Influence of Lead Selection and Population on Automated Measurement of QT Dispersion
Background—The study of QT dispersion (QTd) is of increasing clinical interest, but there are very few data in large healthy populations. Furthermore, there is still discussion on the extent to which QTd reflects dispersion of measurement. This study addresses these problems.
Methods and Results—Twelve-lead ECGs recorded on 1501 apparently healthy adults and 1784 healthy neonates, infants, and children were used to derive normal limits of QTd and QT intervals by use of a fully automated approach. No age gradient or sex differences in QTd were seen and it was found that an upper limit of 50 ms was highly specific. Three-orthogonal-lead ECGs (n=1220) from the Common Standards for Quantitative Electrocardiography database were used to generate derived 12-lead ECGs, which had a significant increase in QTd of 10.1±13.1 ms compared with the original orthogonal-lead ECG but a mean difference of only 1.63±12.2 ms compared with the original 12-lead ECGs. In a population of 361 patients with old myocardial infarction, there was a statistically significant increase in mean QTd compared with that of the adult normal group (32.7±10.0 versus 24.53±8.2 ms; P<0.0001). An estimate of computer measurement error was also obtained by creating 2 sets of 1220 ECGs from the original set of 1220. The mean error (difference in QTd on a paired basis) was found to be 0.28±9.7 ms.
Conclusions—These data indicate that QTd is age and sex independent, has a highly specific upper normal limit of 50 ms, is significantly lower in the 3-orthogonal-lead than in the 12-lead ECG, and is longer in patients with a previous myocardial infarction than in normal subjects.
The QT interval of the 12-lead ECG has long been of clinical interest. It has been known for years that patients with either acquired or congenital long-QT syndrome often suffer sudden cardiac death.1 2 More recently, however, there has been an interest in QT dispersion (QTd), ie, the difference between the maximum and minimum QT intervals in a 12-lead ECG. Mirvis3 showed that these differences could be substantial in normal individuals when 150 electrodes were used on the torso. Furthermore, it has been found that increased QTd is a strong predictor of cardiac death in patients with chronic heart failure4 and peripheral vascular disease.5 It has also been suggested that increased QTd may be a marker of arrhythmia risk in patients with long QT intervals,6 hypertrophic cardiomyopathy,7 and sustained ventricular arrhythmias.8 On the other hand, some recent studies have tended to be less positive. For example, Fei et al9 indicated that increased QTd was not a marker of cardiac death in patients with idiopathic dilated cardiomyopathy. These authors pointed out that atrial fibrillation and bundle-branch block limit the value of QTd in such patients. With respect to the utility of QTd as a clinical tool, Glancy et al10 concluded that its poor reproducibility limited its role, particularly as a predictor of events, whereas Surawicz11 felt that its routine adoption faced serious obstacles, although it was premature to judge them as insurmountable.
The question of whether QTd is merely a reflection of the different projections of resultant cardiac electrical activity onto varying lead axes rather than being related to dispersion at the myocardium has previously been raised.6 12 There is no doubt, however, that the monophasic action potential has a variable duration at different areas of the epicardium, ie, a measurable dispersion as shown by Cowan et al.13
Meaningful data on 12-lead ECG QTd derived from large healthy populations have hitherto not been available. To provide such material and to address some of the points referred to above, a study was undertaken in multiple populations.
To measure QTd, ECGs from different populations were studied. The details of the various populations are as follows:
1. Normal database (adults): There were 1501 individuals (863 men and 638 women) aged between 18 and 78 years in this group. All were apparently healthy for their age. Each underwent a physical examination, at which time blood pressure was checked and a complete history obtained. None was suffering from any abnormality that could have affected cardiovascular status. Details of this population can be found elsewhere.14
2. Normal database (children): This group included 1784 neonates, infants, and children. All were apparently healthy and were examined by a physician either at birth or at postnatal clinics, preschool play groups, or in school. Details of this group are available elsewhere.15
3. CSE study group: There were 1220 ECGs available from individuals enrolled in the European Union Project entitled “Common Standards for Quantitative Electrocardiography” (CSE). There were 831 men and 389 women. The group comprised 382 normal control subjects; the remainder had myocardial infarction or ventricular hypertrophy. Full details are available elsewhere.16
4. Infarct group: A separate group of patients investigated locally for the presence of coronary artery disease after previous hospitalization for myocardial infarction was also studied. All had undergone cardiac catheterization and had been proven to have partially or totally occluded coronary arteries together with ventricular wall motion abnormalities. There was a proven history of inferior myocardial infarction in 180 individuals, namely, 142 men aged 56.8±9.7 years and 38 women aged 61.2±7.5 years. In addition, 181 persons, namely, 131 men aged 57.3±8.8 years and 50 women aged 59.6±8.7 years, had anterior myocardial infarction. There were therefore 273 men and 88 women in this group.
ECGs in groups 1 and 4 were recorded in Glasgow Royal Infirmary with an ECG machine that was designed and developed locally.17 The 12-lead ECG and a hybrid XYZ-lead ECG18 were recorded simultaneously in digital form at 500 samples/s per lead and transmitted to a central computer for analysis.
ECGs in the pediatric age group (group 2) were collected with a Siemens Mingorec 4 digital ECG machine, which also sampled all ECGs at 500 samples/s. Data were written to digital tape by the Mingorec and transferred to the computer laboratory in Glasgow Royal Infirmary for analysis.
ECGs from the CSE study group were recorded at various European centers by use of a variety of electrocardiographs that sampled at 500 samples/s, but all ECGs were available in digital form for analysis at Glasgow Royal Infirmary. This database is available on CD-ROM. Patients in this population had the conventional 12-lead ECG and either a Frank19 orthogonal- or a hybrid18 XYZ-lead ECG recorded.
Computerized ECG Analysis
All ECGs in the study were analyzed by the same locally developed computer program,20 which is used worldwide in commercially available equipment. In brief, the QT intervals are measured from average beats formed from similar cycles. A provisional global QRS onset and T-wave end are determined with all leads, but an individual QRS onset is later determined for each of the 12 leads. For T end, the second peak (P) of the M-shaped T-wave spatial velocity (effectively the sum of the absolute values of the first derivatives of all 12 average beats at each sampling instant) is located. For each lead, the provisional global T end is revised initially on the basis of the noise level. Thereafter, an interval around the revised T end, where T end is later than P, is derived. Then, for each individual lead, a form of second derivative of the signal is used to locate the definitive T end. On the basis of a comparison with other programs published in 1987,21 a small adjustment is then made in the final estimate of T end for each lead. The procedure still applies in the presence of relatively flat T waves. QT intervals are therefore available for all 12 leads individually.
The Glasgow program can analyze 12 or 15 leads recorded simultaneously. Data for the 3-orthogonal-lead ECG were obtained by analysis of all 15 leads (12 leads plus X, Y, and Z leads) simultaneously and then extraction of the 3 orthogonal lead measurements.
Given the availability of orthogonal XYZ leads, it is possible to derive the 12-lead ECG from these leads by the use of generally accepted equations.22 For example, at any sampling instant, lead I can be derived as follows: where a, b, and c are coefficients derived from a model.23 Twelve-lead ECGs obtained in this way are called derived 12-lead ECGs in this article. The derived 12-lead ECG was obtained only from the 1220 CSE ECGs for comparative purposes.
A technique introduced for studying the repeatability of measurements from ECG signals sampled 500 times/s is that of taking every odd sample and constructing a waveform at 250 samples/s and taking every even sample and similarly constructing an ECG waveform.24 This method is known as splitting and effectively creates 2 representations of the same ECG. Thereafter, a process of linear interpolation allows each of the 2 waveforms to be recreated at 500 samples/s. This approach was used for part of the study.
Measurements from all ECGs were written to a file that was used as input to the BMDP Statistical Processing Package. Programs P3D and P7D were used to obtain means, SDs, and 96 percentile ranges. Paired and unpaired tests of significance were used where appropriate, and the level of significance was chosen as 5%.
Table 1⇓ presents the QT intervals calculated from the 1501 adult normal subjects for each lead of the 12-lead ECG in men and women as both corrected and uncorrected values. Data include 96% ranges formed by excluding 2% of values at each extreme of the distribution. The formula of Hodges et al25 was used to correct for rate, viz:
There was a small but statistically significant difference in QT interval, both corrected (7 ms) and uncorrected (4 ms), between men and women, with the latter having the longer intervals. The Bazett formula for QT correction26 would have given a longer QTc and increased the difference between men and women, as shown previously in smaller samples.27 Because of the considerable variation in heart rate in normal neonates, infants, and children, data on the QT interval for this age group are presented in Table 2⇓, corrected by use of the formula of Hodges et al.25
The overall uncorrected QT intervals in anterior infarction and inferior infarction were 390.22±55.0 and 385.24±44.3 ms, respectively, both of which are longer than the overall normal QT interval of 378.8±28.5 ms (P=0.007 for anterior infarction versus normal and P=0.058 for inferior infarction versus normal). Corresponding values for corrected QT intervals were 431.4±29.8, 424.3±22.4, and 402.1±18.6 ms, respectively.
The mean difference in overall QT interval between the derived and actual 12-lead ECG in the 1220 CSE ECGs was 1.25±13.75 ms.
Table 3⇓ presents QTd in the adult normal group for the 12-lead ECG and the orthogonal-XYZ-lead ECG. There was no significant difference between men and women with respect to QTd on the 12-lead ECG (24.67±8.2 versus 24.35±8.2 ms for men and women, respectively). On the other hand, in the 3-orthogonal-lead ECG, corresponding figures were 16.23±9.4 and 14.93±9.2 ms (P<0.01). Table 4⇓ provides data for QTd in the 1784 children.
A comparison was made of the effect of measuring QTd in various subsets of leads of the 1501 adult 12-lead ECGs (Figure 1⇓). The use of all 12 leads gave maximum mean QTd of 24.53±8.2 ms, whereas the use of leads I, II, and V1 through V6 resulted in a small reduction in mean value of QTd to 20.76±7.8 (96% range, 8 to 40) ms. With respect to precordial leads alone, the mean QTd for leads V1 through V6 was 17.13±7.2 (range, 6 to 36) ms. For leads V2 through V6, mean QTd was 14.31±7.2 (range, 4 to 32) ms. For limb leads alone, QTd was 20.11±8.6 (range, 6 to 40) ms.
With respect to the 1220 CSE patients, mean QTd in the conventional 12-lead ECG was 29.1±10.2 (range, 10 to 60) ms, and for the derived 12-lead ECG, QTd was 27.47±10.8 (range, 10 to 66) ms with a mean paired difference in QTd of 1.63±12.22 (range, −30 to 30) ms. This difference is essentially clinically negligible but statistically significant (P<0.0001). QTd in the 3-orthogonal-lead ECGs was 17.1±10.0 (range, 0 to 54) ms, and the paired difference in QTd between the 12-lead ECGs derived from the 3-orthogonal-lead ECGs and the latter themselves was 10.1±13.1 (range, −18 to 38) ms.
With respect to myocardial infarction, QTd was 31.21±9.3 ms in the inferior infarct group and 34.19±10.5 ms in the anterior infarct group. QTd for the combined group of 361 patients was 32.7±10 ms. These values are significantly longer than in normal subjects (P<0.0001 for normal versus anterior infarct and normal versus inferior infarct).
An alternative definition of QTd is the SD of all QT intervals.28 For the 12-lead ECG, by this definition, QTd was 7.79±2.72 ms (range, 3.6 to 15.2 ms) for the 1501 adults and 8.41±3.11 ms (range, 3.23 to 15.87 ms) for the 1784 children. For the 361 patients with old myocardial infarction, the corresponding value was 10.77±3.64 ms (range, 4.42 to 22.31 ms), a highly significant difference (P<0.00001) versus normal.
In recent years, increasing interest has been shown in the technique of measuring QTd to assess the prognosis of certain groups of patients, but its accuracy has come under criticism.10 29 The first point to consider, therefore, is the repeatability of the automated methods used, assessed by the splitting technique. The differences in QTd between the 2 sets of split ECGs were extremely small, with a mean difference of almost 0 ms and a mean absolute difference of only 7.21 ms. These data can be compared with those from repeat manual observations, which give mean differences of 12 ms.10 Intraobserver variation has been measured as 15 ms.10 The data on repeat computer-based measurement of QTd therefore give a certain degree of confidence in interpreting the results derived by automated methods.
It might be questioned whether normal ranges and hence criteria developed with one computer program can be applied to measurements made with another. Although there are undoubtedly differences in strategies adopted by different manufacturers, the program used in the present study has already been shown to be comparable with others in the CSE study21 with respect to interval measurements and overall diagnostic performance.16 No comparative studies on automated measurement of QTd by the major ECG interpretative programs have been undertaken.
3-Lead ECG Versus 12-Lead ECG
The 3 orthogonal leads XYZ produced a smaller QTd than the conventional 12-lead ECG. What is of more interest is the fact that, using the 1220 CSE ECGs whenever both original 12-lead and 3-orthogonal-XYZ-lead ECGs were available, the 12-lead ECG derived from the XYZ-lead ECG gave a significantly longer QTd than the orthogonal XYZ leads themselves (27.47±10.8 versus 17.1±10.0 ms). This difference of ≈10 ms shows that one component of QTd is due to the different projections of the electrical activity on the various lead directions. Thus, given 3 orthogonal leads, it is possible to project the electrical activity onto 12 separate lead axes and produce a much larger QTd purely owing to the mathematical manipulation. This finding supports the argument that part of QTd is due to differing projections of electrical activity and is not due solely to “proximity” effects in precordial leads that would not be reflected in a derived 12-lead ECG.
It might be argued that the differences obtained were due to measurement error. However, given that the 3-orthogonal-lead ECG fiducial points were determined from a 15-lead analysis, this is unlikely to be the case. There can be no doubt that the substantial difference in QTd between the 3-orthogonal and the derived 12-lead ECG is related to the availability of additional lead axes along which cardiac electrical activity is projected.
Normal Group Versus Myocardial Infarction
There was a clear difference in QTd between the myocardial infarction group and the normal subjects. This must be considered along with the fact that the actual QT intervals in the infarct group were also longer than in the normal group. This raises the question as to whether there is any correlation between QTd and QT interval. The data showed this to be the case both for normal subjects (r=0.324, P<0.001) and for patients with myocardial infarction (r=0.282, P<0.001). However, the significance of these correlations is mainly due to the large numbers involved.
Number of Leads Used to Measure QTd
To investigate the effect of the number of leads used in the measurement of QTd, various subsets of the 12-lead ECG were studied. Mean QTd measured from all 12 leads was ≈4 ms longer than QTd measured from leads I, II, and V1 through V6 (20.76±7.8 ms). Considerable dispersion in the 12-lead ECG can be assigned to leads I, II, and V1, there being an additional reduction in mean QTd of ≈6 ms, to 14.31±7.2 ms, if only leads V2 through V6 are considered.
In addition, electrical activity takes longer to travel from the heart to peripheral electrodes, such as those that are used to measure the limb leads, and hence, it could be expected that some of the dispersion may be related to this fact alone.
Redundancy of Limb Leads
The relationships among standard limb leads are well known. For example, Einthoven’s law states that the sum of the potentials in leads I and III at any instant in the cardiac cycle equals the potential in lead II, ie, I+III=II.
Furthermore, if any 2 limb leads are available, all other limb leads can be derived therefrom. For example, if I and II are available, then aVF=II−0.5 I. This therefore suggests that it is only necessary to consider QTd among 8 leads, because if the T wave has ended in 2 limb leads, eg, I and II, then the QT interval in the remaining 4 limb leads can never be substantially greater, allowing for small measurement error and the fact that QRS onset may vary slightly from lead to lead.
QRS Onset Dispersion
There was a mean QTd of 20.11 ms for limb leads only. This may reflect true differences not only in T offsets but in QRS onsets, as well as a component due to measurement error. Mirvis3 found considerable variance around the mean QRS onset derived from 150 individual leads, whereas Cowan et al12 found up to 24-ms variation in QRS onset per 12-lead ECG in their study of only 10 ECGs. The CSE Working Party defined the I segment as the isoelectric section at the start of a QRS complex relative to the earliest QRS onset.30 In manual QT measurement of single leads, an I segment will not be detected, but if multichannel recordings are used with a single QRS onset for all leads, errors may ensue in measurement of QTd.
QT Rate Correction
It is commonplace to use the Bazett formula26 to correct QT interval for heart rate. On the other hand, our own data have shown that the QT interval corrected for heart rate by this formula correlates with heart rate, which defeats the purpose of the correction.27 Correction of the QT interval for rate is a complex area, and many different formulas for it have been suggested. However, for many years, our own experience has been with the formula of Hodges et al,25 which has performed well27 and is gradually gaining more widespread acceptance.31
The formula of Hodges et al25 is linear, which means that rate correction for QTd is unnecessary, ie, with this formula, QTd calculated from corrected QT intervals is identical to QTd calculated from uncorrected intervals. Others have argued that there is no evidence that QTd needs to be corrected with respect to rate,32 and experimental data supporting this conclusion have been presented recently.33
Figure 1⇑ shows the upper limits of normal QTd for various lead groups. These indicate that for use of all 12 leads or a subset of I, II, and V1 through V6, upper limits are similar. A value of 50 ms provides a highly specific estimate of normal QTd in adults and children when an automated method of measurement is used.
There is a generally held view that increased localized dispersion of cardiac electrical activity may lead to arrhythmogenesis,34 but this is different from suggesting that QTd as measured by the ECG can detect the same information. Indeed, the present study shows that QTd is due in part to the variable projection of cardiac electrical activity onto different lead axes. It also shows that differences in QTd between normal subjects and infarct patients can be detected by use of 2 separate definitions, although no attempt has been made to link these with arrhythmic events. From a clinical point of view, however, placing any reliance on small changes in QTd for an individual patient merits caution, because many studies have highlighted measurement problems, possibly due to different paper speed and gain,35 different definitions of the end of the T wave,36 the use of automated versus manual techniques,37 or interobserver and intraobserver variation.10 29 Indeed, the definition of the end of the T wave requires clarification, given that many authors state that in the presence of a U wave, it is the nadir between the T and the U wave that is used as T end.
Nevertheless, as outlined above, many studies have found that increased QTd is a strong predictor of cardiac death. Other studies have found a reduction in QTd in patients receiving therapy.38 39 What is of increasing interest is that data from larger numbers of patients are now becoming available. For example, our own group has recently reported a positive association between increased QTd and nonfatal myocardial infarction or death from coronary heart disease in a study of 6595 men with hypercholesterolemia.40 In a study of 1339 normal subjects and patients with myocardial infarction with and without ventricular tachycardia (VT), Zaidi et al41 found that the group with VT had the highest QTd, whereas those with a history of myocardial infarction but no VT had a mean QTd higher than the normal group but lower than those with VT. In that study, QTd was based on 15 leads. de Bruyne et al,42 in a longitudinal study of 5523 elderly individuals, found that those whose corrected QTd was in the highest tertile (QTd >59.6 ms) had a relative risk of a cardiac death of 2.1 compared with those in the lowest tertile (QTd <39.0 ms). Thus, additional studies on the development of criteria that acknowledge the variability of the measurement technique being used appear to be justified to advance the use of QTd in the individual patient. This point is important because trends can certainly be seen between higher QTd and death from coronary heart disease, but there is still a question of what constitutes a sensitive and specific criterion for QTd in the individual patient.
In the meantime, the present study suggests that the use of a threshold value of 50 ms for abnormal QTd as measured by computer appears to be justified. That does not imply that this value is sensitive and specific for the detection of VT, for example. Nevertheless, QTd is a measurement that should be treated with some caution for the present. It has been suggested that it is perhaps “an electrocardiographic Holy Grail,”43 but there must be considerable dubiety about this. Results of additional investigation of the technique are awaited with interest.
- Received May 27, 1998.
- Revision received July 8, 1998.
- Accepted July 21, 1998.
- Copyright © 1998 by American Heart Association
Schwartz PJ, Wolf S. QT interval prolongation as a predictor of sudden death in patients with myocardial infarction. Circulation.. 1978;57:1074–1077.
Mirvis DM. Spatial variation of QT intervals in normal persons and patients with acute myocardial infarction. J Am Coll Cardiol. 1985;5:625–631.
Darbar D, Luck J, Davidson N, Pringle T, Main G, McNeill G, Struthers AD. Sensitivity and specificity of QTc dispersion for identification of risk of cardiac death in patients with peripheral vascular disease. BMJ. 1996;312:874–879.
Day CP, McComb JM, Campbell RW. QT dispersion: an indication of arrhythmia risk in patients with long QT intervals. Br Heart J. 1990;63:342–344.
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;71:511–514.
Fei L, Goldman JH, Prasad A, Keeling PJ, Reardon K, Camm AJ, McKenna WJ. QT dispersion and RR variations on 12-lead ECGs in patients with congestive heart failure secondary to idiopathic dilated cardiomyopathy. Eur Heart J. 1996;17:258–263.
Glancy MJ, Weston PJ, Bhullar HK, Garratt CJ, Woods KL, de Bono DP. Reproducibility and automatic measurement of QT dispersion. Eur Heart J. 1996;17:1035–1039.
Surawicz B. Will QT dispersion play a role in clinical decision making? J Cardiovasc Electrophysiol. 1996;7:777–784.
Cowan JC, Hilton CJ, Griffiths CJ, Tansuphaswadikul S, Bourke JP, Murray A, Campbell RWF. Sequence of epicardial repolarisation and configuration of the T wave. Br Heart J. 1988;60:424–433.
Macfarlane PW, McLaughlin SC, Devine B, Yang TF. Effects of age, sex, and race on ECG interval measurements. J Electrocardiol. 1994;27(suppl):14–19.
Macfarlane PW, Coleman EN, Pomphrey EO, McLaughlin SC, Houston A, Aitcheson TC. Normal limits of the high-fidelity paediatric ECG: preliminary observations. J Electrocardiol. 1989;22(suppl):163–169.
Willems JL, Abreu-Lima C, Arnaud P, van Bemmel JH, Brohet C, Degani R, Denis B, Gehring J, Graham I, van Herpen G, Machado H, Macfarlane PW, Michaelis J, Moulopoulos SD, Rubel P, Zywietz C. The diagnostic performance of computer programs for the interpretation of electrocardiograms. N Engl J Med. 1991;325:1767–1773.
Watts MP, Shoat DB. Trends in electrocardiographic design. J IERE. 1987;57:140–150.
Macfarlane PW. A hybrid lead system for routine electrocardiography. In: Macfarlane PW, ed. Progress in Electrocardiology. London, UK: Pitman Medical; 1979:1–5.
Frank E. An accurate, clinically practical system for spatial vectorcardiography. Circulation. 1956;13:737–749.
Bailey JJ, Itscoitz SB, Hirshfeld JW Jr, Grauer LE, Horton MR. A method for evaluating computer programs for electrocardiographic interpretation, I: application to experimental IBM program of 1971. Circulation. 1974;50:73–79.
Hodges M, Salerno D, Erlinen D. Bazett’s QT correction reviewed: evidence that a linear QT correction for heart rate is better. J Am Coll Cardiol. 1983;1:694. Abstract.
Bazett HC. An analysis of the time relations of the electrocardiograms. Heart. 1920;7:353–370.
Macfarlane PW, Lawrie TDV. The normal electrocardiogram and vectorcardiogram. In: Macfarlane PW, Lawrie TDV, eds. Comprehensive Electrocardiology. Oxford, UK. Pergamon Press; 1989;1:407–457.
Hnatkova K, Malik M, Kautzner J, Gang Y, Camm AJ. Adjustment of QT dispersion assessed from 12 lead electrocardiograms for different numbers of analysed electrocardiographic leads: comparison of stability of different methods. Br Heart J. 1994;72:390–396.
The Common Standards for Quantitative Electrocardiography Working Party. Recommendations for measurement standards in quantitative electrocardiography. Eur Heart J. 1985;6:815–825.
Wagner G. Marriott’s Practical Electrocardiography. 9th ed. Baltimore, Md: Williams & Wilkins; 1994.
Zabel M, Wolfgang J. Rate dependence of the QT interval and of QT dispersion: comparison of atrial pacing and exercise testing. Circulation. 1997;96(suppl I):I-325. Abstract.
Higham PD, Campbell RWF. QT dispersion. Br Heart J. 1994;71:508–510.
Murray A, McLaughlin NB, Burke JP, Doig JC, Furniss SS, Campbell RWF. Errors in manual measurement of QT intervals. Br Heart J. 1994;71:386–390.
McLaughlin NB, Campbell RWF, Murray A. Comparison of automatic QT measurement techniques in the normal 12-lead electrocardiogram. Br Heart J. 1995;74:84–89.
Moreno FL, Villaneuva T, Karagounis LA, Anderson JL, for TEAM-2. Reduction in QT interval dispersion by successful thrombolysis therapy in acute myocardial infarction. Circulation. 1994;90:94–100.
Macfarlane PW, on behalf of the WOSCOPS Study Group. Electrocardiographic measures including QT dispersion in the West of Scotland Coronary Prevention Study. Circulation. 1997;96:I-325–I-326. Abstract.
Zaidi M, Robert A, Fesler R, Derwael C, Brohet C. Dispersion of ventricular repolarisation: a marker of ventricular arrhythmias in patients with previous myocardial infarction. Heart. 1997;78:371–375.
de Bruyne MC, Hoes AW, Kors JA, Hofman A, Grobbee DE, van Bemmel JH. QTc dispersion predicts cardiac mortality in the elderly: the Rotterdam Study. Circulation. 1998;97:467–472.
Campbell R. Commentary: QT dispersion may reflect vulnerability to ventricular fibrillation. BMJ. 1996;312:878–879.QT dispersion (QTd) was assessed in 3285 children and adults by use of an automated technique. An upper limit of 50 ms was found to be highly specific, whereas no age or sex differences were found. An increased mean QTd was found in patients with a previous myocardial infarction. It was also shown that mean QTd measured in 12-lead ECGs derived from corresponding 3-orthogonal-lead ECGs was ≈10 ms greater, indicating that QTd is related in part to the number of leads used for its measurement, the mean error of which was estimated to be 0.28±9.7 ms.