Reliability and reproducibility of QRS duration in the selection of candidates for cardiac resynchronization therapy.

Background: A QRS >120 ms remains the recommended criterion for the selection of cardiac resynchronization therapy (CRT) candidates. However, the reproducibility of this measurement has not been studied thoroughly.
Methods: QRS duration was measured by 3 experienced cardiologists and by automatic measurement on 228 electrocardiograms (ECGs) randomly collected from 188 subjects, including neonates, healthy adults, patients with complete and incomplete bundle branch block, and CRT candidates. All ECGs were recorded at a 25 mm/s sweep speed. Forty recordings were duplicated and 50 ECGs were recorded at both 25 and 50 mm/s.
Results: Significant interobserver differences (P < 0.001) were found between each combination of paired observers, with an up to 50-ms absolute variability between cardiologists and low concordance with computerized measurements. Intraobserver absolute variability was also significant (P < 0.01) for the 3 observers. These significant differences persisted (P < 0.01) when focusing our interest on the ECGs in the 100-140 ms range (defined as at least one out of the 4 measures in this range). Considering the 120 ms limit, 22 (27.5%) ECGs were differently classified by at least one of the cardiologists. We observed similar interobserver differences between each combination of paired observers with a 50 mm/s sweep speed.
Conclusion: Manual QRS duration measurements were associated with significant inter- and intraobserver variability and low concordance with computerized measurements. The measurement of QRS is, therefore, operator-dependent and a reevaluation of the measurement methods may be essential to develop clinical and investigative standards.