Improvement and limitation of the reliability of automated QT measurement by recent algorithms

Abstract

Background Newer algorithms for automated QT interval measurements may be more reliable than previous algorithms. Objective This study compares Bazett-corrected QTc obtained by an older algorithm (Old12SL) and by 2 newer ones (New12SL and v3.19) to semiautomated measurement performed by experienced cardiologists. Methods A total of 6105 randomly selected electrocardiograms were classified by the cardiologists as normal (4227), borderline (1254), abnormal (575), or not analyzable (49). Errors of automated measurement were defined by more than 30 milliseconds of absolute difference between Bazett-corrected QTc obtained by automated algorithms and semiautomated measurement. Results The Old12SL had approximately twice as many errors (5.25%) as the New12SL (2.33%) and v3.19 (2.30%), P < .0001. Abnormal tracings resulted in more errors than did normal ones (Old12SL: 16.52% vs 3.45%, New12SL: 7.30% vs 1.51%, and v3.19: 10.61% vs 1.21%). Conclusion Newer automated algorithms for QT measurements are highly reliable in normal tracings. However, electrocardiogram abnormalities increase the risk of QT measurement errors.