Author + information
- Received November 24, 2004
- Revision received March 1, 2005
- Accepted March 10, 2005
- Published online December 20, 2005.
- Angel R. León, MD⁎,⁎ (, )
- William T. Abraham, MD†,‡,
- Anne B. Curtis, MD§,
- James P. Daubert, MD∥,
- Westby G. Fisher, MD†,¶,
- John Gurley, MD†,
- David L. Hayes, MD#,
- Randy Lieberman, MD⁎⁎,
- Susan Petersen-Stejskal, BS††,
- Kevin Wheelan, MD‡‡,
- MIRACLE Study Program
- ↵⁎Reprint requests and correspondence:
Dr. Angel R. León, Cardiology, MOT 6th Floor, Emory Crawford Long Hospital, 550 Peachtree Street, Atlanta, Georgia 30308
Objectives The purpose of this study was to evaluate the safety of implanting a cardiac resynchronization therapy (CRT) system.
Background Clinicians and patients require data on the safety of the CRT implant procedure to estimate procedural risk.
Methods We evaluated outcomes of transvenous CRT system implantation in 2,078 patients from the Multicenter InSync Randomized Clinical Evaluation (MIRACLE) study, the MIRACLE Implantable Cardioverter-Defibrillator (ICD) study, and the InSync III study. We compared the MIRACLE study to the InSync III study and the MIRACLE ICD study randomized phase to its general phase to evaluate the effect of new technologies.
Results The implant attempt succeeded in 1,903 of 2,078 (91.6%) patients. Implant time decreased from 2.7 h in the MIRACLE study to 2.3 h in the InSync III study (p < 0.001), and from 2.8 h in the MIRACLE ICD study randomized phase to 2.4 h in the general phase (p < 0.001). The implant procedure produced 62 perioperative complications in 53 (9.3%) MIRACLE trial patients; 159 in 135 (21.1%) MIRACLE ICD study randomized phase patients and 71 in 62 (13.9%) general phase patients (p < 0.05 vs. randomized); and 41 in 37 (8.8%) InSync III study patients (p = NS vs. the MIRACLE study). We observed 73 postoperative complications in 62 (11.7%) MIRACLE trial patients, 77 in 68 (11.9%) MIRACLE ICD study randomized phase patients and 56 in 45 (11.0%) general phase patients (p = NS), and 37 in 34 (8.6%) InSync III study patients (p = NS). A total of 8% of patients required reoperation to treat lead dislodgement, extracardiac stimulation, or infection during follow-up.
Conclusions Transvenous CRT system implantation appears safe, well-tolerated, has a high success rate, and improves with operator experience and the addition of new technologies.
Cardiac resynchronization therapy (CRT) provides effective adjunctive therapy for patients with symptomatic heart failure and a wide QRS complex who remain symptomatic optimal pharmacological therapy. However, few large scale studies analyze the safety of implanting a CRT system in such chronically ill individuals.
Cardiac resynchronization therapy device implant originally combined a surgical approach to direct left ventricular (LV) lead placement with transvenous insertion of the right atrial (RA) and right ventricular (RV) pacing leads (1,2). The surgical approach, requiring general anesthesia and thoracotomy, assured successful LV lead implantation but also introduced increased morbidity and mortality (3). The subsequent introduction of a fully transvenous approach to CRT provided a potentially safer alternative to surgical LV epicardial lead placement (1,4). Early transvenous CRT implants utilized standard RV leads adapted to LV pacing until the development of specialized LV leads and delivery systems provided dedicated leads for CRT. The safety and efficacy of implanting a transvenous CRT system using these leads and delivery systems remains largely unknown, as most published clinical trials to date have provided limited data to adequately assess their use.
Given the potentially large number of patients who might benefit from CRT, patients and clinicians should have access to ample data on the safety and efficacy of the implant procedure to create reasonable expectations of procedural risk. To better understand these risks, we evaluated the success and complication rates for transvenous CRT system implantation in over 2,000 patients from three completed large clinical studies of CRT (collectively referred to as the MIRACLE study program): the Multicenter InSync RAndomized CLinical Evaluation (MIRACLE) study, the MIRACLE Implantable Cardioverter-Defibrillator (ICD) study, and the InSync III study. Here we report the outcomes of the implant procedure, including the safety and effectiveness of the implant technique, and the performance of the components of the biventricular pacing system from the time of implant through the six-month follow-up. We also looked for a learning curve within the MIRACLE study, and compared implant success in the MIRACLE study to the InSync III study and in the MIRACLE ICD study randomized phase to the MIRACLE ICD study general phase to evaluate the effect of operator experience and new technologies on implant efficacy.
The rationale, design, and primary heart failure efficacy results of the MIRACLE and MIRACLE ICD studies appear in print elsewhere (5,6). The InSync III study, a prospective, multicenter, nonrandomized trial, evaluated the safety and efficacy of sequential biventricular pacing. The InSync III study patients received an InSync III system with active CRT. The InSync III study used the MIRACLE study “off” (control) group as the control for comparing the efficacy of the InSync III system (7).
Patients in all three studies underwent testing after hospital discharge and at one month, three months, six months, and every six months thereafter. Each study enrolled patients with LV dysfunction (LV ejection fraction ≤35%), New York Heart Association (NYHA) functional class III or IV heart failure despite use of a stable and optimal heart failure drug regimen, and ventricular dyssynchrony defined by a QRS duration ≥130 ms. Patients gave written informed consent under guidelines approved by the institutional review board of each participating center.
Patients enrolled in the MIRACLE and InSync III studies underwent implantation of a CRT pacing device (InSync or InSync III, Medtronic Inc., Minneapolis, Minnesota); those enrolled in the MIRACLE ICD study received a combined CRT-ICD system (InSync ICD, Medtronic Inc.). The MIRACLE ICD study enrolled patients with a standard ICD indication in two phases: a randomized placebo-controlled phase, which included NYHA functional class II patients, and a general phase that continued the safety evaluation with all patients receiving CRT. The general phase included the additional optional use of the Model 4193 over-the-wire LV lead (Medtronic Inc.). The three studies excluded patients who met standard indications for pacing.
All patients without a preexisting device underwent a transvenous insertion of RA, RV, and LV electrodes and implantation of a CRT. Patients with an existing ICD who met MIRACLE CRT indications underwent an upgrade to CRT-ICD using, when possible, the existing right heart leads. The studies utilized various left-heart leads and left-heart delivery systems (Table 1).The three CRT systems used conventional commercially available RA and RV electrodes.
The defined perioperative period included the day of implant plus the seven subsequent days. The postoperative period included day eight through six months. The predetermined definition of a complication included any event requiring intravenous fluid or medication administration or any invasive intervention.
The procedure took place in either a cardiac electrophysiology laboratory, cardiac catheterization laboratory, or operating suite. Implanter training included the presence of an experienced implanter at each newly established investigational center to mentor the center’s first implant. Conscious sedation with local anesthesia or general anesthesia provided patient comfort and analgesia. The implant began with the establishment of central venous access by subclavian vein puncture and/or cephalic vein cut down for insertion of RA and RV leads and the LV lead delivery system. Commercially available pacing lead introducers facilitated access into the venous circulation. The implanters used standard stylet-driven insertion technique to place the atrial lead in the right appendage, free wall, or septal locations depending on operator preference. The study designs encouraged placement of RV leads at the apex, but also accepted a septal location. Transvenous placement of the LV lead required cannulation of the coronary sinus (CS) with a specialized delivery system consisting of a straight or preshaped CS guide catheter, a flexible guidewire, and a hemostasis valve. In cases where the guide catheter alone failed to enter the CS, cannulation of the CS by a deflectable electrophysiologic mapping catheter or preshaped angiographic catheter inserted through the guide catheter provided access into the CS. The protocol required contrast venography of the CS to delineate the cardiac venous anatomy and identify potential target veins for LV lead insertion (Fig. 1A).The implanter selected the target vein based on the visualized anatomy. Acquisition and storage of venogram images in multiple fluoroscopic projections provided a reference of venous anatomy to guide the LV lead insertion procedure. The anterior-lateral, lateral, or posterior-lateral veins served as primary targets in order to maximize RV and LV electrode separation (Fig. 1B). Implanters selected secondary targets only when venous anatomy, stimulation threshold, stability, or extracardiac (diaphragmatic) stimulation prevented lead delivery to the primary site. The implanting physician designated the location of the LV tip electrode to be recorded at the time of implantation. Each study provided two or three LV lead options (Table 1). When successful delivery of the LV electrode produced acceptable stimulation thresholds (≤3.0 V) in the absence of extracardiac stimulation, removal of the lead delivery system preceded the determination of final electrical implant parameters. The inability to deliver an LV lead into any venous branch defined a failed implant attempt.
A pacing system analyzer (Medtronic Analyzer model 8090, Medtronic Inc.) measured the capture thresholds, R-wave and P-wave amplitudes, and pacing impedance during overdrive pacing through the RA, RV, and LV electrodes. Successful LV lead implantation led to connection of the three electrodes to the CRT device. A final interrogation of the implanted system at the time of hospital discharge confirmed satisfactory capture thresholds for each electrode. The follow-up scheme included comprehensive device interrogation to assess lead performance at one, three, and six months for all patients regardless of assignment to control or CRT.
The primary investigator initially classified the cause of death in patients who died in each study. An adverse Events Review Committee (AERC) independently reviewed each death and served as the final arbiter of the cause.
The MIRACLE trial generated a learning curve by providing data from a center’s first implants without the confounding effects of participation in previous CRT trials or experience with commercialized lead and delivery systems. The learning curve analysis compared implant success within the MIRACLE study, average implant time, and fluoroscopy time as a function of center experience. The centers appear grouped by the number of their implant attempts (1 to 5 attempts, 6 to 10 attempts, and ≥11 attempts). A second analysis compared implant outcomes between chronologically sequential studies in order to determine the effect of increasing operator experience and introduction of new technology on implant outcomes. That analysis compares total procedure duration, time to CS cannulation, and time to first acceptable lead position in the randomized and general phases of the MIRACLE ICD study, and between the MIRACLE and the InSync III studies.
The specific device-related electrical data appear as mean ± SD. We calculated the implant success rate as the number of successful implants divided by the total number of patients undergoing a procedure with an exact two-sided 95% confidence interval. The chi-square test compared differences in the implant success rates and the percentage of patients with complications between the groups. The Wilcoxon rank-sum test compared the differences in the average implant and fluoroscopy times. The Kaplan-Meier method calculated the probability of 30-day mortality. All patients undergoing an attempted implant entered the perioperative analysis of complications; all patients with successfully implanted devices proceeded to postoperative analysis of freedom from device-related (LV lead, system) complications at six months. Values of p < 0.05 denote significant differences.
All together the three studies enrolled a total of 2,078 patients who underwent an attempted CRT system implant at 98 centers throughout the U.S. and Canada with an average of 21 implants per center. Table 1lists the baseline characteristics of this population. The MIRACLE study enrolled patients from November 1998 to December 2000, the MIRACLE ICD study randomized phase from October 1999 to August 2001, the MIRACLE ICD study general phase from August 2001 through June 2002, and the InSync III study from November 2000 to June 2002.
The CRT device implant procedure succeeded in 1,903 of 2,078 (91.6%) patients, with 35 (1.8%) requiring more than one attempt to achieve success. The reasons most commonly cited for unsuccessful implants include inability to access the CS ostium (n = 69), acute dislodgement or unstable lead position (n = 59), or inability to obtain a distal lead position (n = 52). The available final LV lead position in 1,859 (98%) of the patients include 794 (42.7%) in the lateral vein, 547 (29.4%) in the posterior-lateral, 110 (5.9%) in the posterior, 253 (13.6%) in the anterior, 92 (5.0%) in the anterior-lateral, and 63 (3.4%) in the middle cardiac vein. The LV lead utilized in successful implants included the stylet-driven model 2187 lead in 676 (35.5%) patients, the stylet-driven model 4189 in 475 (25.0%) patients, the over-the-wire model 4193 (Medtronic Inc.) in 722 (37.9%) patients, and the stylet-driven model 2188 in 27 (1.4%) patients. A successful implant attempt could have used more than one lead. The implant success rates (95% confidence interval) for each of the studies were: the MIRACLE study, 92.5% (90.0 to 94.5); MIRACLE ICD study randomized phase, 89.2% (86.5 to 91.5); MIRACLE ICD study general phase, 91.5% (88.5 to 94.0); and the InSync III study, 94.1% (91.4 to 96.1).
The MIRACLE ICD study included 357 patients with pre-existing ICD devices for upgrade to CRT-ICD. The upgrade procedure succeeded in 320 patients (89.6%).
The MIRACLE study implant success rate for the first 5 implant attempts at each center averaged 89.6% (84.2 to 93.6), increasing to 94.6% (89.2 to 97.8) for the second 5 implant attempts, and continuing at 93.4% (89.7 to 96.1) for all implant attempts after the 10th. Compared with the first five attempts per center, total procedure time decreased significantly with increased experience as shown in Figure 2.
The overall median implant time in the MIRACLE study program measured 2.6 h. The median implant duration (25th, 75th interquartiles) in the MIRACLE study measured 2.7 (2.1, 3.6) h; 2.8 (2.2, 3.7) h in the MIRACLE ICD study randomized phase; 2.4 (1.8, 3.1) h in the MIRACLE ICD study general phase; and 2.3 (1.8, 3.0) h in the InSync III study. Implant time decreased significantly in the InSync III study (p < 0.0001) compared to the MIRACLE study, and in the MIRACLE ICD study general phase to the randomized phase (p < 0.0001).
Median time to CS cannulation decreased from 11.0 (5.0, 22.0) min in the MIRACLE study to 6.5 (3.0, 15.0) min in the InSync III study (p < 0.001) and from 8.0 (4.0, 15.0) min in the MIRACLE ICD study randomized phase to 5.0 (3.0, 12.0) min in the MIRACLE ICD study general phase (p < 0.001). Median time to first acceptable LV lead position decreased from 16.0 (8.0, 40.0) min in the MIRACLE study to 10.5 (5.0, 22.0) min in the InSync III study (p < 0.001) and from 12.0 (6.0, 26.0) min in the MIRACLE ICD study randomized phase to 10.0 (5.0, 21.0) min in the MIRACLE ICD study general phase (p < 0.001). Median fluoroscopy time decreased from 33.0 (19.0, 55.0) min in the MIRACLE study to 27.6 (16.5 47.0) in the InSync III study (p <0.05) and from 33.9 (20.4, 51.4) min in the MIRACLE ICD study randomized phase to 28.9 (18.1, 45.5) min in the MIRACLE ICD study general phase (p < 0.005).
The LV lead was implanted via subclavian vein puncture (85%), axillary vein puncture (8%), or cephalic vein cut down (7%). At implant, the mean LV lead pacing threshold measured 1.5 V at a 0.5-ms pulse width; mean LV lead impedance measured 747 ohms; and the mean R-wave amplitude sensed from the LV measured 13.5 mV. The LV pacing thresholds for the entire group changed from 1.65 ± 1.23 V before discharge to 1.73 ± 1.17 V at the six-month follow-up. Figure 3shows the LV lead pacing thresholds from each lead model over time. Voltage threshold values rose early and then decreased over time. The steroid-eluting Models 4189 and 4193 demonstrated no apparent peak during the immediate postimplant period and more consistent threshold over time. An analysis of all acute RA lead data demonstrated a mean pacing threshold of 1.1 V at a pulse width of 0.5 ms, a mean impedance of 553 ohms, and a mean P-wave amplitude of 3.2 mV. Right ventricular lead data included a mean pacing threshold of 0.9 V at a pulse width of 0.4 ms, a mean impedance of 629 ohms, and mean sensed R-wave amplitude of 14.6 mV.
Table 2lists all lead-, system-, and procedure-related perioperative complications separately for each study and for the total MIRACLE study program. We observed a total of 333 perioperative complications in 287 patients, and 243 postoperative complications in 209 patients.
The AERC classified a total of eight (0.3%) deaths as procedure-related. Two of the deaths occurred in the MIRACLE study. One patient developed progressive hypotension after implant and died later the same day; the other experienced asystole, required cardiopulmonary resuscitation, did not recover neurologic function, and died one month later. Five deaths occurred during the randomized phase of the MIRACLE ICD study: two as the result of progressive heart failure (one with respiratory and renal failure, the other with severe mitral valve disease), one from tachyarrhythmia, one from a bradyarrhythmic event, and one from an acute coronary syndrome. The one procedure-related death in the InSync III study resulted from tachyarrhythmias initiated during the implant.
A total of 45 patients experienced CS dissections or perforations, cardiac perforations, or cardiac vein dissection or perforation; 21 of the 45 (46.7%) still received a CRT system during the same implant session or at a later attempt. Eight patients required intervention for presumed or confirmed hemopericardium: two received intravenous pressor infusion, four underwent pericardiocentesis, and two required both pericardiocentesis and intravenous fluid or pressor infusion. All patients recovered without further sequelae and continued in the studies. Of the total 45 CS traumas, seven occurred during use of the Model 2187 or 2188 lead, eight with the model 4189, and two with the model 4193. The remaining events resulted from use of other implant access catheters or wires. Seventeen patients in the MIRACLE ICD study received only a standard ICD system after abandonment of the LV lead implant. Insertion of the RA or RV leads produced five cardiac traumatic events: one required pericardiocentesis and lead replacement, and two resolved with lead replacement or repositioning. One RA perforation occurred when the screw of the lead penetrated the anterior atrial wall and punctured the aorta leading to surgical exploration and repair. Table 3lists lead, device or system, and procedure-related postoperative complications.
Thirty-three patients died during the first 30 days after implant, yielding a 30-day mortality rate of 1.5%. The implant procedure led to 8 deaths, 10 resulted from progressive heart failure, 11 from sudden cardiac death (SCD), and 4 from noncardiac causes. An additional 101 patients (4.9%) died between day 31 and the 6-month follow-up, including 51 from progressive heart failure, 28 from SCD, 13 from noncardiac causes, and 9 that remained unclassified.
One hundred sixty-one (7.7%) patients required 175 lead revisions either by repositioning or replacement during the six-month follow-up period. The reasons for revision included dislodgement, loss of LV capture without obvious dislodgement, and uncorrectable extracardiac stimulation. An additional 20 (1.0%) patients developed a pocket infection during the six-month follow-up period, of which 13 required explantation of the device. Of the 13, 9 underwent successful reimplantation of a new system.
The MIRACLE ICD study evaluated the InSync ICD device that incorporates CRT into the ICD. Thirty-two patients in that study experienced a total of 208 shock-receiving episodes for reasons other than actual ventricular tachycardia or fibrillation: 14 patients received 44 shocks for sinus tachycardia, 9 received 133 shocks for atrial fibrillation or flutter with rapid ventricular rates, 7 received 18 shocks for supraventricular tachycardia, and 3 received 13 shocks for noise, electromagnetic interference, or oversensing.
The collective MIRACLE study program represents the largest reported experience of CRT system implant outcomes and demonstrates the safety and efficacy of the transvenous LV lead insertion technique for providing CRT to patients with moderate-to-severe chronic heart failure. The implant technique, the lead delivery systems, and the components of the CRT system produced a high implant success rate (91.6%) and acceptable and stable device and LV lead electrical performance in the large majority of patients. The LV pacing thresholds remained stable over time.
Our predetermined definition of a complication includes any event requiring intravenous or invasive therapy, ranging from intravenous saline bolus or analgesia to cardiac surgery. Adopting this broad definition determined that procedural complications occurred in 496 of 2,078 (23.8%) patients undergoing implant attempts. The incidence of procedure-related deaths (8 of 2,078 or 0.4%) in patients with severe LV dysfunction and poor functional status underscores the safety of the technique in a compromised patient population. The observed 30-day mortality of 1.5% in this population compares favorably with other clinical evaluations of device therapy in patients with heart failure and LV dysfunction (8). Additionally, the incidence of infection, pocket hematoma, pneumothorax, or hemothorax fell within the current profile of perioperative complications of conventional pacing and ICD system implantations (9,10).
Coronary sinus trauma associated with catheterization by the guiding catheter, electrophysiology catheters, or balloon-tipped venogram catheters represents a previously unreported and undesirable outcome of conventional cardiac instrumentation and pacing lead implantation. Conceivably routine use of venography during the LV lead implant process reveals CS trauma that would have otherwise gone unrecognized. Most cases of CS perforation or branch vein perforation associated with guiding catheter or lead trauma may remain clinically silent because the low-pressure venous system does not readily lead to tamponade. The adipose tissue and pericardium overlying the atrial-ventricular groove likely encases and contains the low-pressure bleed. We observed that intimal tears, overt dissection of the CS and LV vein, or CS perforation produced no clinical sequelae in most patients; 6 of 45 patients experiencing CS trauma or cardiac perforation required pericardial drainage, none required surgical exploration, and no patient died as a result. Although the clinical sequelae from CS trauma in this series appear infrequently, the complications from perforation or dissection still pose a potential life-threatening risk to these compromised patients.
Postoperative LV lead dislodgement represents a limitation of transvenous CRT systems in comparison to established RA and RV lead performance. The 5.8% lead dislodgement observed here does not differ significantly from those reported in other studies of CRT (11,12). The observed frequency of extracardiac stimulation and exit-block might represent additional microdislodgement of the LV lead that led to a 7.7% requirement for reoperation. Prior experience with right heart pacing leads suggests that dislodgement rates may decrease with improved technology. Dislodgement rates with first generation RA leads ranged up to 18% (13). More recent reports of RA and RV lead dislodgement rates range between 1.5% to 3.3% (9,10). However, the management of extracardiac stimulation and prevention of LV lead dislodgement present a continuing challenge to implanters and the device industry.
Combining CRT with an ICD did not introduce additional risk of arrhythmia or unnecessary shocks to patients receiving these devices. The occurrence of inappropriate ICD shocks in the InSync-ICD study did not differ from that seen in a clinical evaluation of a dual-chamber pacing-defibrillator (14). Adding CRT to the ICD did not demonstrate any increased incidence arrhythmia or create any apparently new mechanism of proarrhythmia, nor did it interfere with the efficacy of ICD therapy (15).
The overall perioperative (13.8%) and postoperative (10%) complications appear greater than the complication rates seen with conventional pacing and defibrillating devices. However, the reported complication rates in the large single-study experience in the Comparison of Medical Therapy, Pacing and Defibrillation in Heart Failure (COMPANION) trial agree with those we report in the MIRACLE program (16). The cumulative risk of reoperation approached 9% during follow-up (lead dislodgement and extracardiac stimulation [7.7%] and infection [1%]). Also, one should recognize that CS trauma, LV lead dislodgement, and extracardiac stimulation requiring repeat operation represent three major sources of increased complications inherent to CRT device implant not seen in conventional device implantation.
The experience reported here documented the presence of a learning curve within a study and from study to study. Implant success increased with experience, while procedure and fluoroscopy time decreased for first implanters in the MIRACLE study. The initial phase of the learning curve appears to flatten after 10 implant attempts. Additional implanter experience combined with improvements in lead delivery systems and introduction of over-the-wire lead technology contributed to increased implant success and decreased procedure time in the InSync ICD and InSync III studies, probably by decreasing time to CS cannulation and LV lead positioning. We observed a mixed relationship between complication rates and experience: implant-related complications (Table 2) decreased significantly from the initial to the latter phase of the MIRACLE ICD study, but not from the initial the MIRACLE study to the subsequent InSync III study.
The MIRACLE program experience suggests that transvenous implantation of LV pacing leads provides an effective, well-tolerated approach to biventricular pacing and defines a risk profile to set physician and patient expectations: the implant attempt succeeded in approximately 90% of patients with a 30-day all-cause mortality of 1.5% and an additional 9% chance of needing a second surgical procedure for LV lead dislodgement, extra-cardiac stimulation, or infection after an initially successful implant.
The authors express their gratitude to two Medtronic employees, Cindy Geng, MS, for her statistical assistance, and Jane Moore, BA, RCIS, for her editing assistance in the preparation of this manuscript.
This study was supported by Medtronic, Inc., Minneapolis, Minnesota.
- Abbreviations and Acronyms
- Adverse Events Review Committee
- cardiac resynchronization therapy
- coronary sinus
- implantable cardioverter-defibrillator
- left ventricle/ventricular
- Multicenter InSync Randomized Clinical Evaluation
- New York Heart Association
- right atrial/atrium
- right ventricle/ventricular
- sudden cardiac death
- Received November 24, 2004.
- Revision received March 1, 2005.
- Accepted March 10, 2005.
- American College of Cardiology Foundation
- Daoud E.G.,
- Kalbfleisch S.J.,
- Hummel J.D.,
- et al.
- Young J.M.,
- Abraham W.T.,
- Smith A.L.,
- et al.,
- Multicenter InSync ICD Randomized Clinical Evaluation (MIRACLE ICD) Trial Investigators
- León A.R.,
- Abraham W.T.,
- Brozena S.,
- et al.
- Higgins S.L.,
- Hummel J.D.,
- Niazi I.K.,
- et al.
- Wilkoff B.L.,
- Hess M.,
- Young J.B.,
- Abraham W.T.
- MIRACLE ICD Investigators,
- Leon A.R.,
- Young J.B.,
- Abraham W.T.,
- et al.
- Bristow M.R.,
- Saxon L.A.,
- Boehmer J.,
- et al.,
- Comparison of Medical Therapy, Pacing and Defibrillation in Heart Failure (COMPANION) Investigators