Electrogram morphology discriminators in implantable cardioverter defibrillators: A comparative evaluation.

BACKGROUND

Morphology algorithms are currently recommended as a standalone discriminator in single-chamber implantable cardioverter defibrillators (ICDs). However, these proprietary algorithms differ in both design and nominal programming.

OBJECTIVE

To compare three different algorithms with nominal versus advanced programming in their ability to discriminate between ventricular (VT) and supraventricular tachycardia (SVT).

METHODS

In nine European centers, VT and SVTs were collected from Abbott, Boston Scientific, and Medtronic dual- and triple-chamber ICDs via their respective remote monitoring portals. Percentage morphology matches were recorded for selected episodes which were classified as VT or SVT by means of atrioventricular comparison. The sensitivity and related specificity of each manufacturer discriminator was determined at various values of template match percentage from receiving operating characteristics (ROC) curve analysis.

RESULTS

A total of 534 episodes were retained for the analysis. In ROC analyses, Abbott Far Field MD (area under the curve [AUC]: 0.91; P < .001) and Boston Scientific RhythmID (AUC: 0.95; P < .001) show higher AUC than Medtronic Wavelet (AUC: 0.81; P < .001) when tested for their ability to discriminate VT from SVT. At nominal % match threshold all devices provided high sensitivity in VT identification, (91%, 100%, and 90%, respectively, for Abbott, Boston Scientific, and Medtronic) but contrasted specificities in SVT discrimination (85%, 41%, and 62%, respectively). Abbott and Medtronic's nominal thresholds were similar to the optimal thresholds. Optimization of the % match threshold improved the Boston Scientific specificity to 79% without compromising the sensitivity.

CONCLUSION

Proprietary morphology discriminators show important differences in their ability to discriminate SVT. How much this impact the overall discrimination process remains to be investigated.