Ventricular Arrhythmias Detected After Transvenous Defibrillator Implantation in Patients With a Clinical History of Only Ventricular Fibrillation
Implications for Use of Implantable Defibrillator
Background Patients with a history of ventricular fibrillation (VF) have been shown to have a clinical profile, response to electrophysiological testing (EPS), and response to antiarrhythmic therapy that distinguishes them from patients with a history of sustained monomorphic ventricular tachycardia (MVT). Despite these differences, it is not clear whether VF in these patients is triggered by MVT or occurs de novo. The incidence of MVT and VF in such patients after their index VF event has important implications for therapeutic decisions regarding implantable defibrillator selection and programming.
Methods and Results The records of 111 consecutive patients who had undergone transvenous cardioverter/defibrillator (ICD) implantation for malignant ventricular arrhythmias were reviewed retrospectively. For each patient, all device tachyarrhythmia detections were examined and classified as VF, MVT, rapid polymorphic VT, or other. The number of events, time to first arrhythmia detection, and cycle length of MVTs were recorded. There were 55 patients with a history of only VF and 56 with a history that included an episode of MVT. Over 14 months of follow-up, with all patients initially off of antiarrhythmic medications, MVT was detected by only 18% of patients with a history of only VF compared with 54% of those with a history that included MVT (P=.002). Among patients who did detect MVT, those with a history of only VF had fewer episodes (7±7 versus 20±31, P=.001) and a shorter mean MVT cycle length (279 versus 314 ms, P=.03) than those with a clinical history of MVT. Abrupt onset of VF not preceded by MVT was detected in 11% of patients with VF only. In addition to a history of MVT, male sex, age <60 years, and MVT inducible on EPS were all significantly associated with an increased likelihood of MVT detection. On multivariate analysis, the inducibility of MVT was the primary independent predictor of MVT detection but was of minimal incremental predictive value in the subgroup of patients with a history of only VF. When EPS results were not considered, arrhythmia history was the primary independent predictor of MVT detection.
Conclusions Patients with a history of only VF infrequently have MVT detected by their defibrillators. When these patients do detect MVT, it is faster than that detected in patients with a clinical history of MVT before ICD surgery. A significant percentage of VF survivors detected the abrupt onset of VF not preceded by MVT, suggesting that the deterioration of rapid MVT to VF is not the only clinically important mechanism of VF induction. These findings may have important implications for the understanding of the mechanism of VF induction and for use of an implantable defibrillator.
Patients resuscitated from ventricular fibrillation (VF) outside the context of an acute Q-wave myocardial infarction have been shown to have a clinical profile that distinguishes them from patients with a history of sustained monomorphic ventricular tachycardia (MVT).1 2 3 4 5 6 7 Compared with patients with a history of MVT, VF survivors tend to have a higher ejection fraction,1 are less likely to have a history of myocardial infarction,1 2 are less likely to have late potentials detected by signal-averaged ECG,3 and are less likely to have VT induced on electrophysiological study (EPS).1 2 3 4 5 6 When VT is inducible in VF survivors, it is faster and more often polymorphic than in patients with a history of MVT.5 6 In addition, there is evidence that the inducibility of MVT does not predict recurrent events in patients with a history of VF as well as it does in those with a history of MVT.5 7 In one retrospective study, VF survivors with inducible MVT (or VF) suppressed by conventional drug therapy did not have a better prognosis than those who were not suppressed.5
Despite these differences, there is a common impression among physicians that VF survivors simply represent a more malignant end of the spectrum of the manifestations of MVT.8 9 10 11 12 13 14 It is postulated that in some patients, MVT-induced hypotension, ischemia, left ventricular dysfunction, or electrical instability eventually causes the rhythm to deteriorate to VF.8 9 10 In favor of this argument are reports of MVT deteriorating to VF during ambulatory ECG monitoring.11 12 13 14 The incidence of spontaneous MVT and VF in an unselected group of patients resuscitated from VF not taking antiarrhythmic medications is unknown. Determining the incidence of these rhythms in such a patient group might provide insights into the mechanisms of cardiac arrest in VF survivors and have important implications for implantable cardioverter/defibrillator (ICD) use. If VF survivors are unlikely to have MVT, an ICD capable of specific VT detection and therapy may be unnecessary. This patient group may need only defibrillation capabilities.
The purpose of this study was to use the memory and therapeutic capacity of the ICD to document the incidence of MVT and VF in a group of patients with a history of only VF and compare them with patients with a history that included at least one episode of sustained MVT.
The records of 111 consecutive patients who had undergone transvenous ICD implantation for malignant ventricular arrhythmias at the University of Washington Medical Center, Seattle, were examined retrospectively. The clinical arrhythmia history before device implantation was determined and defined as either a history of VF recorded at the time of resuscitation from cardiac arrest with no history of sustained MVT or a history that included any episode of sustained MVT. Arrhythmias observed only during EPS were not used to classify patients. All patients presenting with VF were considered candidates for nonthoracotomy ICD implantation, with the exception of patients in whom there was an indication that revascularization would significantly reduce the risk of recurrence, such as evidence that their arrest had occurred in the context of ongoing ischemia or if their coronary anatomy was such that coronary artery bypass graft surgery (CABG) was judged to be indicated independent of the arrhythmia history. The demonstration of some degree of inducible ischemia was not considered an indication for bypass surgery or angioplasty if exertional symptoms were medically controlled and the arrest was not clearly related to ischemia. Patients in whom a proarrhythmic drug reaction or other reversible cause of their malignant arrhythmia was identified were not considered candidates for ICD implantation.
Baseline data gathered for each patient included age, left ventricular ejection fraction, presence of coronary artery disease, history of myocardial infarction, past history of CABG, and results of EPS. EPSs including up to triple extrastimuli at two drive train cycle lengths at two right ventricular pacing sites were performed as a separate procedure before ICD implantation in 51 (91%) of the MVT patients and 28 (51%) of the VF patients. An EPS via the defibrillator consisting of up to triple extrastimuli at two drive train cycle lengths from the right ventricular apex was performed in 98 (88%) of the patients (48 VF and 50 MVT patients) before hospital discharge, usually 3 days after ICD implantation. Six patients did not have EPS performed either before implantation or before hospital discharge and were excluded from the analysis of the predictive value of EPS testing. MVT induced on either of these studies was considered inducibility in the analysis.
Implantable Cardioverter Defibrillator
A tiered-therapy ICD (model 7216A or 7217B PCD, Medtronic) was implanted in all patients via a transvenous lead system incorporating a right ventricular lead, a subcutaneous patch, and a superior vena cava or coronary sinus lead. A report of implantation methods and survival data in many of these patients was published elsewhere.15 This device detects rhythms for treatment based on the cycle length of ventricular activity as detected by the right ventricular lead. The details of our use of the detection, treatment, and memory capacity of this device have been described.16 The VF zone is the programmed heart rate range within which a tachycardia is detected and treated as VF. VF detection and therapy was programmed as on in all patients. The VT zone is the programmed heart rate range within which a tachycardia is detected and treated as VT. The VT zone was activated only in patients with a clinical history of MVT, reproducible MVT during EPS, or frequent episodes of rapid MVT detected in the VF zone. Antiarrhythmic drugs were discontinued at the time of device implantation and were used in an attempt to reduce the frequency of device discharges in only five patients (all in the MVT group) a mean of 6 months after ICD implantation.
Defibrillators were interrogated after all clinically apparent device treatment events and on routine follow-up every 3 months. The date of an event was defined as the date of symptomatic device therapy, or if asymptomatic, the date of device interrogation. The ICD memory contains the cycle length of the 20 cycles before the most recent arrhythmia detection and the 10 cycle lengths after the last therapy delivered. The device does not store electrograms. Arrhythmias were identified on the basis of these cycle length data in conjunction with clinical information. A rapid arrhythmia detected in the VF or VT zone was classified as MVT if the cycle length of the tachycardia did not vary by >30 ms once initiated. More irregular tachycardias, with a mean cycle length <200 ms or those in which 75% of recorded cycle lengths were <260 ms, were defined as VF. Slower irregular tachycardias were defined as polymorphic VT unless atrial fibrillation was strongly suspected on clinical grounds. Rapid atrial fibrillation could be recognized by variations in the cycle length of a tachycardia, with a mean cycle length usually >300 ms in a patient known to have chronic or paroxysmal atrial fibrillation. Sinus tachycardia was suspected in the proper clinical setting when the cycle length of the tachycardia detected equaled the lower limit of the VT zone, suggesting acceleration of the sinus rate into the VT zone. The cycle length of the 10 beats after the last therapy is also recorded by the device and was used to differentiate tachycardia type. For instance, variable cycle lengths persisting after therapy just below the VT zone detection interval would help confirm atrial fibrillation if suspected. Similarly, a regular posttherapy cycle length just below the VT zone cycle length would help confirm sinus tachycardia if suggested by the clinical history and the pretherapy cycle length data. Supraventricular tachycardias, including sinus tachycardia and atrial fibrillation, were eliminated from this analysis.
If on interrogation more than one MVT or VF episode had been detected, then all the episodes were judged to have been of the mechanism determined from the cycle length data available for the most recent tachycardia detected. If both MVT and VF had been detected, then cycle length data would be available only for the last detected tachycardia. Frequently, this occurred when the last episode was detected in the VF zone and resulted in a high-energy shock prompting the interrogation. In most of these instances, examination of the available cycle length data revealed that it had actually been rapid MVT that had precipitated the “VF” detection. In such an instance, the MVT episodes for which no cycle length data were available were presumed to be true MVT detections. If the last episode had been MVT and there were VF episodes for which no cycle length data were available, differentiation between acceleration of MVT by antitachycardia pacing, rapid MVT in the VF zone, and primary VF could not be made, and these episodes were classified as “unknown.” This occurred in eight MVT patients and prevents the accurate estimation of the frequency of primary VF in the MVT patients; it may lead to a slight undercounting of the number of MVT episodes in some MVT patients. The infrequency of MVT in VF survivors was such that this occurred in only two patients. In both patients, primary VF episodes were documented on separate occasions. Similarly, in no VF survivors did VF episodes prevent examination of cycle length data for evaluation of possible MVT in a patient who had not had MVT documented on another occasion. The cycle length of an MVT episode was determined as the mean of the cycle lengths recorded before therapy once MVT was established. Episodes of VT for which cycle length data were not available were not used in the calculation of the mean MVT cycle length.
The χ2 test was used for analysis of categorical variables, and Student’s t test was used for analysis of continuous variables. Life table analysis was performed according to the Kaplan-Meier method. Stepwise Cox regression was used to identify independent predictors of the presence or absence of MVT detection on follow-up. Analysis of the independent predictors of the number of MVT episodes detected was performed by stepwise linear regression.
The clinical characteristics of the study patients are outlined in Table 1⇓. Fifty-five patients had a history of only VF, and 56 had a history that included a clinical episode of MVT (19 of the 56 MVT patients had experienced an episode of VF before ICD implantation). The clinical profiles of the groups are shown in Table 1⇓. The mean follow-up time for both groups was 14 months. Patients with a history of only VF had a higher ejection fraction and a lower frequency of induction of MVT during EPS than patients with a history of MVT.
MVT was detected significantly more often in MVT than in VF patients (Table 2⇓). MVT was detected in 18% (10 of 55) of VF patients and 54% (30 of 56) of MVT patients (P=.002). Life table analysis of the time to first MVT detection revealed that MVT was detected earlier in MVT than in VF patients (Figure⇓). Among patients detecting MVT, patients with a clinical history of MVT had more episodes than VF patients. The mean number of MVTs detected by MVT patients was 20±31 compared with 7±7 in VF survivors (P<.001). Only 5% (3 of 55) of VF survivors detected more than five episodes of MVT compared with 38% (21 of 56) of MVT patients (P=.002). No clinically apparent episodes of sustained MVT not treated by the device occurred in the VF patient group. The mean cycle length of the MVT detected in VF survivors was significantly shorter than that in MVT patients, 279±39 versus 314±43 ms (P=.03). There was no significant difference in MVT detection or MVT cycle length among the patients with a history of MVT as a function of whether they had experienced an episode of VF before ICD implant. Abrupt-onset VF was detected in 11% (6 of 55) of the VF survivors, including 2 patients who also detected MVT. The mean cycle length of these episodes of VF was 194 ms. Polymorphic VT with a mean cycle length of 265 ms was detected in 7% (4 of 55) of the patients with a history of VF only. All but 1 of these 4 patients also detected MVT. The acceleration of MVT with antitachycardia pacing in 14 MVT patients with multiple episodes of MVT hampered accurate determination of the incidence of primary VF and polymorphic VT in the MVT patients. Sinus tachycardia was detected in 3 patients (1 VF and 2 MVT). Atrial fibrillation was detected in 8 patients (1 VF and 7 MVT). True MVT was documented at other times in all but 1 of these 11 patients. Thus, these false detections tended to occur in patients who had detected true ventricular arrhythmias, and in most cases, their VT detection zone had been expanded as a result of frequent episodes of slow MVT.
Clinical Predictors of MVT Detection
The univariate relative risk of MVT detection as a function of baseline characteristics in all patients is presented in Table 3⇓. A clinical arrhythmia history of MVT, age <60 years, male sex, and MVT induced on EPS were associated with an increased risk of MVT detection. A history of coronary artery disease, ejection fraction <30%, and a history of myocardial infarction were not statistically associated with an increased risk of MVT. On stepwise Cox regression, inducibility of MVT on EPS was the most powerful independent predictor of MVT detection (Table 4⇓). After the results of EPS were taken into account, a lower ejection fraction and no history of myocardial infarction were independently associated with a higher risk of MVT detection. If inducibility on EPS is left out of the model, then arrhythmia history becomes the most powerful predictor of MVT detection. Virtually identical results were found when stepwise linear regression was used to determine the independent predictors of the number of MVT episodes on follow-up. If inducibility is included in the model, then it becomes the only independent predictor; otherwise, the arrhythmia history is the only independent predictor. Only 6 of 14 VF survivors (43%) with MVT induced on EPS had MVT detected on follow-up. Furthermore, 4 of the 10 VF survivors who detected MVT had not had MVT induced on EPS; 2 of these 4 patients had undergone EPS both before and after ICD implant. Thus, EPS was of minimal incremental predictive value once the arrhythmia history had been taken into consideration.
These findings lend support to the assertion that patients with a history of only VF should be considered a unique patient population and not simply a subset of MVT patients. In our study, VF survivors were much less likely to detect MVT during follow-up, and those VF survivors who did have MVT had fewer episodes than did patients with a clinical history of MVT before ICD implant. The observation that a significant number of patients with a history of only VF detected abrupt-onset VF not preceded by MVT suggests that in many such patients, VF is not the result of hemodynamic or electrical deterioration of MVT but rather is a primary event.
Our findings can also be interpreted to support the hypothesis that some episodes of VF are the end result of MVT deteriorating to VF. We found that MVT detected in patients with a history of only VF was faster than in patients with a history of only MVT, 279±39 versus 314±43 ms (P=.03). This faster MVT may be more likely to precipitate VF.8 9 10 Because the ICD treated MVTs early in their course, we were not able to document rapid MVT deteriorating to VF and can make no assessment as to the frequency of such an event. Thus, there may be several ECG mechanisms by which VF occurs clinically.
Only a few investigators have examined the differences between patients who present with MVT and those who present with VF. As in our patients, those studies showed that VF survivors tend to have a higher ejection fraction1 and are less likely to have MVT induced with EPS1 2 3 4 5 6 than are MVT patients. Adhar et al1 reported the results of EPS performed on 66 survivors of out-of-hospital VF and compared the results with those in 35 patients presenting with hemodynamically well-tolerated sustained VT. Sustained MVT was induced in only 25% of the VF survivors, compared with 69% of the patients with a history of VT, whereas polymorphic VT or VF was induced in 50% and 12%, respectively. Although, perhaps, fewer of the patients with a history of VT had MVT induced than one might expect,4 the difference in inducibility between them and the VF survivors persists. On multivariate analysis, the inducibility of rapid polymorphic rhythms was the only clinical or EPS variable that independently distinguished the two groups. These findings again suggest that a basic difference exists in the electrophysiology of patients who present with VF compared with those who have had sustained VT.
To investigate the value of EPS-guided therapy in cardiac arrest patients, Poole et al5 reviewed mortality and arrhythmia recurrence rates as a function of the results of EPS in 241 VF survivors followed for 30±15 months. VF was induced in 39 patients (16%) and sustained VT in 66 (27%). The suppression of inducible VT or VF with type 1 antiarrhythmic medications was not associated with either improved arrhythmia-free survival or improved total mortality compared with patients who were not suppressed. One interpretation of these results is that the rhythm induced and suppressed in most of these patients, MVT, was not the rhythm responsible for either the patients’ presenting cardiac arrest or their recurrent episode and that antiarrhythmic drug suppression of MVT inducibility is not a marker of effective therapy in VF survivors. This hypothesis is supported by our present findings that on follow-up after defibrillator implantation, spontaneous MVT is much less common in VF survivors compared with patients who present with MVT and that VF survivors have a relatively high incidence of primary VF not precipitated by MVT.
Clinical Predictors of MVT Detection
Although the inducibility of MVT on EPS was the most powerful independent predictor of the detection of MVT on follow-up in the patient population as a whole, when the analysis was restricted to patients with a clinical history of only VF, the results of EPS were less helpful. In only 6 of 14 VF survivors with MVT induced on EPS was it detected during follow-up, whereas 4 of 10 patients who detected MVT had not had MVT induced with EPS. Thus, the results of EPS were of minimal incremental value in assessing the risk of MVT detection once the clinical arrhythmia history was considered.
Past myocardial infarction has been identified as a marker for the inducibility of MVT in patients undergoing EPS. Similarly, in our study, on univariate analysis both coronary artery disease and a history of myocardial infarction were associated with a trend toward an increased likelihood of MVT detection on follow-up. But by stepwise Cox regression, after correction for ejection fraction and the results of EPS, a history of myocardial infarction was associated with a reduced likelihood of MVT detection. This finding should not be interpreted to suggest that a history of myocardial infarction somehow reduces the risk of MVT on follow-up. More likely, the ejection fraction and the results of EPS are in part surrogate markers for a history of myocardial infarction; once these factors are corrected for the absence of a history of myocardial infarction, they may identify patients with idiopathic nonreentrant MVT who may not be readily induced with EPS but are still at significant risk for spontaneous MVT on follow-up.
We were not able to assess the risk of MVT or VF detection as a function of the degree of inducible ischemia or the severity of coronary artery disease beyond the information contained in the ejection fraction and the history of past myocardial infarction. Our patient population included some patients with inducible ischemia, but we were careful to exclude patients who were judged to have had an ischemic trigger to their presenting arrhythmia. We excluded patients undergoing epicardial ICD implantation with or without CABG because of concerns that the potential proarrhythmic effects of such procedures might bias outcome. Our findings, therefore, cannot necessarily be extrapolated to patients who undergo epicardial ICD implantation or have revascularization as part of their arrhythmia therapy, but they are applicable to patients who have previously undergone CABG (40% percent of our population).
Implications for ICD Use
The observed differences between VF survivors and MVT patients may have significant implications for ICD selection and programming. If VF survivors are unlikely to have MVT on follow-up, then a tiered-therapy ICD capable of antitachycardia pacing and/or low-energy cardioversion may be unnecessary. When VF survivors in our study did have MVT, the cycle length was short (279 ms). Because available defibrillators classify arrhythmias largely on the basis of their cycle lengths, MVTs with cycle lengths <300 ms frequently fall into the programmed VF detection zone. Thus, differentiating rapid MVT from VF becomes a significant problem in patients with fast MVT and further limits the value of implanting a tiered-therapy ICD in VF patients. Finally, as the cycle length of MVT shortens, the likelihood of success of antitachycardia pacing therapy decreases.17 Thus, the infrequent occurrence and the rapid rate of the MVT detected in patients with a history of only VF may significantly limit the utility of tiered-therapy ICDs in this patient group. On the other hand, the results of EPS were of minimal incremental value in predicting the occurrence or frequency of MVT in VF survivors after consideration of clinical arrhythmia history. This makes EPS of doubtful utility in assessing the value of implanting a tiered-therapy ICD versus a shock-only ICD in VF survivors. Any cost savings achieved by using a shock-only device would be offset by the cost of EPS used to determine the type of ICD to implant.
Several aspects of this study may limit the power of our conclusions. First, the ICD used stores cycle length data but does not store electrograms or time of therapy. The classification of arrhythmias on the basis of these cycle length data is most likely to cause errors in determining the timing and number of episodes per patient and in differentiating rapid atrial fibrillation from polymorphic VT. We did not use polymorphic VT as a major end point and classified detected tachycardias as atrial fibrillation in only eight patients. All but one of these eight patients had true MVT documented at another time. Furthermore, the relative infrequency of events in VF survivors combined with routine interrogation, usually within 24 hours of clinical events in all patients, led us to be able to examine cycle length data such that an ensuing episode of VF or other arrhythmia did not prevent examination of a possible episode of MVT in any patient who had not otherwise had MVT documented.
A second potential limitation is that our assessment of the predictive value of EPS may have been affected by the fact that EPS was not performed preoperatively in our later VF patients. Instead, single-site EPS was performed through the ICD in the first post-operative week. Although 20 (40%) of the VF survivors and 4 (7%) of the patients with a history of MVT included in the analysis of EPS had EPS performed at only a single site at a postoperative study, only 1 patient from each group detected MVT after a negative single-site EPS study. Thus, although the yield of pacing at a single site is less than that of pacing at two sites,4 we do not feel that this difference in yield had a significant impact on our analysis of the predictive value of EPS for MVT detection either in MVT or in VF patients. We have chosen not to routinely perform EPS before ICD implantation in patients presenting with VF because the results of EPS in such patients have not proved to provide information that is critical to therapeutic decision making.18 The yield of EPS for finding otherwise unsuspected, treatable causes of VF, such as inducible rapid supraventricular tachycardia, has been very low in patients without other historical or clinical factors to suggest it. Moreover, the results of the CASCADE study19 suggested that EPS-guided drug therapy in VF survivors is inferior to empirical amiodarone therapy, making the demonstration of inducible arrhythmias relatively unimportant in therapeutic decision making. In addition, we have not found the success or failure of attempts at MVT pace termination in the electrophysiology laboratory to be predictive of the efficacy of ICD antitachycardia pacing therapies after device implantation.
Another potential limitation of our study is related to our population. As a referral center in an area served by an emergency medical system with rapid response times, it is possible that our patient group may include more true spontaneous VF survivors than would be found in other areas in which prompt resuscitation is not available. In areas with less rapid emergency response, more survivors of out-of-hospital VF may have had MVT that, over time, deteriorated to VF. If this is the case, physicians in other regions may find a higher incidence of spontaneous MVT in VF patients. Last, the 14-month follow-up time is relatively short; longer follow-up might have resulted in a higher or lower relative incidence of MVT and VF.
The results of this study strongly suggest that after ICD implantation, MVT is much less likely to be detected in patients with a history of only VF than in those with a clinical history of MVT. This lower incidence of MVT detection, the shorter cycle length of the MVT that is detected, and the detection of spontaneous VF in patients with a history of only VF support the concept that these patients form a unique population that should be distinguished from patients with a history of MVT. Furthermore, the results of EPS were found to be of minimal value in predicting the detection of MVT in VF-only patients. These findings may have important implications for understanding the natural history of patients resuscitated from VF, the value of EPS before defibrillator implantation, and the type of defibrillator best suited for certain patient groups.
The authors thank Charles Troutman and Jill Anderson for their nursing care of the patients.
- Received July 7, 1994.
- Revision received October 17, 1994.
- Accepted November 6, 1994.
- Copyright © 1995 by American Heart Association
Josephson ME. Recurrent ventricular tachycardia. In: Josephson ME. Clinical Cardiac Electrophysiology: Techniques and Interpretations. Philadelphia, Pa: Lea & Febiger; 1993:181-274.
Poole JE, Mathisen TL, Kudenchuk PJ, McAnulty JH, Swerdlow CD, Bardy GH, Greene HL. Long-term outcome in patients who survive out of hospital ventricular fibrillation and undergo electrophysiologic studies: evaluation by electrophysiologic subgroups. J Am Coll Cardiol. 1990;16:657-665.
Stevenson WG, Brugada P, Waldecher B, Zehender M, Wellens HJJ. Clinical, angiographic and electrophysiologic findings in patients with aborted sudden death as compared with patients with sustained ventricular tachycardia after myocardial infarction. Circulation. 1985;71:1146-1152.
Lima JAC, Weiss JL, Guzman PA, Weisfeldt ML, Reid PR, Traill TA. Incomplete filling and incoordinate contraction as mechanism of hypotension during ventricular tachycardia in man. Circulation. 1983;68:928-938.
Gradman AH, Bell PA, DeBusk RF. Sudden death using ambulatory monitoring. Circulation. 1977;55:210-211.
Lahiri A, Balasubramian V, Rafferty EB. Sudden death during ambulatory monitoring. Br Med J. 1979;1:1676-1678.
Nikolic G, Bishop RL, Singh JB. Sudden death recorded during Holter monitoring. Circulation. 1982;66:218-225.
Bardy GH, Hofer B, Johnson G, Kudenchuk PJ, Poole JE, Dolack GL, Geleva M, Mitchel R, Kelso D. Implantable transvenous cardioverter-defibrillators. Circulation. 1993;87:1152-1168.
Bardy GH, Troutman C, Poole JE, Kudenchuk PJ, Dolack GL, Johnson G, Hofer B. Clinical experience with a tiered-therapy multiprogrammable antiarrhythmia device. Circulation. 1992;85: 1689-1698.
Dolack GL, Poole JE, Kudenchuk PJ, Johnson G, Mitchell R, Bardy GHQ. Management of ventricular fibrillation without baseline electrophysiologic testing off antiarrhythmic drugs using transvenous implantable cardioverter-defibrillators. Circulation. 1993;88(suppl I):I-167. Abstract.