Defibrillation Failure Research Articles

ANALYSIS OF THE VIRTUAL ELECTRODE PHENOMENA USING BIDOMAIN MODEL: BASIC CHARACTERISTICS FOR PASSIVE MEMBRANE

The virtual electrode (VE) has been recognized as an important factor for success or failure of cardiac defibrillation. Many researches have been performed to study characteristics of the VE. However, there are some questions which remain unanswered. In this study, we developed a simulator to solve a three-dimensional bidomain model and performed several simulations to elucidate the basic characteristics of VE in a simplified cardiac tissue with passive membrane when a constant unipolar cathodal stimulus was applied. The results showed that for smaller electrodes, VE has a typical dog-bone shaped virtual cathode (VC) and two egg-shaped virtual anodes (VAs). The distributions both in intra- and extracellular potentials have concentric ellipsoidal isosurfaces, but their ellipticities are subtly different, producing VE. For larger electrodes, VC becomes larger and has a flat-dish shape rather than dog-bone, and VA becomes smaller and also flattens and collapses. The peak values of VE are larger for smaller electrodes, but their time courses show similar tendency among the different sized electrodes. The change of stimulus strength and polarity only affects the magnitude of VE in a linear manner and the distribution pattern is unchanged. These results provide us fundamental knowledge about VE.

Multiple Failed External Defibrillation Attempts During Robot-Assisted Internal Mammary Harvest for Myocardial Revascularization

We describe multiple failed external defibrillation attempts via rescue defibrillation pads for ventricular fibrillation that occurred secondary to direct electrical current transmission through the pericardial sac from electrocautery during robot-assisted internal mammary harvest. Only after resumption of two-lung ventilation and decompression of the iatrogenic pneumothorax was the patient successfully defibrillated. Conditions necessary for robotic intrathoracic surgery may make defibrillation and resuscitation difficult if they become necessary.

Effects of compression depth and pre-shock pauses predict defibrillation failure during cardiac arrest

Cardiopulmonary resuscitation (CPR) and electrical defibrillation are the primary treatment options for ventricular fibrillation (VF). While recent studies have shown that providing CPR prior to defibrillation may improve outcomes, the effects of CPR quality remain unclear. Specifically, the clinical effects of compression depth and pauses in chest compression prior to defibrillation (pre-shock pauses) are unknown. A prospective, multi-center, observational study of adult in-hospital and out-of-hospital cardiac resuscitations was conducted between March 2002 and December 2005. An investigational monitor/defibrillator equipped to measure compression characteristics during CPR was used. Data were analyzed from 60 consecutive resuscitations in which a first shock was administered for VF. The primary outcome was first shock success defined as removal of VF for at least 5s following defibrillation. A logistic regression analysis demonstrated that successful defibrillation was associated with shorter pre-shock pauses (adjusted odds ratio 1.86 for every 5s decrease; 95% confidence interval 1.10-3.15) and higher mean compression depth during the 30s of CPR preceding the pre-shock pause (adjusted odds ratio 1.99 for every 5mm increase; 95% confidence interval 1.08-3.66). The quality of CPR prior to defibrillation directly affects clinical outcomes. Specifically, longer pre-shock pauses and shallow chest compressions are associated with defibrillation failure. Strategies to correct these deficiencies should be developed and consideration should be made to replacing current-generation automated external defibrillators that require long pre-shock pauses for rhythm analysis.

P4-24: Vulnerability to electric shocks in regional ischemia

Defibrillation is the only effective therapy against sudden cardiac death. However, defibrillation research has focused mostly on normal hearts, while the majority of patients who undergo the procedure exhibit cardiac pathophysiology secondary to coronary artery disease. Little is known about defibrillation in the setting of ischemic disease. The goal of this study is to aid understanding of defibrillation failure in ischemic hearts by studying changes in cardiac vulnerability to electric shocks during the first 10min following LAD occlusion.

Lessons From the Failure and Recall of an Implantable Cardioverter-Defibrillator

Multiple clinical trials have shown that a properly functioning implantable cardioverter-defibrillator (ICD) is capable of interrupting sudden death caused by ventricular tachyarrhythmias. Unfortunately, ICDs are complex medical devices, and they do not always perform as expected. For example, only 5% of ICD batteries last >7 years, and most dual-chamber ICD models must be replaced for battery depletion every 3 to 5 years.1 Normal battery depletion, however, is reliably predicted by routine follow-up methods before an ICD fails. In contrast, electronic malfunctions are unpredictable and may not be detected by standard follow-up techniques before an ICD is unable to deliver effective therapy. Thus, sudden cardiac arrest or death may be the first and only sign that an ICD has failed.2 Recently, we reported the death of a 21-year-old patient who received a Prizm 2 DR model 1861 ICD pulse generator (Guidant, Inc) in 2001 to prevent sudden cardiac death resulting from hypertrophic cardiomyopathy.3 In March 2005, this young man experienced a witnessed arrest and could not be resuscitated. His ICD was returned to Guidant, which found that the device had failed during the delivery of a shock. The cause of failure was massive electronic damage caused by electrical overstress that occurred when a short circuit developed between a high-voltage wire and a tube used to test the housing during manufacturing (see the Figure). At the time of our patient’s death, Guidant had knowledge of 25 similar Prizm 2 DR model 1861 failures in patients, 3 of whom had required rescue defibrillation. Indeed, Guidant had first observed this mode of failure 3 years earlier, in February 2002, when 2 returned Prizm 2 DR pulse generators exhibited the same short circuiting that caused our patient’s device to fail. Guidant was sufficiently concerned about these failures that manufacturing changes were made …

Evidence That Activation Following Failed Defibrillation Is Not Caused by Triggered Activity

Earliest postshock activation following failed defibrillation shocks slightly lower than the defibrillation threshold (DFT) in large animals appears to arise from a focus. We tested the hypothesis that these foci are caused by early or delayed afterdepolarizations (EADs or DADs) by performing epicardial electrical mapping and giving the EAD inhibitor pinacidil or the DAD inhibitor flunarizine to see if the foci were extinguished or altered in timing or location. A sock containing 504 electrodes was placed over the entire ventricular epicardium of 12 open-chested pigs. After the DFT was determined and additional shocks given, pinacidil was administered to 6 pigs and flunarizine to 6 pigs. Then, the DFT was again determined and additional shocks were given. Pinacidil significantly shortened the effective refractory period (ERP) (162 +/- 16 vs 130 +/- 28 msec) and action potential duration (APD(90)) (179 +/- 6 vs 149 +/- 19 msec) and significantly increased the peak frequency of the power spectrum of a left ventricle (LV) electrode during ventricular fibrillation (VF) (9.3 +/- 0.6 vs 10.5 +/- 1.0 Hz), while flunarizine did not significantly alter the ERP (162 +/- 8 vs 167 +/- 18 msec) or APD(90) (187 +/- 12 vs 191 +/- 20) but significantly reduced the peak frequency (9.2 +/- 0.5 vs 7.5 +/- 1.0 Hz). These findings suggest the drugs had their expected electrophysiological effects. However, the DFT was not significantly changed by either drug. Following the same strength shock 10% below the predrug DFT, earliest postshock activation arose in a focal epicardial pattern from the anterior-apical LV both before and after the drugs. The time from the shock until the appearance of this activation was not significantly different before and after either drug. The lack of change in DFT as well as the lack of change in the incidence, location, and timing of the postshock focus with sub-DFT strength shocks before and after pinacidil and flunarizine provide evidence that these foci are not caused by triggered activity.

Enhancing Electrical Cardioversion and Preventing Immediate Reinitiation of Hemodynamically Deleterious Atrial Fibrillation with Class III Drug Pretreatment

Nifekalant for shock-resistant atrial fibrillation. In severely ill patients, the development of atrial fibrillation (AF) may provoke lethal hemodynamic instability requiring immediate electrical defibrillation, which often is unsuccessful. Using the novel potassium channel blocking agent nifekalant, we prospectively assessed the hypothesis that class III antiarrhythmic drugs facilitate electrical cardioversion and suppress the immediate recurrence of hemodynamically deleterious AF. Among 1896 adults admitted to the intensive care unit for cardiovascular diseases, hemodynamically destabilizing new-onset AF (systolic blood pressure<90 mm Hg) resistant to conventional electrical cardioversion occurred in 27 patients, and of these, 24 patients (70+/-12 years) were enrolled. Twenty-one patients had congestive heart failure and 11 patients had been mechanically ventilated. After three failed transthoracic cardioversions due to failure of conversion to SR (11 patients) or immediate reinitiation (13 patients), nifekalant (0.25+/-0.04 mg/kg) was administered intravenously, and electrical defibrillation was reattempted. In 18 patients (75%), sinus rhythm was restored and maintained after nifekalant infusion (6 patients) or subsequent transthoracic cardioversion (12 patients). Nifekalant administration significantly decreased the heart rate and increased systolic blood pressure during AF (P<0.001), and successful cardioversion rapidly further ameliorated these parameters (P<0.001). Logistic regression analysis showed that atrial defibrillation failure (relative risk [RR] 19.34, P=0.05) and age of >75 years (RR 15.25, P=0.03) were independent predictors of in-hospital death. Nifekalant renders electrical defibrillation and the prevention of the early recurrence of hemodynamically unstable AF more successful without deteriorating hemodynamics, and successful defibrillation is associated with a more favorable patient outcome. Pretreatment with other class III drugs, e.g., ibutilide or dofetilide, would also be efficacious in patients with failed urgent electrical cardioversion.

Unsuccessful Internal Defibrillation in Brugada Syndrome: Focus on Refractoriness and Ventricular Fibrillation Cycle Length

In patients with Brugada syndrome, implantable cardioverter defibrillator (ICD) is the only reliable treatment to prevent sudden death though, in some cases, internal defibrillation may be unsuccessful. The aim of this study was to examine the determinants of defibrillation failure, with a focus on electrophysiologic characteristics. The study included 51 patients treated with ICD: 22 with Brugada syndrome and 29 with structural heart disease (SHD). The prevalence of defibrillation energy requirement precluding the programming of a 10-J safety margin, the mean right ventricular effective refractory period (ERP), and mean induced ventricular fibrillation cycle length (VFCL) from the stored ICD electrograms, were compared between the two patient groups. High defibrillation requirements were observed in 18% of patients with Brugada syndrome versus 0% of patients with SHD. However, the patients with SHD had larger heart size than those with Brugada syndrome. Mean VFCL and mean ERP were both significantly shorter in patients with Brugada syndrome than in patients with SHD, and ERP and VFCL were significantly correlated. Patients with Brugada syndrome have a high prevalence of high defibrillation energy requirement, and short ventricular ERP and VFCL.

Value of Pre‐Hospital Discharge Defibrillation Testing in Recipients of Implanted Cardioverter Defibrillators

Opinions vary regarding the need to perform defibrillation testing prior to hospital discharge in recipients of state-of-the-art cardioverter defibrillators (ICDs). Our protocol is to perform predischarge ICD testing 1 day after implant. This report includes 682 consecutive implants. Adverse observations at testing were grouped into (1) risk of defibrillation failure, (2) surgical complications, (3) sensing/pacing issues or narrow defibrillation margin warranting closer follow-up, or (4) findings correctable by device reprogramming. Among the 682 patients, 63% had single-chamber and 37% dual-chamber or biventricular ICDs. In 48 patients (7%) there were 69 concerns and/or interventions, with overlaps among the four categories, including one failure to defibrillate (0.15%), and six other patients at risk. Surgical complications included 11 hematomas (1.6%), and six lead dysfunctions. Closer follow-up was indicated in 19 patients (2.7%), for high pacing thresholds in seven, sensing issues in seven, and <10 J defibrillation margin in five. Device reprogramming was needed in 31 patients (4.5%), for tachycardia detection and therapy settings in 12, and for pacing/sensing functions in 22 patients. In eight patients ventricular fibrillation could not be induced. There was no morbidity or mortality due to testing. The state-of-the-art ICDs delivering biphasic shocks are remarkably reliable. The routine pre-hospital discharge defibrillation testing of such ICDs may be optional and left to the physicians' discretion.

Deaths associated with implantable cardioverter defibrillator failure and deactivation reported in the United States Food and Drug Administration Manufacturer and User Facility Device Experience Database

The aim of this study was to understand the causes of implantable cardioverter defibrillator (ICD) failure and complications so that adverse events, including unnecessary death, can be prevented. Sudden death may occur if an ICD fails to treat life-threatening ventricular arrhythmias. The United States Food and Drug Administration Manufacturer and User Facility Device Experience Database was searched for ICD devices and the search term "death." The search yielded 212 death events involving 100 ICD pulse generator and lead models from five manufacturers. These death events were associated with (A) ICD devices for which pulse generator interrogation data and/or the results of the manufacturers analysis of returned devices were available; (B) ICD devices for which neither interrogation data nor the results of the manufacturer's analysis were reported; and (C) normally functioning ICDs that had been deactivated. (A) A total of 103 (69%) of 150 death events were associated with defective pulse generators or high-voltage leads. Most (34/42 [81%]) apparently sudden or arrhythmic death events were associated with high-voltage lead failure; other deaths were related to pulse generator failure (8/42 [19%]) caused by electronic component defects. (B) A total of 21 of 51 death events were related to a manufacturer's recall; all deaths were arrhythmic but without allegation of device failure. (C) Eleven death events occurred in patients whose pulse generators were found to be off or deactivated; these devices appeared to have been deactivated accidentally or by exposure to magnetic fields, or they were not reactivated after elective surgery. ICD device failure and unintended pulse generator deactivation have resulted in unnecessary deaths. Although these deaths may be infrequent, improved devices and follow-up techniques are needed. The magnet deactivation feature probably is unsafe, and health professionals and patients should be cautioned.

Effects of shock strengths on ventricular defibrillation failure.

The mechanism of defibrillation is controversial. Reentry appearing immediately after the shock has been shown to be responsible for defibrillation failure in some studies while other studies have demonstrated that a rapid train of focal activations with the first focus appearing >50 ms after the shock is responsible for failed defibrillation. We tested the hypothesis that both patterns can occur, but at different shock strengths. Biphasic 6/4 ms shocks of 100-900 V in 100-V increments were given after 10 s of ventricular fibrillation from electrodes in right ventricular apex and right atrium in five isolated pig hearts. Transmembrane activity was optically mapped from the anterior and posterior epicardium using two CCD cameras. The defibrillation threshold (DFT) was 786+/-199 V. The interval from the shock to the earliest post-shock activation was zero for shocks <400 V but increased with increasing shock voltage to 62+/-6 ms at 800 V. The number of post-shock phase singularities, which is related to reentry incidence, decreased continuously from pre-shock values for 100-V shocks to zero as the shock strength increased to 600 V. Focal activations were observed after shocks >600 V with no epicardial reentry present. Reentry is responsible for defibrillation failure for low-strength shocks. As the shock strength approaches the DFT, a focal epicardial activation pattern becomes responsible for failed defibrillation. Thus, the mechanism of defibrillation failure depends on shock strength, with focal activation as the mechanism for the clinically important near-DFT strength shocks.

Postshock arrhythmogenesis in a slice of the canine heart.

Recent evidence has demonstrated that defibrillation shocks terminate or reset reentrant activity in the myocardium through the generation of virtual electrode polarization (VEP). Previous research has revealed that the shock establishes phase singularities (PSs) in the tissue via the VEP mechanism. The aim of this study was to examine, as a function of shock strength and electrode configuration, the relationship between end-shock PSs and the reentrant circuits established after failed defibrillation attempts. The study uses a complex three-dimensional finite-element bidomain model of a slice of the canine heart characterized by realistic geometry and fiber architecture and undergoing a single scroll wave. Defibrillation shocks of increasing strength are delivered through three different electrode configurations. The results demonstrated that >98% of all PSs have a lifetime of half a reentrant cycle or less. Stronger shocks result in a faster rate of annihilation of postshock PSs. For failed shocks, the surviving PSs underlie the activity of one or more scroll waves, which remain stationary in the slice. For all electrode configurations tested, the increase in shock strength leads to a rapid initial increase in the number of postshock reentries followed by a slower decrease; similar behavior is observed with regard to end-shock PSs. These results present new evidence regarding the mechanisms underlying failure of defibrillation shocks.

Persistent automaticity as a mechanism of ventricular defibrillation failure

Persistent automaticity as a mechanism of ventricular defibrillation failure

Shock coordinated with high power of morphology electrogram improves defibrillation success in patients with implantable cardioverter defibrillators.

Animal studies have suggested that the success of defibrillation may depend on the properties of VF waveform obtained from the morphology electrogram (ME) at the time of the shock. The reliable identification of depolarization events in the fibrillatory signal can be achieved using adaptive estimation of the instantaneous signal power (P). The aim of this study was to investigate if a high P of the ME (P(ME)) was related to ventricular DFT and if the upslope in ME can be associated with the depolarization event. A total of 575 VF (mean duration 10 s) episodes recorded and stored during ICD implantation in 77 patients with ventricular arrhythmias were used for analysis. The DFT was defined using a double step-down test. The values of P(ME) immediately before pulse delivery (P(shock)) and shock outcomes were registered. The differences between P(shock) of successful and failed defibrillation were tested with the Mann-Whitney U test. The relationship between individual medians of P(shock) (P(med)) and DFT was analyzed using the Kruskall-Wallis H-test. The coincidence between identified depolarization and upslope in ME was tested using the chi-square test. A P value of 0.05 was set for an error probability. The P(shock) in case of failed defibrillation was significantly lower than P(shock) in successful cases by the pulses of any strength (P < 0.001). The test revealed a significant inverse correlation between P(med) and DFT with P < 0.001. The depolarization corresponded to the upslope of ME in 85% of cases. This study demonstrated that a high value of instantaneous power of ME indicates the optimal time for shock delivery. The implementation of this algorithm in ICDs may improve the defibrillation efficacy.

Mechanisms of shock-induced arrhythmogenesis during acute global ischemia.

Little is known about the mechanisms of vulnerability and defibrillation under ischemic conditions. We investigated these mechanisms in 18 Langendorff-perfused rabbit hearts during 75% reduced-flow ischemia. Electrical activity was optically mapped from the anterior epicardium during right ventricular shocks applied at various phases of the cardiac cycle while the excitation-contraction decoupler 2,3-butanedione monoxime (BDM; 15 mM) was used to suppress motion artifacts caused by contraction of the heart. During ischemia, vulnerable window width increased [from 30-90% of the action potential duration (APD) in the control to -10 to 100% of the APD in ischemia]. Moreover, arrhythmia severity increased along with the reduction of APD (176 +/- 9 ms in control and 129 +/- 26 ms in ischemia, P < 0.01) and increased dispersion of repolarization (45 +/- 17 ms in control and 73 +/- 28 ms in ischemia, P < 0.01). Shock-induced virtual electrode polarization was preserved. Depolarizing (contrary to hyperpolarizing) response time constants increased. Virtual electrode-induced wavefronts of excitation had much more tortuous pathways leading to wavefront fractionation. Defibrillation failure at all shock strengths was observed in four hearts. Optical mapping revealed that the shock extinguished the arrhythmia; however, the arrhythmia self-originated after an isoelectric window of 339 +/- 189 ms. In conclusion, in most cases, virtual electrode-induced phase singularity (VEIPS) was responsible for shock-induced arrhythmogenesis during acute global ischemia. Enhancement of arrhythmogenesis was associated with an increased dispersion of repolarization and altered deexcitation. In four hearts, arrhythmogenesis could not be explained by VEIPS.

Long-term structural failure of coaxial polyurethane implantable cardioverter defibrillator leads.

Transvene models 6936/6966, a coaxial polyurethane ICD lead, may be prone to structural failure. These models comprise 54% of ICD lead failures in the authors' Multicenter Registry database. Because ICD leads perform a vital function, the clinical features, causes, and probability of Transvene 6936/6966 lead failure were determined. The Registry and United States Food and Drug Administration databases were queried for the clinical features and structural causes of the Transvene 6936/6966 lead failure, and a five-center substudy estimated the survival probability for 521 Transvene 6936/6966 implants. The mean time to failure was 4.8 +/- 2.1 years, and the estimated survival at 60 and 84 months after implant were 92% and 84%, respectively. Oversensing was the most common sign of failure (76%), and 24 patients experienced inappropriate shocks. The manufacturer's reports indicated that high voltage coil fracture and 80A polyurethane defects were the predominant causes of lead failure. Transvene models 6936 and 6966 coaxial polyurethane ICD leads are prone to failure over time. Patients who have these leads should be evaluated frequently. Additional studies are needed to identify safe management strategies.

Cellular mechanism of reentry induced by a strong electrical stimulus: implications for fibrillation and defibrillation.

The objective of this review article is to describe the graded response hypothesis of reentry induced by a strong single electrical stimulus in the normal canine ventricular myocardium. It is shown that the graded responses (subthreshold depolarization during phase 3 of the action potential) induced at a site distant (S2) from the regular S1--S1 pacing site, propagate slowly over short distances (approximately 5 mm) and initiate a regenerative action potentials in recovered cells near the S1 site. Activation wave then blocks near the S2 site (unidirectional block) but reenters when the S2 site recovers it excitability. Super strong S2 currents do not induce reentry (upper limit of vulnerability). Since similar activation sequence and properties are shown to exist in intact canine hearts during induction of ventricular fibrillation with a similar S2 stimulus, the graded response hypothesis may have relevance to vulnerability to fibrillation. Furthermore, since the upper limit of vulnerability is closely related to defibrillation threshold, the graded response hypothesis might also be relevant to defibrillation mechanism. Other proposed mechanisms of fibrillation and defibrillation (critical point hypothesis, the progressive depolarization hypothesis and the hypothesis of phase singularity of defibrillation failure) are also discussed in this review paper and compared to the graded response hypothesis.

Mechanism of VF reinitiation after failed defibrillation shocks: An optical mapping study in isolated swine hearts Nipon Chattipakorn, Isabelle Banville, Richard A. Gray, Raymond E. Ideker. University of Alabama at Birmingham, Birmingham, AL

Mechanism of VF reinitiation after failed defibrillation shocks: An optical mapping study in isolated swine hearts Nipon Chattipakorn, Isabelle Banville, Richard A. Gray, Raymond E. Ideker. University of Alabama at Birmingham, Birmingham, AL

Virtual electrode polarization in the far field: implications for external defibrillation.

We recently suggested that failure of implantable defibrillation therapy may be explained by the virtual electrode-induced phase singularity mechanism. The goal of this study was to identify possible mechanisms of vulnerability and defibrillation by externally applied shocks in vitro. We used bidomain simulations of realistic rabbit heart fibrous geometry to predict the passive polarization throughout the heart induced by external shocks. We also used optical mapping to assess anterior epicardium electrical activity during shocks in Langendorff-perfused rabbit hearts (n = 7). Monophasic shocks of either polarity (10-260 V, 8 ms, 150 microF) were applied during the T wave from a pair of mesh electrodes. Postshock epicardial virtual electrode polarization was observed after all 162 applied shocks, with positive polarization facing the cathode and negative polarization facing the anode, as predicted by the bidomain simulations. During arrhythmogenesis, a new wave front was induced at the boundary between the two regions near the apex but not at the base. It spread across the negatively polarized area toward the base of the heart and reentered on the other side while simultaneously spreading into the depth of the wall. Thus a scroll wave with a ribbon-shaped filament was formed during external shock-induced arrhythmia. Fluorescent imaging and passive bidomain simulations demonstrated that virtual electrode polarization-induced scroll waves underlie mechanisms of shock-induced vulnerability and failure of external defibrillation.

Direct evidence of the role of virtual electrode-induced phase singularity in success and failure of defibrillation.

We recently demonstrated that virtual electrode-induced phase singularity is responsible for arrhythmogenesis during T wave shocks and explains the upper and lower limits of vulnerability. Furthermore, we suggested that the same mechanism might be responsible for defibrillation failure. The aim of this study was to experimentally support this hypothesis. We used the voltage-sensitive dye di-4-ANEPPS and fast imaging to assess electrical activity in Langendorff-perfused rabbit hearts. Ventricular arrhythmias were induced by monophasic shocks applied during T wave. Three types of defibrillation shocks (n = 79) were delivered from an intravenous right ventricular electrode: monophasic (8 msec), optimal biphasic (8/8 msec, 2/1 leading-edge voltage ratio), and nonoptimal biphasic (8/8 msec, 1/1 leading-edge voltage ratio). We found that a monophasic shock extinguished arrhythmic pattern of electrical activity via a virtual electrode polarization effect. However, the virtual electrode polarization was likely to produce phase singularities, leading to another arrhythmia and defibrillation failure. Nonoptimal biphasic shocks produced similar effects. Optimal biphasic shocks were successful because the first phase of the shock erased the arrhythmia via the virtual electrodes effect, whereas the second phase canceled the virtual electrodes, eliminating the substrate for phase singularities and arrhythmia resulting from them. Our data provide the first experimental support of the hypothesis implicating virtual electrode-induced phase singularity in defibrillation failure in the Langendorff-perfused rabbit heart. Optimal biphasic shock has a higher defibrillation efficacy because it does not produce virtual electrode-induced phase singularities.