Prospective evaluation of a microprocessor-assisted cardiac rhythm algorithm: results from one clinical center

A new microprocessor-assisted cardiac rhythm algorithm (MAC-RHYTHM), based on median QRST subtraction from 10-second electrocardiographic (ECG) data, was prospectively tested on 10,761 ECGs acquired at one of three clinical centers. Atrial waves (P waves, atrial fibrillatory waves, and flutter waves) were detected from the median QRST-subtracted residual signals. Rhythm criteria were applied to the detected atrial waves and their temporal relation to QRS complexes for generating rhythm interpretations. Rhythm statements generated by MAC-RHYTHM were compared against the true rhythm of the ECGs as read by an experienced cardiologist. The results of prospective testing were compared with the results of an earlier retrospective testing using MAC-RHYTHM and a released commercial ECG analysis program on stored ECGs. The prospective results were very similar to the results of MAC-RHYTHM on retrospective data for all the rhythms examined (sinus rhythms, atrial fibrillation, atrial flutter, junctional rhythms, second degree atrioventricular blocks). For three of the abnormal rhythms, namely, atrial fibrillation, junctional rhythms, and second degree atrioventricular blocks, MAC-RHYTHM gave significantly higher sensitivity in both prospective (87.5%, 92.2%, and 80.8%, respectively) and retrospective (82.0%, 81.2%, and 79.6% respectively) testing than the released commercial ECG analysis program (65.0%, 39.6%, and 12.0% respectively). Similarly, for sinus rhythms, MAC-RHYTHM had significantly higher specificity (prospective, 91.0% and retrospective, 91.7%) than the released commercial program (86.5%). The specificity for the abnormal rhythms remained very high with MAC-RHYTHM (prospective, 99.4% to 99.7% and retrospective, 99.1% to 99.7%) compared to the released commercial program (99.0% to 99.9%). This prospective study also indicated that further work is needed to improve the detection of pacemaker spikes and the interpretation of paced rhythms in 10-second resting ECGs.