Automated measurement of QT interval dispersion from hard-copy ECGs

J Electrocardiol. 1993 Oct;26(4):321-31. doi: 10.1016/0022-0736(93)90053-g.


Increased "dispersion" of the QT interval of the electrocardiogram has been proposed as a marker for increased risk of cardiac arrhythmias, but definitive identification of its independent predictive value requires accurate and reproducible measurement in large numbers of cases. A personal computer-based technique for (1) converting hard-copy electrocardiograms to digital records and (2) automatically measuring QT interval dispersion from the digitized records has been developed and validated. Hand measurements of the RR interval from the original tracing and cursor or automated measurements from digitized waveforms correlated to within 1%. QT intervals measured by cursor on digitized waveforms were a mean of 14 ms (95% confidence interval, 10-19 ms) longer than manual measurements on original tracings. Automatic QT interval measurements were a mean of 5 ms longer than cursor measurements (95% confidence interval, 3-7 ms). Automated measurements were observer independent and repeatable (coefficient of variation for repeat measurements, 0.137% RR and 0.370% QT). Estimates of QT dispersion (expressed as coefficient of interlead QT variation) were made for 14 patients with documented recurrent ventricular and 15 control subjects. The median coefficient of interlead QT variation was 8.8% (range, 4.4-12.4%) for arrhythmia patients and 3.6% (range, 2.7-6.3%) for the control group (P < .001). The automatic measurements were more conservative and less likely to give spuriously large values for QT dispersion than manual measurements. Automated QT dispersion measurements should facilitate future studies on predicting the risk of ventricular arrhythmias.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Analog-Digital Conversion*
  • Arrhythmias, Cardiac / diagnosis*
  • Arrhythmias, Cardiac / epidemiology
  • Electrocardiography / methods*
  • Electronic Data Processing
  • Female
  • Humans
  • Male
  • Microcomputers*
  • Reproducibility of Results
  • Risk Factors
  • Signal Processing, Computer-Assisted*
  • Software