhibitors [2,5]. The two most common mechanisms of death in patients with chronic heart failure are by progressively increasing pump failure or by a sudden, unexpected deterioration [6]. In patients with mild to moderate chronic heart failure sudden death is the more common fatal event; however, current clinical techniques do not accurately predict these events [7]. Among patients fulfilling the clinical criteria for sudden cardiac death the underlying mechanism is often due to malignant ventricular arrhythmias [8,9].A common substrate for these arrhythmias is regional variation of repolarization times within the ventricular myocardium. It has been suggested that inhomogeneity of ventricular repolarization can be detected on a standard surface 12-lead ECG by measuring the QT interval and QT dispersion, which is a measure of the range of QT intervals across all 12 ECG leads [10].QT dispersion has been shown to correlate well with epicardial monophasic action potential recordings for repolarization variation [11]. QT dispersion appears to be predictive of ventricular arrhythmias in patients with ischaemic heart disease [12 15], hypertrophic cardiomyopathy [16] and congenital long QT syndromes [10]. Recently a few small studies have suggested that QT dispersion on a 12-lead ECG may be able to identify a group of chronic heart failure patients at increased risk of sudden arrhythmic death [17,18] ; however, not all studies are in agreement [19,20]. There is little data concerning the predictive power of the QT interval itself in patients with chronic heart failure, although the QT interval has been reported to be an important indicator of a high risk of sudden death in patients with ischaemic heart disease [21 24]. The United Kingdom Heart Failure Evaluation and Assessment of Risk Trial (UK-HEART) was designed to assess the value of non-invasive predictors of mortality in patients with chronic heart failure. We have previously reported our data relating to heart rate variability and mortality [25]. In this analysis we have examined the relationship of QT interval parameters measured on a single 12-lead ECG as independent predictors of all-cause mortality and mode of death, in comparison with known and commonly measured demographic, haemodynamic and biochemical risk factors. Methods Study design, entry criteria and organization The protocol for UK-HEART specified a prospective, multicentre design, with sample size and duration of follow-up based on a power calculation designed to ensure that the relationship between QT interval parameters and mortality would be adequately evaluated. Patient recruitment and data collection were carried out in four U.K. cardiology centres (Leeds, Nottingham, Doncaster and Edinburgh) between 1 December 1993 and 31 April Ambulant outpatients of either sex, aged between 18 and 80 years, with symptoms and objective evidence of chronic heart failure (pulmonary venous congestion, pulmonary oedema or a cardiothoracic ratio >0 55 on at least one chest radiograph, or left ventricular ejection fraction <0 45) were eligible for entry into the trial. All patients were in sinus rhythm at the time of recruitment, and were receiving diuretic therapy. In order to avoid possible confounding effects, patients were excluded if they had chronic renal impairment (creatinine >350 μmol.l 1 ), hypertrophic cardiomyopathy, myocardial infarction within the last 3 months, a history of alcohol abuse, diabetes mellitus or clinical evidence of autonomic neuropathy. QT interval measurements At enrolment a 12-lead ECG was recorded at a paper speed of 25 mm. s 1. Measurements were made using a high-resolution digitizing tablet (Summasketch III; Summagraphics) linked to an IBM compatible PC running software developed locally (Mr S. Freear, Department of Electronic Engineering, University of Leeds). All measurements were made by a single operator unaware at the time of analysis of the patient s demographics, diagnosis and other investigation results. Analysis was not possible in 59 cases, because of either unexpected atrial fibrillation, multiple ectopic beats, severely prolonged PR interval, low T wave voltages, or poor recording quality. QT intervals were measured in all 12 leads if possible, but in at least in six leads. Analysis of 10 or more leads was possible in 87% of ECGs measured. The average of three consecutive beats for each lead was taken wherever possible. The QT interval was measured from the beginning of the QRS complex to the end of the T wave. The end of the T wave was defined as the return of voltage to the isoelectric line. If a U wave was present, measurements were taken to the nadir of the curve or notch between the T and U waves. QT intervals were corrected for heart rate using Bazett s formula (QTc=QT/RR 1/2 ). The QT intervals reported are the maximum QT and QTc found on the 12-lead ECG (QT max and QTc max ). QT dispersion (QTd) and rate corrected QT dispersion (QTcd) were defined as the difference between the maximum and minimum QT and QTc intervals, respectively [10]. The JT interval, which is the time from the end of the QRS complex to

3 The UK-HEART trial 1337 the end of the T wave, was also measured. JT dispersion (JTd) was then calculated in the same way as for QTd. Patients with bundle branch block have previously been excluded from some QTd studies. These patients (n=141) were not excluded from this study, but we elected to perform a subsidiary analysis for patients without bundle branch block to determine whether the presence of an intraventricular conduction defect had a significant effect on the QT data. In patients with bundle branch block no difference between QTd (76 27 ms) and JTd (mean 75 SD 31 ms) was found (paired t-test; P=0 93), as might be expected if the abnormal QRS complex (i.e. ventricular depolarization) significantly influenced QTd. Further, no significant difference was found between the QTd measured in the bundle branch block group (76 27 ms) and patients with normal QRS morphology (71 27 ms; P=0 006), even though the maximum QT interval (QT max ) was significantly longer in the bundle branch block group ( vs ms; P<0 0001). Baseline characteristics At the recruitment visit, clinical data and demographics, NYHA class and current medical therapy were recorded. Serum was sent for biochemical analysis of renal and hepatic function. A transthoracic echocardiogram was performed and left ventricular dimensions measured at end-systole (LVESD) and end-diastole (LVEDD) according to the American Society of Echocardiography recommendations. The cardiothoracic ratio was measured from a PA chest radiograph. Twenty-four hour ambulatory electrocardiograms were recorded using a miniature tape recorder (Tracker, Reynolds Medical Ltd., U.K.). The average number of ventricular ectopic beats per hour and the presence of non-sustained ventricular arrhythmias, as defined in the V-HeFT (Vasodilator Heart Failure Trial) studies [26], was recorded for each ambulatory electrocardiogram. Mortality Patients were registered with the U.K. national death reporting scheme (Office of Population Censuses and Surveys). The primary end-point of this study was all-cause mortality. Precise sub-classification of mortality in chronic heart failure is difficult; however, previous studies have indicated that QT dispersion is only related to sudden death and not to death due to progressive heart failure. We have therefore classified deaths using criteria previously defined in chronic heart failure studies [27,28]. All deaths were classified by two physicians who were unaware of the QT interval analysis and who determined the probable mode of death after review of death certificates, autopsy findings, hospital and general practitioners records. The mode of death was classified as: (a) sudden cardiac death if it occurred within 1 h of a change in symptoms or if it occurred during sleep or while unobserved, if circumstantial evidence pointed to death from cardiovascular causes in the absence of clinical or post-mortem evidence of acute myocardial infarction or increasing heart failure; (b) progressive heart failure if death occurred after a documented period of symptomatic or haemodynamic deterioration; (c) other cardiovascular death if it did not occur suddenly and was not associated with progressively worsening heart failure; (d) non-cardiovascular death. Power calculation The number of patients studied was based upon evidence from previous work which suggested that the expected difference in QTd between survivors and those dying suddenly is around 30 ms [17]. The expected intersubject standard deviation for QTd for the two groups was 30 ms. The estimated mortality rate was 15% per annum and of these approximately 30% were likely to die suddenly. The study therefore required 500 patients to be followed for 1 year to be likely to detect a difference of 30 ms between the groups with a power of 95% and a significance level of 5%. Statistical methods Statistical analyses were undertaken by an independent statistical centre, which did not participate in patient recruitment or data collection. Descriptive group data are presented as mean standard deviation unless otherwise stated. The Cox s proportional hazards regression model [29] was used to determine which variables were related to mortality and mode of death. In order to adjust for the effects of covariates, variables that were independently associated with cardiac mortality were identified using multivariate analysis with both forward and backward stepping to protect against the omission of important predictors. Results A total of 554 patients were recruited between 1 December 1993 and 1 April Of these the 12-lead ECG was not analysable for QT parameters in 59 cases, data were therefore complete in the remaining 495 patients. Patients were followed-up for days, during which time there were 71 deaths (14 3% mortality). No patients received either an implantable defibrillator or cardiac transplant during the follow-up period. Baseline characteristics are detailed in Table 1. The majority of patients were mildly to moderately symptomatic (NYHA class: I 2 2; II 59 7%; III 37 1%; IV 1 1%). Ischaemic heart disease was the most common aetiology of chronic heart failure (76%); other causes were

4 1338 P. Brooksby et al. Table 1 Baseline characteristics of the population (n=495) Variable Mean SD Min Max Follow-up (days) Age (years) LVEDD (cm) LVESD (cm) CTR (%) NYHA class Sodium (mmol. l 1 ) Potassium (mmol. l 1 ) Creatinine (μmol.l 1 ) Albumin (mmol. l 1 ) Heart rate (beats. min 1 ) QTc max (ms) JTc max (ms) QTd (ms) QTcd (ms) JTd (ms) LVEDD=left ventricular end-diastolic diameter; LVESD=left ventricular end-systolic diameter; CTR=cardiothoracic ratio; NYHA=New York Heart Association; QTc=corrected QT intervals; max=maximum; JTc=corrected JT interval; QTd=QT dispersion; QTcd=rate corrected QT dispersion. idiopathic dilated cardiomyopathy (12%), hypertension (4%), valvular heart disease (5%) and mixed aetiology (3%). Bundle branch block was seen in 27% of the ECGs (LBBB 17%, RBBB 10%). The mean frusemide dose for the population was 73 9 mg ( 68 7) and the majority of patients were treated with an angiotensin converting enzyme inhibitor (81%). Other notable treatments for the population were digoxin (18 9%), other antiarrhythmics (chiefly amiodarone (14 6%)), and beta-blockers (8%). Univariate predictors of all-cause mortality The predictive accuracy of variables derived from clinical assessment, chest radiography, serum biochemistry, echocardiograms and ECGs were investigated initially using univariate analysis with respect to all-cause mortality. Ten variables were statistically significantly associated with all-cause mortality, and they are listed in Table 2. Of the QT parameters only QTc max and QTcd were related to mortality. However, none of the nonheart rate corrected QT parameters and none of the JT interval parameters were predictive of mortality. Multivariate predictors of all-cause mortality Multivariate analysis was undertaken in order to determine which variables were independently predictive of all-cause mortality. The five variables, which proved to be significantly related to mortality, are listed in Table 3. None of the QT parameters remained in the model as Table 2 Statistically significant univariate predictors of all cause mortality (mean SD or percentage of the population) Alive Dead P value CTR (%) < Creatinine (μmol.l 1 ) < Sodium (mmol. l 1 ) < QTc max (ms) < LVEDD (cm) HR (beats. min 1 ) Couplets 59% 81% Albumin (mmol. l 1 ) Bundle branch block 25% 39% QTcd (ms) For abbreviations see Table 1. independent predictors. The rate corrected QT parameters QTc max and QTcd lost their statistical significance when heart rate was entered into the model. This indicates that there was a significant component of heart rate, which allowed the parameters QTc max and QTcd to be univariate predictors. The significant correlations between QTc max and QTcd and other variables entered into the model are shown in Table 4. Apart from BBB and heart rate with QTc max associations that one would expect, there were no strong correlates with either of these variables. As this study included patients with bundle branch block and this might have weakened the predictive power of the QT parameters (though not the JT parameters), the analyses were repeated excluding the patients with bundle branch block. This further analysis again indicated that none of the QT parameters were independent predictors of mortality, and the same parameters as in the first analysis were predictive (i.e. cardiothoracic ratio, creatinine, heart rate, LVEDD, and the presence of couplets). Finally, the presence or absence of bundle brunch block was tested for interaction between each of the variables included in the model and no statistically significant interaction was detected. Similarly, antiarrhythmic and digitalis therapy may have influenced the results. However, when the population not taking digitalis or antiarrhythmic therapy were entered into the model there was no change in the variables that were predictive of mortality. Sudden and progressive heart failure deaths There were 24 events classified as sudden death and 30 progressive heart failure death. In patients with mild heart failure as graded by NYHA class (I and II) there were slightly more sudden deaths (3 6%) than progressive heart failure deaths (2 5%). Whereas the opposite was true for patients within NYHA class III or IV (sudden death rate 16 1%; progressive heart failure death rate 18 9%).

5 The UK-HEART trial 1339 Table 3 Statistically significant multivariate predictors of all-cause mortality Variable P value Hazard ratio (95% confidence intervals) For increase of: Cardiothoracic ratio (1 03,1 11) 1% Creatinine (1 03,1 13) 10 μmol.l 1 Heart rate (1 06,1 43) 10 beats. min 1 LVEDD (1 12,2 03) 1 cm Couplets (1 20,5 15) Present vs absent As for all-cause mortality the only QT parameters found to be significantly predictive of sudden death and progressive heart failure death on univariate analysis were QTcd (P=0 029 and respectively) and QTc max (P=0 002 and , respectively). Again, none of the QT parameters were significant independent markers of sudden death or progressive heart failure death. The variables found to be independent predictors of mode of death in the multivariate analysis are detailed in Table 5. Discussion Patients with severe chronic heart failure or multiple arrhythmias have already declared themselves to be at high risk of subsequent mortality, with survival rates at Table 4 QTc max Variable Significant correlates with (a) QTcd and (b) Pearson s correlation coefficient P value (a) QTcd Urate Heart rate LVESD LVEDD CTR BBB SVT Age Digoxin NYHA class (b) QTc max BBB Heart rate CTR LVESD LVEDD No of couplets/24 h Urate No of VEs/24 h SVTs NYHA class AST Age Albumin BBB=bundle branch block; SVT=supraventricular tachycardia; AST=asyptomatic supraventricular tachycardia. 1 year of only 40% [3]. However, among ambulant patients with clinical features of mild to moderate chronic heart failure, treated with an angiotensin converting enzyme inhibitor and diuretics, outcome varies substantially [5]. This study set out to determine whether the QT interval and QT dispersion on a single 12-lead ECG independently predict outcome in this group of patients. Previous studies addressing this issue were all limited in size and have produced conflicting results. This prospective and adequately powered study has shown that although QTc max and QTcd were strong univariate markers, they do not provide independent information regarding all-cause mortality in patients with mild to moderate chronic heart failure, and do not independently predict sudden death or death due to progressive heart failure. These parameters, therefore, do not add any further information to that which can be gained from simple radiographic, biochemical, echocardiographic and Holter data. The variables that significantly correlated with QTcd and QTc max were the main indicators of the degree of pump failure. All of the correlations were weak, however, apart from that between QTc max and the presence of bundle branch block, and QTc max and heart rate which, of course, would be expected to correlate strongly. Previous studies have also shown weak correlations between QT interval parameters and indices of left ventricular function [18,30]. It is possible that the QT interval, therefore, provides some information about the degree of left ventricular damage, possibly as a consequence of fibrotic change within the myocardium. QT intervals may also indirectly indicate the extent of cardiac pump failure by tracking the rise in catecholamine levels found in patients with chronic heart failure. Although catecholamine levels were not measured in this study, previous studies have indicated that QT dispersion does correlate well with a number of indices of sympathetic activation [31 33]. The presented results agree with some [19,20], but not all [17,18,34] published data. There are a number of potential explanations for this variability in results. The previously published studies have all had smaller sample sizes, were in the main retrospective and most did not undertake multivariate analysis. The relatively small number of deaths classified as sudden death in this study and previous studies reduces the statistical power of multivariate analyses. Indeed, in the paper by Fu et al. a combined end-point of sudden death and sustained

6 1340 P. Brooksby et al. Table 5 Statistically significant multivariate predictors of (a) sudden death and (b) progressive heart failure death Variable P value Hazard ratio (95% confidence intervals) For increase of: (a) Cardiothoracic ratio (1 03,1 16) 1% LVEDD (1 13,3 13) 1 cm Potassium (1 08,6 42) 1 mmol. l 1 (b) Creatinine (1 04,1 18) 10 μmol.l 1 Sodium (1 05,1 29) 1 mmol. l 1 Cardiothoracic ratio (1 01,1 12) 1% Couplets (1 15,9 74) Present vs absent ventricular tachycardia had to be used [18]. The definition of sudden death is particularly difficult in chronic heart failure patients and ventricular arrhythmias are responsible for only a proportion of these cases [9,35]. It follows that some of the variability in the results could be explained by differences in the classification of the mode of death, which in the majority of cases has to be made retrospectively. The mix of primary aetiologies for chronic heart failure may have had some influence, as previous work has indicated that QT dispersion was predictive of malignant arrhythmias in patients with ischaemic and not idiopathic dilated cardiomyopathy [36]. However, there does not appear to be a clear pattern as the papers by Barr et al. and Fu et al. both had a similar case mix to that of UK-HEART. The all-cause mortality rate in the studies of Barr et al., Pinsky et al. and Fu et al. were higher (17%, 27% and 30% per annum respectively) than in our study. It is possible, therefore, that QT dispersion is more predictive in patients with chronic heart failure who are at higher risk of early mortality. Unlike previous studies, we included patients with bundle branch block. However, when these patients were excluded from the analysis the results were unaffected. Finally previous studies have indicated the poor reproducibility and repeatability of QT dispersion measurements [37], and this may have a large part to play in producing the variability between the reported studies. It is important to note that the current study merely assessed the predictive value of QT parameters on an isolated, single 12-lead ECG and no serial ECGs were recorded. We cannot, therefore, determine whether dynamic temporal changes in QT parameters have prognostic significance, and this will need to be addressed by future studies. This study has demonstrated the importance of simple measures of left ventricular damage (cardiothoracic ratio and LVEDD in this study) in defining the prognosis of patients with mild to moderate symptoms of chronic heart failure. Cardiothoracic ratio was not only the most significant variable for predicting all-cause mortality, but was also the only variable, which was a significant independent variable for both sudden and progressive heart failure deaths. This result is in accordance with the V-HeFT data, which also showed the cardiothoracic ratio to be a significant independent marker of all-cause mortality, in a population of patients with a similar mortality rate [26]. We did not test the prognostic value of peak oxygen consumption or catecholamine assays in our population, since they are not routinely available to most clinicians. The data presented here for the secondary analysis of sudden and progressive heart failure deaths should be interpreted with caution, since the numbers of events are smaller and the precise definition of an individual s mode of death is exceedingly difficult even when autopsy results are available [9,35]. As previously shown in the V-HeFT study [6], sudden and progressive heart failure deaths seem mainly to be related to the severity of cardiac dysfunction. It is, however, interesting to note that extra-cellular potassium levels were significant independent risk markers for sudden death in this study. It is well known that hypokalaemia increases the risk of ventricular arrhythmias [38,39] and it is possible that this is a mechanism by which sudden deaths occur in the context of chronic heart failure. It is notable, however, that the magnitude of extra-cellular potassium fall was small and although it may predict the behaviour of a population, it is of limited value as an isolated marker of prognosis in the individual patient with chronic heart failure. Conclusion The results presented therefore indicate that QT or JT parameters, measured on a single 12-lead ECG from patients with chronic heart failure, are not significant independent markers of all-cause mortality, sudden cardiac death or progressive heart failure death. These indices do not add further information to that which can be gained from simple radiographic, biochemical, echocardiographic and Holter data. We are indebted to Debbie Blakey RGN, Liz Batin RGN and G. Borthwick for their help in data collection and to the cardiologists in Nottingham, Leeds, Edinburgh, Doncaster, Wakefield and Grimsby for referring us their patients for this study. Financial support was provided by the Chest, Heart and Stroke Association

7 The UK-HEART trial 1341 (Scotland) and the Northern and Yorkshire Research and Development Directorate. References [1] McKee PA, Castelli WP, McNamara PM, Kannel WB. The natural history of congestive heart failure: the Framingham study. N Engl J Med 1971; 285: [2] The SOLVD Investigators. Effect of enalapril on survival in patients with reduced left ventricular ejection fractions and congestive heart failure. N Engl J Med 1991; 325: [3] The CONSENSUS trial study group. Effects of enalapril on mortality in severe congestive heart failure. Results of the Cooperative North Scandinavian Enalapril Survival Study (CONSENSUS). N Engl J Med 1987; 316: [4] Rector TS, Cohn JN. Prognosis in congestive heart failure. Annu Rev Med 1994; 45: [5] Carson P, Johnson G, Fletcher R, Cohn J. Mild systolic dysfunction in heart failure (left ventricular ejection fraction >35%): baseline characteristics, prognosis and response to therapy in the vasodilator in heart failure trials (V-HeFT). J Am Coll Cardiol 1996; 27: [6] Goldman S, Johnson G, Cohn JN, Cintron G, Smith R, Francis G. Mechanism of death in heart failure. The vasodilator-heart failure trials. Circulation 1993; 87 (Suppl VI): V124 V131. [7] Gradman A, Deedwania P, Cody R, Massie B, Packer M, Goldstein S. Predictors of total mortality in congestive heart failure. J Am Coll Cardiol 1989; 14: [8] De Luna AB, Coumel P, Leclerq JF. Ambulatory sudden cardiac death: mechanisms of production of fatal arrhythmia on the basis of data from 157 cases. Am Heart J 1989; 117: [9] Luu M, Stevenson WG, Stevenson LW, Baron K, Walden J. Diverse mechanisms of unexpected cardiac arrest in advanced heart failure. Circulation 1989; 80: [10] Day CP, McComb JM, Campbell RWF. QT dispersion: an indication of arrhythmia risk in patients with long QT intervals. Br Heart J 1990; 63: [11] Higham PD, Hilton CJ, Aitcheson JD, Furniss SS, Bourke JP, Campbell RWF. Does QT dispersion reflect dispersion of ventricular recovery? Circulation 1992; 86: I-392. [12] Higham PD, Furniss SS, Campbell RWF. Acute myocardial infarction, reperfusion and QT dispersion. J Am Coll Cardiol 1992; 19: 304A. [13] Potratz J, Wiegand U, Brandes A et al. Late potentials, heart rate variability and QT-dispersion in patients with implanted cardioverter-defibrillator (Abstr). Eur Heart J 1996; 17 (Abstr Suppl): 384. [14] Perkiomaki JS, Koistinen MJ, YliMayry S, Huikuri HV. Dispersion of QT interval in patients with and without susceptibility to ventricular tachyarrhythmias after previous myocardial infarction. J Am Coll Cardiol 1995; 26: [15] Pye M, Quinn AC, Cobe SM. QT interval dispersion: A non-invasive marker of susceptibility to arrhythmia in patients with sustained ventricular arrhythmias? Br Heart J 1994; 71: [16] Buja G, Miorelli M, Turrini P, Melacini P, Nava A. Comparison of QT dispersion in hypertrophic cardiomyopathy between patients with and without ventricular arrhythmias and sudden death. Am J Cardiol 1993; 72: [17] Barr CS, Naas A, Freeman M, Lang CC, Struthers AD. QT dispersion and sudden unexpected death in chronic heart failure. Lancet 1994; 343 (8893): [18] Fu G-S, Meissner A, Simon R. Repolarization dispersion and sudden cardiac death in patients with impaired left ventricular function. Eur Heart J 1997; 18: [19] Fei L, Goldman JH, Prasad K, Keeling PJ et al. QT dispersion and RR variations on 12-lead ECGs in patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy. Eur Heart J 1996; 17: [20] Strunk-Mueller C, Gietzen F, Kuhn H. QTc dispersion in dilated cardiomyopathy a new method for stratifying the risk of sudden cardiac death (Abstr). Eur Heart J 1996; 17 (Abstr Suppl): 276. [21] Puddu PE, Bourassa MG. Prediction of sudden death from QTc interval prolongation in patients with chronic ischemic heart disease. J Electrocardiol 1986; 19: [22] Schwartz PJ, Wolf S. QT interval prolongation as predictor of sudden death in patients with myocardial infarction. Circulation 1978; 57: [23] Peters RW, Byington RP, Barker A, Yusuf S. Prognostic value of prolonged ventricular repolarisation following myocardial infarction: the BHAT experience. J Clin Epidemiol 1990; 43: [24] Moller M. QT interval in relation to ventricular arrhythmias and sudden death in postmyocardial infarction patients. Acta Med Scand 1981; 210: [25] Nolan J, Batin PD, Andrews R et al. A prospective study of heart rate variability and mortality in chronic heart failure; Results of the United Kingdom heart failure evaluation and assessment of risk trial (UK-HEART). Circulation 1998; In press. [26] Cohn JN, Johnson GR, Shabetai R et al. Ejection fraction, peak exercise oxygen consumption, cardiothoracic ratio, ventricular arrhythmias, and plasma norepinephrine as determinants of prognosis in heart failure. Circulation 1993; 87 (6 S VI): V15 V16. [27] Cleland JGF Dargie HF, Ford I. Mortality in heart failure; Clinical variables of prognostic value. Br Heart J 1987; 58: [28] Greene H, Richardson D, Barker A et al. Classification of deaths after myocardial infarction as arrhythmic or nonarrhythmic. Am J Cardiol 1989; 63: 1 6. [29] Cox DR. Regression models and life tables. J Roy Stat Soc 1972; 34: [30] Lee AFC, Hanson J, Kennedy N, Struthers AD. QT dispersion is related to an increased end diastolic volume on MUGA scan (Abstr). Eur Heart J 1996; 17 (Abstr Suppl): 277. [31] Bonnar CE, Gillespie ND, MacFayden RJ, Ogston SS, Struthers AD. QT-dispersion is significantly increased between 6am and 8am in heart failure patients a possible role in sudden death? J Am Coll Cardiol 1997; 29 (Suppl A): 510A. [32] Bonnar CE, MacFayden RJ, Robson JM, Duncan AM, Struthers AD. QT-dispersion is related to autonomic tone in patients with stable chronic heart failure (Abstr). Eur Heart J 1997; 18 (Abstr Suppl): 200. [33] Lowe MD, Ludman PF, Rowland E, Grace AA. B-adrenoceptor stimulation increases dispersion of cardiac repolarisation (Abstr). Clin Sci 1997; 93. [34] Pinsky DJ, Sciacca RR, Steinberg JS. QT dispersion as a marker of risk in patients awaiting heart transplantation. J Am Coll Cardiol 1997; 29: [35] Pratt CM, Greenway PS, Schoenfeld MH, Hibben ML, Reiffel JA. Exploration of the precision of classifying sudden cardiac death. Implications for the interpretation of clinical trials. Circulation 1996; 93: [36] Cui G, Sen L, Fonarow GC et al. QT dispersion and malignant ventricular arrhythmia in patients with end-stage heart failure caused by ischemic vs. idiopathic cardiomyopathy. J Am Coll Cardiol 1996; 27 (Suppl. A): 121A. [37] Fei L, 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: [38] Tsuij H, Veditti FJ, Evans JC, Larson MG, Levy M. The associations of levels of potassium and magnesium with ventricular premature complexes (the Framingham Heart Study). Am J Cardiol 1994; 74: [39] Gettes L. Electrolyte abnormalities as triggers for lethal ventricular arrhythmias. In: Akhtar M, Myerburg RJ, Ruskin JN, eds. Sudden cardiac death. Philadelphia: Williams & Wilkins, 1994: