13th International Congress
THE "NEW FRONTIERS"
OF ARRHYTHMIAS 1998

January 24-31, 1998
Marilleva, Trento, Italy

S-167

Future perspectives for detection criteria and algorithmic evolution in ICDs

Eraldo Occhetta, Fabio Cundari*.
Dipartimento di Cardiologia, Ospedale Maggiore della Carita, Novara,
*Ela Medical S.p.A., Milan, Italy

To perform tachycardias classification, third generation of implantable cardioverter defibrillators are provided with pure ventricular signal-based detection criteria.

This family of devices shows technological and performance differences under detection algorithms point of view, based not only upon criteria design, configuration and programming setup, but also upon different detection criteria combination and availability.

Apart from these considerations a list of criteria, built on an historical availability basis, should mainly include the following ones: rate, stability, sudden onset and morphology analysis.

Inside third generation implantable single chamber cardioverter defibrillators, the different arrangement and combination of these criteria allowed a very high sensitivity on ventricular tachycardia, but many supraventricular tachycardias were treated, mainly atrial fibrillation and sinus tachycardia^1-6.

New generations of automatic cardioverter defibrillators, provided with dual chamber pacing-sensing functions allow the design of dual chamber arrhythmias detection algorithms.

In addition to the clear advantages in haemodynamic performance obtained by a dual chamber pacing modality, with important consequences on patients life quality and duration, in such devices, atrial sensing showed its efficacy in ventricular tachycardias detection specificity enhancement when it is flanked to the well known "stability" and "onset"^7,8 inside a full dual chamber architecture-based detection algorithm⁹.

Defender^TM 9001 was the first implanted device adding atrial based criteria: atrioventricular association and chamber of origin of the tachycardia. Its PARAD^TM algorithm, refer to figure 1, showed a very high specificity on sinus tachycardia, 1:1 associated supraventricular tachycardia and atrial flutter¹¹; anyway not rare inappropriate treatments were still observed on fast atrial fibrillation with a nominal persistence programming (8 ventricular cycles). In this case, ventricular rhythm was classified as stable and not associated, because regular PR intervals could not be detected during atrial fibrillation. A reprogramming of persistence duration up to 20 ventricular cycles after a first inappropriate therapy on atrial fibrillation increased specificity, but lengthened therapy time.

Fig. 1: PARAD^TM decision tree.

In order to increase its specificity, without lowering sensitivity, PARAD^TM underwent an evolutive designing process.

Atrial rate, as another classification criterion to be inserted in the full detection algorithm, was excluded, because double tachycardias (atrial fibrillation and ventricular tachycardia) are not exceptional events in patients implanted with defibrillators, and must be treated.

Referring to figure 2, the evolution (PARAD+^TM) of the dual-chamber classification algorithm included a new criterion based on the detection of RR intervals significantly longer than mean tachycardia interval. If at least one long RR interval has been observed during the previous persistence duration, therapy is inhibited. This criterion shows a higher specificity on atrial fibrillation at all persistence durations that all other criteria combination, without significantly decreasing ventricular sensitivity. The optimal parameters programming in terms of sensitivity and specificity is obtained with standard PARAD+^TM with persistence set at 16 cycles. In this case, 25/25 fast atrial fibrillation are rejected from therapy, but 1/33 ventricular tachycardia is not treated, as for PARAD^TM. It is of interest to precise that this non-treated ventricular tachycardia has been induced and recorded during an Electro-Physiological Study⁷. The higher specificity with a 100% specificity with a 100% sensitivity obtained with PARAD+^TM at 8 cycles persistence (92%).

Fig. 2: PARAD+^TM decision tree.

Ventricular tachycardia persistence appears to be another important parameter, in addition to criteria combination in order to increase full algorithm specificity.

This can be explained by atrioventricular conduction variability during atrial fibrillation. On one hand, fast and regular ventricular rhythm is followed by slower and more irregular RR intervals, correctly detected as supraventricular tachycardia by RR stability criterion alone. In this case, lengthening of persistence may allow atrial fibrillation rejection. On the other hand, if long RR intervals occurrence cannot simply be predicted during stable rhythm phases, it is clear that the probability of detecting such intervals increases with rhythm analysis duration: therapy is inhibited by this new criterion.

However, ventricular tachycardia detection delay must be shorter for patients suffering from these arrhythmias compromising haemodynamics. A persistence of 16 cycles, which corresponds to approximately 10 sec for slow ventricular tachycardias, seems to be the best compromise in terms of ventricular tachycardia therapy delay, sensivity and supraventricular tachycardia specificity with PARAD+^TM.

Key Words

Implantable cardioverter defibrillator – new perspectives  
dual chamber ICD, ventricualr signal-based detection criteria, atrial-sensing-pacing, atriventricular association algorithm, PARAD decision tree, detection of RR intervals, ventricular tachycardia persistence, R