The Evidence Against QT Dispersion

Gerard van Herpen, Henk J. Ritsema van Eck, and Jan A. Kors

Department of Medical Informatics, Erasmus MC - University Medical Centre Rotterdam,
Rotterdam, The Netherlands

Correspondence: G van Herpen, Department of Medical Informatics, Erasmus MC, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands.
E-mail: vanherpen@mi.fgg.eur.nl, phone +31 10 4087045, fax +31 10 4089447

---

---

1.  Introduction

The ST-T segment of the ECG is regarded as the resultant of the combined repolarisation activities of the whole ventricular myocardium. It is the treasure chest in which all information is hoarded about the ventricular recovery process, normal or pathological. To lay hands on its precious contents is a tantalizing challenge. A key to this treasury seemed to be offered by Campbell and his group in 1990 in the form of QT dispersion (QTD) [Day et al., 1990]. The idea was that intramyocardial differences in action potential duration would be mapped onto surface ECG leads, resulting in differences in QT duration between the leads. QTD, the difference between longest and shortest QT among the leads, was thought to reflect this intramyocardial heterogeneity of repolarisation and its increase to reveal a pathological process (Fig. 1). Indeed, Campbell, after some hesitation, spoke of “a glimpse of the Holy Grail”. His initial misgivings, however, were justified. The idea of QTD rests on a fallacy. Nevertheless, the concept was enthusiastically embraced and a tidal wave of publications followed in which QTD was statistically correlated with all sorts of pathological conditions, present or future. For a survey, see [Malik and Batchvarov, 2000].

Figure 1. QTD as illustrated by Higham and Campbell (Br Heart J 71: 509, 1994). Different QT durations have been measured between the extremity leads.

2.  Why QTD Does Not Work

The repolarisation processes in the heart are accompanied by electrical fields, which produce potentials in the exploring electrodes. The QTD hypothesis presupposes that the repolarisation potential vanishes at different times in different electrodes. In the heart, if there is dispersion of repolarisation duration, there must be a tissue element that is the last to finish its repolarisation [Kors and VanHerpen, 1998; Coumel et al., 1998; Kors et al., 1999]. The electrical field of this element will extend throughout the volume conductor of the body. Wherever an electrode, it will sense the electrical field and only cease to do so when the tissue element is fully repolarised and the field vanishes. In some electrodes it might be difficult to define the end of a weak signal, but that is a matter of precision of measurement and does not detract from the physical truth that there is one simultaneous end to all electrode activity so that QTD cannot exist.

Those who are reluctant to accept this reasoning might perhaps be convinced by merit of a “reductio ad absurdum”: let us assume that there is indeed a difference in repolarisation duration between electrodes.

1. The so-called unipolar precordial leads are, of course, not really unipolar: they use Wilson’s central terminal, the combination of the R, L and F electrodes, as the electrode opposing the chest electrodes. If the longest repolarisation potential duration happens to occur in one of these extremity electrodes, this imposes one and the same QT duration on all V leads. The tool of QTD, therefore, works one-directionally: only if repolarisation potentials last longer in V electrodes then in extremity electrodes will there be QTD between them, not if they are shorter.

2. Let us assume that among the extremity electrodes there is a difference in repolarisation potential duration and that it is longest in R. Then, since R takes part in leads I and II, these leads must be of equal QT duration while III, between F and L, cannot be longer, only equal or shorter. The “augmented’ leads aVR =R–(F+L)/2, aVL= L–(F+R)/2 and aVF= F–(L+R)/2 all contain R and must be equal in duration to I an II. The same reasoning holds if R is replaced by L or F. In any case, five extremity leads must have the same QT duration; in only one can it be shorter. Regardless of this simple truth, in numerous publications on QTD, without any qualms, QT duration differences were measured between all of the extremity leads (Fig.1).

3. The above reasoning can be put to use to assess the effect of measuring error on QTD determination [Kors and Van Herpen, 1998]. As argued, five of the six limb leads must have equal QT duration; only one can be shorter. Thus, between these five QTD = 0 and if any QTD > 0 is measured this must be due to measurement error. QTD was measured by computer in the 5 “non-shortest” leads in the 1220 ECGs of the CSE database [Willems et al., 1991]. Mean and SD turned out to be 20.4 ms and 11.5 ms. By means of a statistical stratagem the data could be used to estimate the error in the 12-lead ECGs as well. The mean error (SD) with respect to an assumed true QTD (0 or any other figure) was 29.4 ± 14.9 ms. These error figures give an impression of the intrinsic unreliability of QT measurement and are dauntingly high, considering the QTD measurements reported in the literature where characteristic values for normals are in the order of 30 ± 10 ms [Van de Loo et al., 1994] and 43 ± 12 ms [Linker et al., 1992] and range from 38 ± 13 ms [Higham et al., 1995] to 94 ± 41 ms [Glancy et al., 1995] for pathological cases. The uncertainty of the measurement of course also leads to poor reproducibility within and between observers.

4. An interesting way to persuade us that QTD originates from measurement inaccuracy has been shown by Lee et al. (1998) and by Macfarlane et al. (1998). From the 12 leads of a standard ECG, using linear transformation, one can derive three leads in the way of the orthogonal components of a vectorcardiogram. Conversely, from this VCG the 12 leads can be reconstructed. These leads resemble the original ones, but are not identical. As in this reconstruction every electrode potential is derived from all 3 orthogonal leads it is impossible that any QTD should exist between the reconstructions. Nevertheless, when QT durations were measured QTD appeared and it was identical with the QTD found in the original ECGs. This suggests that QTD in the original ECGs can entirely be accounted for by measurement error.

3.  Why QTD Nevertheless Seems to Work

As said, no electrode ceases to sense the electrical field before ventricular repolarisation has been fully completed. But a lead registers the difference between two electrode potentials and such voltage can very well be zero. In that case the ECG becomes isoelectric. In the VCG this means that the vector is perpendicular to the lead. In a normal VCG the T loop is elongated and slender with its main axis to the left-anterior and downward, more or less perpendicular to leads V1 and III. In these leads the projection of the T loop tends to be flat and its end will be marked prematurely, where the curve seems to join the 0-line. However, accepted procedure for QTD measurement prescribes that flat T waves are excluded from measurement, precisely those that would produce a short QT duration and therewith an increase in QTD. A pathological T loop, to the contrary, is often wide and distorted. When its mean axis is perpendicular to a lead the sideward extension of the loop, parallel to the lead, may produce a T wave amplitude large enough to be included in the measurements. Still, its QT duration is shortened, especially if the terminal part of the loop is again perpendicular to the lead. This explains why in normals QTD might be lower than in pathological cases: it is kept artificially low by the measurement rules. The concept [Kors et al., 1999] is illustrated by Fig. 2.

Figure 2. A and B: normal T loop, C: pathological, wide T loop. To the right: T waves as projection of the T loops on the lead axis. Arrow marks end of T. In B the T wave is flat, due to perpendicular projection and will be excluded from QTD measurement. In C the pathological T wave will be measured but QT will be shortened.

Essentially the same mechanism seems to underlie the findings of Priori et al. (1997). Using principal component analysis as a measure of “QT complexity” they found QTD about proportional to the ratio between 2nd and 1st eigen values.

4.  Enters the U Wave

The problem of low ST-T voltage was aptly solved by ignoring it. A second nuisance is often the U wave. In the presence of a prominent U wave the end of T is taken at the signal nadir between T and U. By doing this one shortens QT duration and increases QTD. But is the U wave perhaps something more than a nuisance? Ritsema van Eck in our institute has tackled the problem of the U wave by means of computer modeling. U waves then appear as a normal constituent of repolarisation. T and U form a continuum and not before the end of the U wave is ventricular repolarisation complete. This means that QTD, even if it made sense in physical terms, has no meaning in terms of dispersion of repolarisation: the end of T is not the end of repolarisation.

5.  Conclusion

It is saddening to see how many investigators have devoted time and effort to what, in the words of Rautaharju (1999) was “the greatest fallacy in electrocardiography in the 1990's”, and that the basic physical teachings of electrocardiography have been lost. It seems that the opportunity to measure some quantity and apply statistics to it suffices to put hundreds to laborious work thinking that, if so many invest in the selfsame business it must be worth your money. Afterwards it is hard to write off a once so promising venture. Good science is often inductive, i.e. proceeds from correctly observed facts by way of a creditable model to general conclusions, and in this sense has always been evidence-based. But deductive verification is equally necessary to clean the path from observation to conclusion. Deductive physical reasoning would have warned against welcoming QTD as a harbinger of repolarisation disorder. Its physical model is based on nothing more than vague electrophysiological notions, its physical justification only that the magnitude of its measuring error is somehow in relation to what is measured. New insight into the nature of the U wave as an integral part of ventricular repolarisation even asks for rethinking of the entire concept of QT duration and prolongation.

References

Coumel P, Maison-Blanche P, Badilini F. Dispersion of ventricular repolarisation. Reality? Illusion? Significance? Circulation, 97: 2491-2493, 1998.

Day CP, McComb JM, Campbell RWF. QT dispersion: an indication of arrhythmia risk in patients with long QT intervals. Br Heart J, 63: 342-344, 1990.

Glancy JM, Garratt CJ, Woods KL, et al. QT dispersion and mortality after myocardial infarction. Lancet, 345: 945-948, 1995.

Higham PD, Furniss SS, Campbell, RWF. QT dispersion and components of the QT interval in ischaemia and infarction. Br Heart J, 73: 32-38, 1995.

Kors JA, Van Herpen G. Measurement error as a source of QT dispersion. Heart, 80: 453-458, 1998.

Kors JA, Van Herpen G, Van Bemmel JH. QT dispersion as an attribute of T-loop morphology. Circulation, 99: 1458-1463, 1999.

Lee KW, Kligfield P, Okin PM, Dower GE. Determinants of precordial QT dispersion in normal subjects. J Electrocardiol, 31 Suppl.: 128-133, 1998.

Linker NJ, Colonna P, Kekwick CA, Till J, Camm AJ, Ward DE. Assessment of QT dispersion in symptomatic patients with congenital long QT syndromes. Am J Cardiol, 69: 634-638, 1992.

Macfarlane PW, McLaughlin SC, Rodger C. Influence of lead selection and population on automated measurement of QT dispersion. Circulation, 98: 2160-2167, 1998.

Malik M, Batchvarov V. Interpretation and clinical potential of QT dispersion. J Am Coll Cardiol, 36: 1749-1766, 2000.

Rautaharju PM. QT and dispersion of ventricular repolarisation: the greatest fallacy in electrocardiography in the 1990s. Circulation, 99: 2477-2478, 1999.

Van de Loo A, Arendts W, Hohnloser S. Variability of QT dispersion measurements in the surface electrocardiogram in patients with acute myocardial infarction and in normal subjects. Am J Cardiol, 74: 1113-1118, 1994.

Willems JL, Abreu-Lima C, Arnaud P, et al. The diagnostic performance of computer programs for the interpretation of electrocardiograms. N Engl J Med, 325: 1767-1773, 1991.