Author + information
- Received June 2, 1989
- Revision received October 4, 1989
- Accepted October 18, 1989
- Published online March 1, 1990.
- ↵∗Address for reprints: Paul Kligfield, MD, Cornell Medical Center, 525 East 68th Street, New York, New York 10021.
Although electrocardiographic (ECG) voltage can be used to estimate left ventricular mass, day-to-day variability of voltage combinations used for this purpose must be established before ECG changes are taken as evidence of progression or regression of hypertrophy. Accordingly, serial ECGs (mean 8 days apart), derived from 10 s samples digitized at 250 Hz, were examined in 78 patients with no intercurrent change in clinical status. The coefficient of variation was calculated as 1 SD of the difference between paired voltage measurements, divided by the average mean value.
Coefficient of variation for single leads was 22.3% for Sv1 27.0% for Rv5 or Rv6 27.1% for RevL and 34.7% for Sv3. Coefficient of variation was lower for voltage combinations than for individual lead measurements: 18.5% for Sokolow-Lyon voltage (Sv1 + Rv5 or Rv6), 22.3% for Gubner-Ungerleider voltage (R1 + S3) and 24.8% for Cornell voltage (RevL + Sv3). Serial reclassification due to variation above and below standard criteria for left ventricular hypertrophy occurred in only 3% of patients for Sokolow-Lyon voltage and 4% of patients for Cornell voltage in this group. Minute to minute reproducibility of voltage was assessed with electrodes in place in a separate group of 26 patients, and the coefficient of variation was 2.6% for Sokolow-Lyon voltage, 5.9% for Gubner-Ungerleider voltage and 2.9% for Cornell voltage.
These data indicate that serial variability of computer-measured ECG voltage combinations is high, due primarily to changes in lead placement and body position, but less than the variability of computer-measured voltage in individual leads.
- Received June 2, 1989.
- Revision received October 4, 1989.
- Accepted October 18, 1989.