Accuracy of computer-calculated and manual QRS duration assessments: Clinical implications to select candidates for cardiac resynchronization therapy

Background: QRS duration (QRSD) plays a key role in the field of cardiac resynchronization therapy (CRT). Computer-calculated QRSD assessments are widely used, however inter-manufacturer differences have not been investigated in CRT candidates.

Methods: QRSD was assessed in 377 digitally stored ECGs: 139 narrow QRS, 140 LBBB and 98 ventricular paced ECGs. Manual QRSD was measured as global QRSD, using digital calipers, by two independent observers. Computer-calculated QRSD was assessed by Marquette 12SL (GE Healthcare, Waukesha, WI, USA) and SEMA3 (Schiller, Baar, Switzerland).

Results: Inter-manufacturer differences of computer-calculated QRSD assessments vary among different QRS morphologies: narrow QRSD: 4 [2-9] ms (median [IQR]), p=0.010; LBBB QRSD: 7 [2-10] ms, p=0.003 and paced QRSD: 13 [6-18] ms, p=0.007. Interobserver differences of manual QRSD assessments measured: narrow QRSD: 4 [2-6] ms, p=non-significant; LBBB QRSD: 6 [3-12] ms, p=0.006; paced QRSD: 8 [4-18] ms, p=0.001. In LBBB ECGs, intraclass correlation coefficients (ICCs) were comparable for inter-manufacturer and interobserver agreement (ICC 0.830 versus 0.837). When assessing paced QRSD, manual measurements showed higher ICC compared to inter-manufacturer agreement (ICC 0.902 versus 0.776). Using guideline cutoffs of 130ms, up to 15% of the LBBB ECGs would be misclassified as <130ms or ≥130ms by at least one method. Using a cutoff of 150ms, this number increases to 33% of ECGs being misclassified. However, by combining LBBB-morphology and QRSD, the number of misclassified ECGs can be decreased by half.

Conclusion: Inter-manufacturer differences in computer-calculated QRSD assessments are significant and may compromise adequate selection of individual CRT candidates when using QRSD as sole parameter. Paced QRSD should preferentially be assessed by manual QRSD measurements.