Methods
Agreement and Reproducibility of Automatic Versus Manual Measurement of QT Interval and QT Dispersion

https://doi.org/10.1016/S0002-9149(97)00927-2Get rights and content

Abstract

To determine whether the automatic measurement of the QT interval is consistent with the manual measurement, this study evaluated the reproducibility and agreement of both methods in 70 normal subjects and 54 patients with hypertrophic cardiomyopathy. The mean, minimum, and maximum QT interval and QT dispersion were computed in a set of 6 consecutive electrocardiograms (3 in the supine and 3 in the standing position) obtained from each subject. The automatic method determined the T-wave end as the intersect of the least-squares-fit line around the tangent to the T-wave downslope with the isoelectric baseline. Manual measurements were obtained using a high-resolution digitizing board. QT dispersion was defined as the difference between the maximum and minimum QT interval and as standard deviations of the QT interval duration in all and precordial leads. In patients with hypertrophic cardiomyopathy, the absolute values of the QT interval and QT dispersion were significantly higher than those in normal subjects (p <0.0001). In both groups, the intrasubject variability of the QT interval was significantly lower with automatic than with manual measurement (p <0.05). The agreement between automatic and manual QT interval measurements was surprisingly poor, but it was better in patients with hypertrophic cardiomyopathy (r2 = 0.46 to 0.67) than in normal subjects (r2 = 0.10 to 0.25). In both groups, the reproducibility and agreement of both methods for QT dispersion were significantly poorer than for QT interval. Hence, the automatic QT interval measurements are more stable and reproducible than manual measurement, but the lack of agreement between manual and automatic measurement suggests that clinical experience gained with manual assessment cannot be applied blindly to data obtained from the automatic systems.

Section snippets

Methods

Seventy healthy volunteers (34 men, mean age 39 ± 14 years, range 13 to 59) and 54 patients with documented hypertrophic cardiomyopathy (HC) (35 men, mean age 37 ± 13 years, range 12 to 64) were entered into the study. All healthy volunteers had a normal physical examination and a normal ECG. The diagnosis of idiopathic HC was based on the World Health Organization definition.[14] All patients had left ventricular hypertrophy (≥1.5 cm) on 2-dimensional echocardiography, in the absence of any

Results

Of the total 744 recordings, 159 (20%) electrocardiographic tracings with T-U–wave pattern were excluded from the analysis. Thus, 597 electrocardiographic recordings, 339 (81%) obtained from normal subjects and 258 (80%) obtained from patients with HC were analyzed.

QT Interval and QT Dispersion in Normal Subjects and HC Patients:

The mean values of the QT interval and QT dispersion that were observed in normal subjects are similar to those noted by other investigators.17, 19 A normal range of QT dispersion between 30 and 50 ms was proposed.[20] In our study, the mean value of the manually assessed global QT dispersion was 44 ± 16 ms. Compared with normal subjects, HC patients had significantly higher values of the mean, minimum, and maximum QT interval and of all index of QT dispersion. The mean value of the global QT

References (27)

  • J Morganroth

    Relations of QTc prolongation on the electrocardiogram to torsade de pointesDefinitions and mechanisms

    Am J Cardiol

    (1993)
  • E Lepeshkin et al.

    The measurement of the Q-T interval of the electrocardiogram

    Circulation

    (1952)
  • P Laguna et al.

    Automatic QT interval analysis in postmyocardial infarction patients

    J Amb Monitoring

    (1991)
  • Cited by (140)

    • Machine read frontal QRS-T angle and QTc is no substitute for manual measurement of QTc in pro-arrhythmic drug overdose

      2021, Journal of Electrocardiology
      Citation Excerpt :

      This is supported by the low sensitivity observed in this study for machine QTcRTH measurements in differentiating between normal and prolonged QT. Similarly, other studies have also found wide clinically significant limits of agreement [7,10,37]. Possible explanations for the disagreement between manual and machine QT interval measurements include the presence of noise in the ECG and abnormalities in the shape of the T wave, which can interfere with how machine algorithms measure the QT interval [9].

    • Electrocardiographic repolarization abnormalities and increased risk of life-threatening arrhythmias in children with dilated cardiomyopathy

      2016, Heart Rhythm
      Citation Excerpt :

      One possible explanation for the difference in QTc interval may be attributed to discrepancies between manual and automated ECG measurements, as ECG data entered into databases may not be routinely verified by manual measurements. In one series33 of 339 ECG readings of normal adult patients, the correlation coefficient between manual and automated measurements was 0.1–0.25 (depending on the ECG lead used for interpretation), and in Bland-Altman analysis, the mean difference was ~25 ms with the interval for the limits of agreement of >100 ms. In the setting of abnormal T-wave morphologies, present in 30% of our patient population, manual measurements allow for more accurate representation of the QT and JT intervals. The findings that our interrater correlation was good (ICC = 0.84) and that the mean difference between reviewers was small (7 ms) support the use of manual over automated measurements.

    • Justification of an introductory electrocardiogram teaching mnemonic by demonstration of its prognostic value

      2014, American Journal of Medicine
      Citation Excerpt :

      The rule supplements teaching the assessment of acute ST shifts and arrhythmias. Other important skills done poorly by computers that students need to learn include determining QT interval by the use of the tangent line fit to the downslope of the T wave.13 Our experience in teaching is that students usually “relax” after learning the rule and are more confident with the ECG, making their learning progress more rapidly.

    • Electrocardiographic differentiation of early repolarization from subtle anterior ST-segment elevation myocardial infarction

      2012, Annals of Emergency Medicine
      Citation Excerpt :

      The mean T-wave amplitudeavg:R-wave amplitudeavg ratio was calculated as mean R-wave amplitude in leads V2 to V4/mean T-wave amplitude in leads V2 to V4. The computerized QTc was recorded; it measures the longest of the 12 QT intervals on the 12-lead ECG, and in the normal range it is more accurate than manual measurement.41 Bazett correction divides by the square root of the R-R interval, measured in seconds.

    View all citing articles on Scopus

    The study was supported in part by the European Society of Cardiology, Sophia Antipolis, France, the National Heart Research Fund, Leeds, United Kingdom, and by an Educational Grant of Marquette Medical Systems, Milwaukee, Wisconsin.

    View full text