Considering the effects of circadian rhythm may improve tachycardia discrimination performance in implantable cardioverter defibrillators

This analysis characterized the difference in circadian variation between ventricular and supraventricular tachyarrhythmia (VT and SVT) and examined how tachyarrhythmia discrimination algorithms in implantable cardioverter defibrillators may be improved by considering the time of day an episode occurs. Training and test sets consisting of 3886 tachyarrhythmia episodes from 653 patients were labeled as SVT or VT through visual inspection of intracardiac electrograms. A logistic regression was fit to the training episodes to estimate the probability that an episode is SVT given only the time of day it occurred. A software model of the current discrimination algorithm was enhanced with the logistic regression. The test episodes were input to both the original and enhanced models to calculate sensitivity and specificity to VT. Coefficients of the logistic regression were statistically significant. Circadian variation was different for SVT versus VT. The enhanced model showed an increase in specificity with no change in sensitivity with respect to the original model.