Underestimation of duration of ventricular activation by 12-lead ECG compared with direct measurement of activation duration derived from implanted pacemaker leads

AimTo determine extent to which 12-lead ECG QRS duration (QRSd) reflects ventricular activation duration compared with time relations from unpaced ventricular myograms in cardiac resynchronisation therapy (CRT) patients.MethodsLeft (LV) and right ventricular (RV) myograms were recorded during spontaneous rhythm from in-situ pacemaker leads in 77 patients receiving CRT; 14 'normal activation' (unpaced QRSd < 120 ms), 10 'simple left bundle branch block' (LBBB, QRSd 120-149 ms), 40 'advanced LBBB' (QRS ≥ 150 ms) and 13 right bundle branch block. Delay in onset (Q-LV, Q-RV) and duration (dur-LV, dur-RV) of activation were measured. Interventricular delay (ΔT: Q-LV minus Q-RV) and 'LV-overrun' (time between end 12-lead QRS and Q-end LV myogram) were calculated.Results'Normal activation': Neither Q-LV, Q-RV (38 ± 6 ms, 39 ± 11 ms), nor dur-LV, dur-RV (66 ± 9 ms, 81 ± 25 ms) differed. ΔT (− 1 ± 11 ms) was not different from zero, nor was Q-end LV (104 ± 10 ms) different from QRSd (p = 0.09).'Simple LBBB': Q-LV (102 ± 28 ms) was longer than 'normal activation' (p < 0.001), but Q-RV, dur-LV, and dur-RV were no different. ΔT (54 ± 23 ms) was increased (p < 0.001) and Q-end LV (187 ± 48 ms) was longer than QRSd (p = 0.005).'Advanced LBBB': Q-LV (115 ± 52 ms) was longer than 'normal activation' (p < 0.001) but Q-RV was no different, so ΔT (72 ± 47 ms) was increased (p < 0.001 compared to normal, p = 0.04 compared to simple LBBB). Dur-LV (102 ± 27 ms) was also prolonged, so Q-end LV (218 ± 48 ms) was longer than QRSd (p < 0.001). Longer LV-overrun was associated with longer ΔT (p < 0.001).ConclusionsProlonged LV myopotential duration, associated with interventricular delay, is electrically silent on 12-lead QRSd. Unpaced surface QRSd underestimates true duration of native LV activation in CRT patients.