| Abstrakt: |
Background: The study evaluated interobserver differences in the classification of the T-U wave repolarization pattern, and their influence on the numerical values of manual measurements of QT interval duration and dispersion in standard predischarge 12-lead ECGs recorded in survivors after acute myocardial infarction. Methods: Thirty ECGs recorded at 25 mm/s were measured by six independent observers. The observers used an adopted scheme to classify the repolarization pattern into 1 of 7 categories, based on the appearance of the T wave, and/or the presence of the U wave, and the various extent of fusion between these. In each lead with measurable QRST(U) pattern, the RR, QJ, QT-end, QT-nadir (i.e., interval between Q onset and the nadir or transition between T and U wave) and QU interval were measured, when applicable. Based on these measurements, the mean RR interval, the maximum, minimum, and mean QJ interval, QT-end and/or QT-nadir interval, and QU interval, the difference between the maximum and minimum QT interval (QT dispersion [QTD]), and the coefficient of variation of QT intervals was derived for each recording. The agreement of an individual observer with other observers in the selection of a given repolarization pattern were investigated by an agreement index, and the general reproducibility of repolarization pattern classification was evaluated by the reproducibility index. The interobserver agreement of numerical measurements was assessed by relative errors. To assess the general interobserver reproducibility of a given numerical measurement, the coefficient of variance of the values provided by all observers was computed for each ECG. Statistical comparison of these coefficients was performed using a standard sign test. Results: The results demonstrated the existence of remarkable differences in the selection of classification patterns of repolarization among the observers. More importantly, these differences were mainly related to the presence of more complex patterns of repolarization and contributed to poor interobserver reproducibility of QTD parameters in all 12 leads and in the precordial leads (relative error of 31%-35% and 34%-43%, respectively) as compared with the interobserver reproducibility of both QT and QU interval duration measurements (relative error of 3%-6%, P < 0.01). This observation was not explained by differences in the numerical order between QT interval duration and QTD, as the reproducibility of the QJ interval (i.e., interval of the same numerical order as QTD was significantly better (relative error of 7.5%-13%, P < 0.01) than that of QTD. Conclusions: Poor interobserver reproducibility of QT dispersion related to the presence of complex repolarization patterns may explain, to some extent, a spectrum of QT dispersion values reported in different clinical studies and may limit the clinical utility in this parameter. [ABSTRACT FROM AUTHOR] |
