Safety of Atrial Defibrillation Shocks Synchronized to Narrow and Wide QRS Complexes During Atrial Pacing Protocols Simulating Atrial Fibrillation in Dogs
Background The potential ventricular proarrhythmic effect of atrial defibrillation shocks (ADS) remains a concern with automatic internal atrial defibrillation. Optimal R-wave synchronization alone may not be sufficient to prevent the induction of ventricular fibrillation (VF).
Methods and Results The proarrhythmic effect of ADS synchronized to normally conducted QRS complexes (NQRS) and to supraventricular complexes with left or right bundle-branch block (L/RBBB) was investigated in a canine atrial pacing study. Short-long-short, single premature, and burst pacing protocols from the high right atrium were performed at baseline, during isoproterenol infusion, and after intravenous procainamide. The ADS were delivered between decapolar catheters in the coronary sinus and lateral right atrium. They were initially delivered 20 milliseconds (ms) after the end of the last conducted QRS complex and then scanned decrementally through that complex until VF was induced. For NQRS complexes, VF occurred only when the ADS were delivered at or before the onset of the QRS complex and never during the complex itself. In the presence of LBBB or RBBB, VF was induced by ADS delivered at the onset of or within the first 45 ms of the QRS complex in 16 animals. The longest RR (VV) intervals preceding ADS-induced VF were 345 ms at baseline and 380 ms after procainamide.
Conclusions In this study, ADS synchronized to NQRS complexes appeared to be safe regardless of the preceding RR interval. In the presence of LBBB or RBBB, RR intervals preceding the ADS of >345 ms at baseline and >380 ms in the presence of procainamide would have been required to avoid VF. Alternatively, ADS delivered 50 ms after the onset of the RV electrogram appeared to be safe in all circumstances regardless of the preceding RR interval.
Atrial fibrillation is the most common supraventricular arrhythmia, increasing in prevalence with advancing years.1 2 3 It is associated with increased morbidity and mortality and is the most common arrhythmic cause for hospital admission.4 Currently available pharmacological and nonpharmacological therapies have shortcomings, so the search for new approaches continues; one new treatment is the implantable atrial defibrillator.5 6
Development of an implantable atrial defibrillator has gained impetus from the success of the implantable cardioverter-defibrillator in terminating episodes of life-threatening ventricular arrhythmias.7 8 Experimental research and clinical studies have shown that internal atrial defibrillation is feasible and that the energy requirement is significantly lower than that for external defibrillation.9 10 11 12 13 14 15 16
Safety issues, predominantly the potential ventricular proarrhythmic effect of ADS, remain a major concern.5 6 10 Early studies showed VF induction after poorly synchronized shocks that were delivered during the vulnerable period of the preceding beat.11 13 More recent work has suggested that even with adequate synchronization, VF may result if the preceding RR interval is too short.17 However, these results were derived from ventricular pacing protocols and so may have overestimated the ventricular proarrhythmic effect of ADS.
The aim of the present study was to determine the risk of VF induction with ADS synchronized to QRS complexes of supraventricular origin that were normally conducted or conducted with R/LBBB and to refine the criteria for ADS safety. Different atrial pacing protocols were used to mimic the RR cycle length changes seen in atrial fibrillation.
The study protocol was approved by the Animal Care and Use Committee of Sinai Samaritan Medical Center and conformed to the American Heart Association position on research animal use guidelines adopted on November 11, 1994.
Forty-one adult mongrel dogs were anesthetized with intravenous thiopental sodium (25 mg/kg), intubated, and ventilated with oxygen (5 L/min) and 1% to 1.5% halothane with a Harvard model 613 respirator pump. Using a cutdown technique, 8F sheaths were passed into the right femoral artery, both femoral veins, and one external jugular vein. Heart rate and blood pressure were monitored continuously, and arterial blood gas samples were drawn when clinically indicated. Maintenance intravenous fluids (lactated Ringer’s, 150 mL/hr) were infused throughout each study. The heart was exposed through a median sternotomy, and the pericardial sac was opened. With direct manual and fluoroscopic guidance, a 7F quadripolar deflectable tip catheter (Cordis Webster, Inc) was passed into the RV apex, and two 8F decapolar catheters with 2-mm electrode spacing (Electro-Catheter Corp) were positioned, one in the distal coronary sinus and the other in the lateral right atrium with the tip in the right atrial appendage. A quadripolar pacing patch (1×1 cm with 0.5-cm electrode spacing) was sutured to the external right atrial appendage. Two standard ventricular epicardial defibrillation patches were sutured to the anterior and posterior ventricular surfaces. To prevent interruption of the atrial pacing protocols, the sinus node was crushed.
Recordings and Data Analysis
Three surface leads (I, II, and V1), bipolar intracardiac electrograms (from the high right atrium and RV apex), and time lines were simultaneously displayed on a multichannel oscilloscope (Gould ES 2000) and recorded on a magnetic tape for subsequent reproduction. Pacing was performed using a programmable digital stimulator (Bloom Associates Ltd). All measurements were made at paper speeds of 100 or 250 mm/s. For studies involving the generation of bundle-branch block, the appropriate ECG changes were confirmed by recording standard 12-lead ECGs.
The ADS were delivered between the coronary sinus and lateral right atrial catheters. All electrodes of each decapolar catheter were connected through an external junction box so that each catheter formed a single, large-surface-area defibrillation electrode. The ADS were delivered as biphasic truncated exponential waveforms from an external defibrillator (Ventritex HVS-02). Each phase had a 3-ms pulse width, and the leading edge voltage of the second phase was half that of the initial phase. To ensure prompt and efficient defibrillation from induced VF, a monophasic shock was delivered between the epicardial patches. A second external defibrillator (Ventak ECD Model 2806, Cardiac Pacemakers) was used as the power source.
The LLVV to ADS was first determined using a step-down protocol. A 6-beat train of paced beats was delivered from the RV apex at a cycle length of 350 ms, and the ADS were synchronized to the peak of the T wave of the last paced beat. An initial ADS of 2 J was delivered. If VF was induced, the shock strength was decreased in 0.2- to 0.5-J decrements until VF was no longer induced. This energy value was defined as the LLVV. Because of the limitations inherent to external defibrillators, the lowest energy value that could be delivered was 0.2 J. For the initial study protocol, the ADS energy was set at LLVV plus 20% J or 0.2 J, whichever was higher. This was then tested three times to confirm consistent VF induction at this ADS energy level.
To examine the effect of higher ADS energy on the ventricular vulnerability, the protocols were repeated using ADS set at 20% below the ULVV to ADS or 10 J, whichever was lower. The ULVV to ADS was determined using the same RV pacing protocol as for the LLVV, with the ADS starting at 10 J. If three successive 10-J ADS induced VF, this energy setting was used. If 10 J was above the ULVV, the energy was stepped down in 0.2- to 0.5-J decrements until VF was induced. The energy setting immediately above this value was defined as the ULVV. The external defibrillator was then set at 20% below the ULVV and tested three times to confirm consistent VF induction.
Pacing was performed from the high right atrium. Three protocols were used: (1) S-L-S, (2) SP, and (3) BP (Fig 1⇓). For the S-L-S protocol, the basic drive consisted of six stimuli (S1) delivered at a cycle length 50 ms greater than the shortest 1:1 AV conduction interval followed by a stimulus (S′1) delivered after the longest attainable pause without interruption by junctional or ventricular escape complexes. For the SP protocol, the basic drive consisted of a train of six stimuli (S1) delivered at a cycle length 50 ms greater than the shortest 1:1 AV conduction interval. For both protocols, a premature stimulus (S2) was then delivered at progressively shorter coupling intervals to determine the functional refractory period of the AV node. The S2 was timed to the functional refractory period of the AV node to attain the shortest RR interval for both protocols. During the BP protocol, a train of 20 stimuli (S1) was delivered at a cycle length 20 ms greater than the shortest 1:1 AV conduction interval.
For each protocol, the ADS were coupled to the last atrial pacing stimulus and timed relative to the respective RV electrogram. They were initially delivered 20 ms after the end of the last QRS complex (on the surface ECG) and were then decreased in decrements through that QRS complex until VF was induced (Fig 1⇑). The point of VF induction was determined to within 5 ms.
The pacing protocols were performed at baseline, during isoproterenol infusion (1 to 2 μg/min, to decrease the baseline cycle length of the junctional escape rhythm by 20%), and after intravenous procainamide (5 mg/kg). The AV nodal conduction parameters were remeasured for each medication, and the pacing intervals were adjusted appropriately.
Generation of LBBB and RBBB
Two groups of 10 animals were studied after induction of either LBBB or RBBB. Once the ADS energy level was determined and the sinus node was crushed, bundle-branch block was generated with DC ablation. A 7F deflectable tip catheter with a 4-mm tip (Cordis Webster) was passed into the right ventricle or retrogradely across the aortic valve into the left ventricle and manipulated under fluoroscopic guidance until a right or left bundle potential was identified in the distal bipole. This was confirmed by the absence of an atrial electrogram on the tracing and an interval of <30 ms from the potential to the local ventricular electrogram. Bundle-branch ablation was performed by delivery of a 200-J DC shock through the tip of the ablation catheter to a large-surface-area adhesive-patch electrode positioned posteriorly over the spine. A 12-lead ECG was obtained before and after energy application, and ablation was considered successful only if the appropriate ECG (ie, RSR’ in V1 for RBBB or QS in V1 for LBBB) were seen together with a ≥40-ms increase in QRS duration. If necessary, the catheter was repositioned, and energy delivery was repeated. Decremental pacing was performed from the high right atrium before and after ablation to ensure that there was no significant change in the shortest 1:1 AV conduction interval. After generation of RBBB or LBBB, the pacing protocols were performed at baseline, during isoproterenol infusion, and after intravenous procainamide as detailed above.
Definition of Terms
1. The onset of the RV electrogram was defined as the point of departure from the isoelectric baseline at the time of inscription of the surface QRS complex.
2. The V-ADS interval at the point of VF induction was defined as the interval between the onset of the RV electrogram of the last paced atrial impulse (to which the ADS were timed) and the ADS itself. A positive V-ADS value means that the ADS were delivered during or after the final paced RV electrogram, indicating that electrogram fell within the vulnerable period of the preceding beat. A negative V-ADS value means that the ADS were delivered before inscription of the RV electrogram, so that the last paced RV electrogram fell outside the vulnerable period of the preceding beat. A V-ADS value of 0 ms means that the onset of the ventricular electrogram coincided with the outer limit of the ventricular vulnerable period of the preceding beat.
3. To estimate the minimum VV interval that would have been required to avoid VF for those episodes with a V-ADS interval of ≥0 ms at the point of VF induction, the V-ADS interval was added to the VV interval preceding that ADS. Any VV interval greater than this value was then considered to be safe for the energy levels tested in this study.
Data are expressed as mean±SD. A paired Student’s t test was used to compare changes within each group, and an unpaired Student’s t test was used to compare the mean values between different groups. The results of the three pacing protocols were analyzed using ANOVA. Tukey’s posthoc test with Tukey-Kramer adjustments was used to test the differences between the mean values of the different groups. Values of P<.05 were considered statistically significant.
Normally Conducted Supraventricular Complexes
Ventricular vulnerability to ADS synchronized to NQRS was studied at baseline in 21 dogs. In addition, pacing protocols were repeated during isoproterenol infusion in 19 animals and after intravenous procainamide in 10 animals. The results that follow are for the lower-energy ADS set at 0.5±0.7 J. Table 1⇓ shows the basic ECG and electrophysiological intervals for each phase. In the baseline state, the initial junctional escape interval after sinus node crush was 744±235 ms with a QRS width of 46±4 ms, local RV electrogram duration of 38±6 ms, and QT interval of 334±46 ms. The shortest 1:1 AV conduction interval was 239±31 ms. The pacing intervals for the three protocols (S-L-S, SP, and BP) are shown in Table 2⇓. For the three pacing protocols, the V-ADS interval at the time of VF induction was always <5 ms (Table 3⇓). Therefore, regardless of the preceding VV interval, the final paced RV electrogram always fell outside the vulnerable period of the preceding QRS complex.
During isoproterenol infusion, the junctional escape interval shortened to 476±70 ms, and the QT interval shortened to 283±27 ms (Table 1⇑). Both were significantly shorter than at baseline (P<.001). There was no change in the QRS width or local RV electrogram duration (46±4 and 39±6 ms, respectively). The shortest 1:1 AV conduction interval was 220±25 ms, which was shorter than that at baseline (P=.04). The pacing intervals are shown in Table 2⇑. On two occasions, VF was induced by ADS delivered at the onset of the local RV electrogram (V-ADS=0 ms), one each during the S-L-S and BP protocols (Table 3⇑). The ADS were delivered within R2 (V-ADS interval >0 ms) on >250 occasions without induction of VF.
After intravenous procainamide, the junctional escape interval was 641±86 ms, the QRS width was 51±6 ms, the local RV electrogram duration was 41±6 ms, and the QT interval was 341±44 ms (Table 1⇑). None were statistically different from those at baseline. The shortest 1:1 AV conduction interval increased significantly from baseline to 275±35 ms (P=.007). The pacing intervals are shown in Table 2⇑. The V-ADS interval at the point of VF induction was negative for all animals (Table 3⇑), indicating that the RV electrogram of the last paced impulse always fell outside the vulnerable period of the preceding beat.
QRS Complexes With Bundle-Branch Block
A median of two DC shocks was required for left bundle ablation, and a median of one shock was required for right bundle ablation. The QRS width was 99±12 ms after left bundle ablation and 98±8 ms after right bundle ablation (Table 1⇑). Both were significantly increased compared with NQRS (P<.001). When compared with the values for NQRS, there was an increase in the RV electrogram duration for both LBBB (48±11 ms, P=.02) and RBBB (62±16 ms, P<.001). There was no significant change in the baseline QT interval (LBBB=316±30 ms, RBBB=344±29 ms) or junctional escape interval (LBBB=626±115, RBBB=707±108 ms) compared with NQRS.
After induction of LBBB, pacing protocols were performed at baseline in 10 animals, during isoproterenol infusion in 9 animals, and after intravenous procainamide in 6 animals. Again, the results that follow are for the lower-energy ADS, set at 0.5±0.4 J. The pacing intervals for the three protocols are shown in Table 2⇑. When the V-ADS intervals at the point of VF induction are compared with the corresponding values for NQRS, they are greater for all protocols, reaching significance in four of the nine pacing protocols (Table 3⇑).
Table 4⇓ details the episodes of VF induced by ADS delivered at the onset of or within the last paced RV electrogram. There were eight episodes of VF at baseline (Fig 2⇓ shows examples), six episodes on isoproterenol, and six episodes after procainamide, all with a V-ADS interval of ≥0 ms. The longest V-ADS intervals at the point of VF induction were 40 ms in the baseline state and on isoproterenol and 25 ms after procainamide (Table 4⇓). As defined in “Methods,” the minimum preceding VV intervals that would have been required to avoid VF were >300 ms at baseline, >290 ms on isoproterenol, and >330 ms after procainamide (Table 4⇓).
For the RBBB group, pacing protocols were performed at baseline and during isoproterenol infusion in 10 animals and were repeated after intravenous procainamide in 8 animals. Lower-energy ADS of 0.4±0.3 were tested. The timing intervals for the three pacing protocols are presented in Table 2⇑. The VV intervals preceding ADS and the V-ADS intervals at the point of VF induction are shown in Table 3⇑. All the V-ADS intervals are significantly greater than the corresponding values for NQRS (all values P<.05).
Each episode of VF induced by ADS delivered at the onset or within the last paced RV electrogram (V-ADS ≥0 ms) is detailed in Table 4⇑. In the baseline state, there were 14 episodes of VF in which the V-ADS interval was ≥0 ms. There were 13 such episodes during isoproterenol infusion (Fig 3⇓ shows examples) and 6 episodes after procainamide. The longest V-ADS intervals at the point of VF induction were 30 ms in the baseline state, 45 ms on isoproterenol, and 45 ms after procainamide (Table 4⇑). For these pacing protocols, the minimum VV interval preceding ADS required to avoid any instance of VF would have been >345 ms in the baseline state, >325 ms during isoproterenol infusion, and >380 ms after intravenous procainamide (Table 4⇑).
During each experimental setting (baseline state, isoproterenol infusion, and procainamide), the pacing protocols were also performed using higher-energy ADS. The energy values were 10±4 J for the normally conducted complexes, 9±2 J for LBBB, and 9±3 J for RBBB. There were no significant statistical differences between the data presented above for the lower-energy ADS and those obtained with the higher-energy ADS (data not shown).
This study was designed to address in a systematic fashion the issue of the ventricular proarrhythmic potential of ADS coupled to conducted supraventricular complexes. The findings are that although ADS synchronized to NQRS may be safe regardless of the preceding RR interval, the presence of bundle-branch block greatly increases the risk of VF induction. The previously documented RR interval preceding ADS of >300 ms17 would not be sufficient to prevent all cases of VF induction by ADS in this study. A minimum RR interval of 350 ms would have been required at baseline, and this increased to 385 ms after intravenous procainamide. These findings were consistent for both lower-energy (0.5±0.6 J) and higher-energy (9±3 J) ADS.
The existence of a “vulnerable phase of ventricular systole” was confirmed by Wiggers and Wegria,18 who demonstrated the proarrhythmic potential of electrical discharges delivered during inscription of the T wave. This led to the synchronization of external defibrillation shocks to the R wave to avoid the vulnerable phase.19 20 Although this approach allows safe external atrial defibrillation, there are several reports in the literature documenting the serious and even fatal consequences of imperfect synchronization.21 22 23
Early studies in which the safety of internal atrial defibrillation was examined again highlighted the importance of adequate ADS synchronization to the R wave or ventricular electrogram.11 13 The episodes of VF occurred when synchronization failed and ADS were delivered on the T wave. More recent work suggests that perfect synchronization alone is not sufficient to prevent VF. In their pacing study in sheep, Ayers et al17 found that even properly synchronized ADS could induce VF when the preceding RR interval was <300 ms. However, in that study, ventricular pacing protocols were used to mimic the cycle length changes of atrial fibrillation, and these may not represent a true measure of ventricular vulnerability to ADS delivered during conducted supraventricular impulses.
For a properly synchronized ADS to induce VF during atrial fibrillation, the R wave to which it is synchronized must be inscribed during the vulnerable period of the preceding beat. Therefore, the supraventricular impulse that initiates that R wave must be transmitted through the AV node and Purkinje system to a region of excitable ventricular myocardium while another region is still in its relative refractory period. Because the functional refractory period of both the AV node and His-Purkinje system exceeds that of ventricular myocardium, this is unlikely to occur under normal circumstances.24 Our finding that in the baseline state ADS synchronized to NQRS were safe regardless of the preceding RR interval confirms this.
Circumstances in which some parts of the ventricular myocardium might recover later than others could alter this dramatically. An example is during RV pacing, when the impulse may be conducted slowly through some parts of the ventricle and those areas that were depolarized relatively late will also recover later. A more clinically relevant problem is presented by bundle-branch block, which in patients with atrial fibrillation, may be preexisting or rate related as a manifestation of underlying His-Purkinje system disease or may be functional occurring during abrupt S-L-S cycle length variation. The finding of a significantly narrower window of safety in the presence of bundle-branch block confirms the greater proarrhythmic potential of ADS in this setting.
In this study, we also examined the effect of pharmacological manipulation of conduction parameters on the safety of ADS. Isoproterenol is α-agonist, which enhances conduction velocity and shortens the refractory period of the AV node and ventricular myocardium. It may therefore mimic the augmented adrenergic state often accompanying the onset of atrial fibrillation. The potential detrimental effect is shown by the occurrence of two episodes of VF after ADS delivered at the onset of the RV electrogram of NQRS, although in both instances the preceding RR intervals were relatively short (270 and 220 ms). There were 19 instances of VF induction by ADS delivered at the onset of or within the V2 electrogram in the presence of bundle-branch block. A minimum RR interval preceding ADS of >325 ms would have been required to prevent all these episodes.
Intravenous procainamide slows conduction velocity and prolongs action potential duration and the effective refractory period. Our findings may have relevance in the clinical setting as patients with paroxysmal atrial fibrillation may be on an antiarrhythmic drug that prolongs refractoriness and results in a prolonged QT interval. In this study, intravenous procainamide did not have any detrimental effect on the safety of ADS for NQRS. However, in the presence of bundle-branch block, there were 12 episodes of VF induced by ADS delivered within the final paced RV electrogram. To prevent all such episodes, a minimum preceding RR interval of >380 ms would have been required.
A New Criterion of Safety
Based on our data, the safe administration of ADS requires a minimum preceding RR interval of 350 ms in the baseline state and this increases to 385 ms after intravenous procainamide. However, this is based on the assumption that the ADS are delivered at the onset of the RV electrogram. An alternative strategy is to delay the ADS by a predetermined value so that it is delivered after the onset of the local RV electrogram. This would correspond to the mid-to-terminal portion of the QRS complex on the surface ECG and so would still be well before the inner limit of the zone of ventricular vulnerability that occurs on the upstroke of the T wave.18 25 In this study, the longest V-ADS interval at the point of VF induction was 45 ms in an animal with RBBB. Therefore, if ADS delivery had been timed to occur 50 ms after the onset of the RV electrogram, all episodes of VF observed in this study would have been prevented. This value would still ensure that the ADS would have been delivered within or just after the QRS complex.
The main limitation of this study was that selected atrial pacing protocols in normal canine hearts were used to mimic the abrupt cycle length changes seen in spontaneous atrial fibrillation. The extent to which the results may be extrapolated to the clinical setting in humans is therefore unknown. However, by ensuring that the ADS were coupled to the QRS complexes with the shortest preceding RR intervals, the pacing protocols were intended to reproduce the most deleterious cycle length variations encountered in atrial fibrillation. Despite this, allowance could not be made for the possible effect of coexistent myocardial pathology on ventricular vulnerability to ADS in human atrial fibrillation.
DC energy was used to produce LBBB and RBBB. One could argue that this technique could potentially enhance ventricular vulnerability to ADS unrelated to the presence of the bundle-branch block. We cannot dismiss this possibility entirely. However, no animal showed any spontaneous ventricular arrhythmias, and there were no unexpected episodes of ADS-induced VF outside the vulnerable period of the preceding beat. Furthermore, postmortem examination of explanted hearts showed only macroscopic evidence of myocardial injury limited to the region of the upper interventricular septum.
The V-ADS interval was defined as the interval between the onset of the RV electrogram (itself defined as the point of departure from the isoelectric baseline) and the ADS at the time of VF induction. Because implantable defibrillators rely on both amplitude and slew rate to recognize intracardiac electrograms, there will inevitably be a small delay between the onset of the RV electrogram and the point of electrogram recognition by a device. This means that synchronized ADS will not be delivered at the exact onset of the RV electrogram. Therefore, some of the episodes of VF at which the V-ADS interval was 0 ms at the time of VF induction (ie, ADS delivered at or after the onset of the RV electrogram) would not have occurred with an implantable device. However, this small intrinsic delay in RV electrogram detection and ADS delivery would not provide reliable protection against all episodes of VF.
A possible concern not addressed in this study is whether the last conducted QRS complex (during which the ADS were delivered) might exert an effect on the ventricular vulnerability. We have since examined this issue as part of a separate study and have found that there is no change when the ADS are delivered either in the presence or absence of a conducted QRS complex.26 This provides further confirmation that it is the timing of the ADS relative to the vulnerable period of the preceding QRS complex that is critical. This has relevance to another possible concern related to the effect of synchronizing ADS to ventricular ectopic beats rather than conducted supraventricular complexes. Although we did not examine this issue, we feel that the weight of the evidence indicates that as long as it does not fall within the preceding vulnerable period, it should not have any significant impact.
Isoproterenol and procainamide were administered to examine the effects of altered AV conduction and ventricular refractory period on the proarrhythmic potential of ADS. How closely these pharmacological interventions mimic conditions in the clinical setting is also unknown.
This study demonstrates that the conduction properties of the AV node and His-Purkinje system protect against VF induction by ADS synchronized to NQRS. However, in the presence of bundle-branch block, this protection is no longer sufficient to ensure that synchronized ADS will not be delivered within the vulnerable period of the preceding beat that would cause VF. To avoid this, ADS should not be synchronized to QRS complexes with a preceding RR interval of ≤380 ms. Alternatively, the results of this study suggest that deliverance of the ADS in the terminal portion or just after the QRS complex (with a minimum delay from the onset of RV electrogram of 50 ms) will prevent VF regardless of the preceding cycle length. We propose that the safest and most efficient approach would combine both strategies.
Selected Abbreviations and Acronyms
|ADS||=||atrial defibrillation shock(s)|
|LBBB||=||left bundle-branch block|
|LLVV||=||lower limit of ventricular vulnerability|
|NQRS||=||normally conducted QRS complexes|
|RBBB||=||right bundle-branch block|
|ULVV||=||upper limit of ventricular vulnerability|
We are grateful to Brian Schurrer and Brian Miller for their help in preparing the illustrations and to Inga Hawkins for her secretarial assistance.
- Received June 3, 1996.
- Revision received December 5, 1996.
- Accepted December 16, 1996.
- Copyright © 1997 by American Heart Association
Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke. 1991;22:983-988.
Bialy D, Lehmann MH, Schumacher DN, Steinman RT, Meismar MD. Hospitalization for arrhythmias in the United States: importance of atrial fibrillation. J Am Coll Cardiol. 1992;19:41A.
Wharton JM, Johnson EE. Catheter based atrial defibrillation. PACE. 1994;17:1058-1066.
Akhtar M, Avitall B, Jazayeri M, Tchou P, Troup P, Sra J, Axtell K. Role of implantable cardioverter defibrillator therapy in the management of high-risk patients. Circulation. 1992;85(suppl I):I-131-I-139.
Akhtar M, Jazayeri M, Sra J, Tchou P, Rovang K, Blanck Z, Dhala A, Deshpande S, Axtell K. Implantable cardioverter defibrillator for prevention of sudden cardiac death in patients with ventricular tachycardia and ventricular fibrillation: ICD therapy in sudden cardiac death. PACE. 1993;16:511-518.
Levy S, Camm J. An implantable atrial defibrillator: an impossible dream? Circulation. 1993;87:1769-1771.
Levy S, Lauribe P, Dolla E, Kou W, Kadish A, Calkins H, Pagannelli F, Moyal C, Bremondy M, Schork A, Shyr Y, Das S, Shea M, Gupta N, Morady F. A randomized comparison of external and internal cardioversion of chronic atrial fibrillation. Circulation. 1992;86:1415-1420.
Ayers GM, Alferness CA, Ilina M, Wagner DO, Sirokman WA Adams JM, Griffin JC. Ventricular proarrhythmic effects of ventricular cycle length and shock strength in a sheep model of transvenous atrial defibrillation. Circulation. 1994;89:413-422.
Wiggers CJ, Wegria R. Ventricular fibrillation due to single, localized induction and condenser shocks applied during the vulnerable phase of ventricular systole. Am J Physiol. 1940;128:500-505.
Lown B, Amarasingham R, Neuman J. New method for terminating cardiac arrhythmias: use of synchronized capacitor discharge. JAMA. 1963;182:150-157.
Lown B, Perlroth MG, Kaidbey S, Abe T, Harken DE. ‘Cardioversion’ of atrial fibrillation. N Engl J Med. 1963;269:325-331.
Killip T. Synchronized DC precordial shock for arrhythmias. JAMA. 1963;186:1-7.
Ross EM. Cardioversion causing ventricular fibrillation. Arch Intern Med. 64;114:811-814.
Moore EN, Preston JB, Moe GK. Durations of transmembrane action potentials and functional refractory periods of canine false tendon and ventricular myocardium: comparison in single fibers. Circ Res. 1965;17:259-273.
Hou CJ-Y, Chang-Sing P, Flynn E, Martinez L, Peterson J, Ottoboni LK, Liem B, Sung RJ. Determination of ventricular vulnerable period and ventricular fibrillation threshold by use of T-wave shocks in patients undergoing implantation of cardioverter/defibrillators. Circulation. 1995;92:2558-2564.
Keelan ET, Krum D, Hare J, Mughal K, Sra J, Akhtar M, Jazayeri MR. Determination of the minimum heart rate deceleration for safe atrial defibrillation. Circulation. 1996;94(suppl I):I-71. Abstract.