QRS width in right bundle branch block. Accuracy and reproducibility of manual measurement.

Background: The QRS prolongation and its relation to malignant ventricular arrhythmias are topics of interest. Controversies exist about the methodology of measuring the QRS. The aim of this study was to assess the accuracy and reproducibility of manual measurement of the QRS in standard electrocardiograms in patients with right bundle branch block and compare results with computer reading.
Methods and Results: Five experienced cardiologists at different levels of training were required to measure QRS duration in 30 electrocardiograms with different degrees of right bundle branch block collected from 24 randomly selected patients who had had radical repairs of tetralogy of Fallot. In each set of electrocardiograms there were six records which had been duplicated. The observers were neither told the purpose of the study nor how the electrocardiograms had been obtained, nor informed that some of the electrocardiograms were duplicates. Photocopies were identified by number, covering the patient's name and computerised measurement. Significant differences were found in the measurement of QRS in the same ECG calculated twice by the same observer (with an absolute variation up to 50 ms), within different observers (P=0.037) and measured manually or by computer (P=0.019). The width of the QRS did not influence the measurements as the biggest intra-observer variation (50 ms) was observed for relatively wide complex (median value between the two measurements 155 ms) and the biggest inter-observer (60 ms) for narrow complex (median value between the five measurements 110 ms). The QRS morphology appeared to influence the measurements, as the intra- and inter-observer variations were more consistent in the presence of obvious notching, slurrings and terminal slow vectors.
Conclusions: Measurement of QRS is difficult, can be operator dependent and influenced by the presence of conduction abnormalities which reduce its accuracy and reproducibility.