Design and Results of the Antiarrhythmics vs Implantable Defibrillators (AVID) Registry
Background—The Antiarrhythmics Versus Implantable Defibrillators (AVID) Study compared treatment with implantable cardioverter-defibrillators versus antiarrhythmic drugs in patients with life-threatening ventricular arrhythmias (VAs). AVID maintained a Registry on all patients, randomized or not, with any VA or unexplained syncope who could be considered for either of the treatment strategies. Trial-eligible arrhythmias were the categories of VF cardiac arrest, Syncopal VT, and Symptomatic VT, below.
Methods and Results—Of 5989 patients screened, 4595 were registered and 1016 were randomized. Mortality follow-up through 1996 was obtained on the 4219 Registry patients enrolled before 1997 through the National Death Index. Crude mortality rates (mean±SD, follow-up, 16.9±11.5 months) were: VF cardiac arrest, 17.0% (n=1399, 238 deaths); Syncopal VT, 21.2% (n=598, 127 deaths); Symptomatic VT, 15.8% (n=1065, 168 deaths); Stable (asymptomatic) VT, 19.7% (n=497, 98 deaths); VT/VF with transient/correctable cause, 17.8% (n=270, 48 deaths); and Unexplained syncope, 12.3% (n=390, 48 deaths).
Conclusions—Patients with seemingly lower-risk or unknown-risk VAs (asymptomatic VT, and VT/VF associated with a transient factor) have a (high) mortality similar to that of higher-risk, AVID-eligible VAs. The similar (and poor) prognosis of most patients with VT/VF suggests the need for reevaluation of a priori risk grouping and raises the question of the appropriate arrhythmia therapy for a broad range of patients.
Randomized trials comparing new and control (or standard) treatments provide the most accurate basis for statistical and clinical inferences about the utility of new therapies. Although randomization helps to ensure the internal validity of a trial, its external validity depends on how well the randomized patients represent the general population with the disease under consideration. That may be difficult to determine. Typically, only a minority of eligible patients enter randomized trials. In this regard, a registry of all patients presenting with the disease at the participating centers may be extremely informative. Not only can a registry address to what extent the enrolled study sample compares with the diseased population, but it also can provide information about the natural history of the disease in both eligible and noneligible patients. A well-designed and -maintained registry thus may provide critical information in evaluating to what extent the study results should be generalized and highlight areas in which additional investigation is needed.
Survivors of ventricular fibrillation (VF) or symptomatic, sustained ventricular tachycardia (VT) represent an important therapeutic target and public health problem1 because of their known high risk of arrhythmia recurrence and sudden death.2 3 Traditional antiarrhythmic drugs have not provided effective prophylaxis against sudden death (even when guided by Holter recordings or electrophysiological studies) and have sometimes been harmful.4 5 6 7 In the past few years, treatment for these patients has consisted mainly of the mixed class II/III agents amiodarone and sotalol or implantable cardioverter-defibrillators (ICDs).8 9 10 11 12 13 14 Randomized trials comparing these therapies are only now being performed and results reported.15 16 17 18 19 20 21 The Antiarrhythmics Versus Implantable Defibrillators (AVID) study was a large, randomized, secondary prevention trial.21 In AVID, ICD therapy was associated with 39%, 27%, and 31% reductions in mortality at 1, 2, and 3 years, respectively, compared with antiarrhythmic drug therapy (mostly empirical amiodarone).
We present here the design and initial observations of the AVID Registry study, performed in parallel with the randomized trial. We confirm preliminary observations that the randomized patients are representative of the general high-risk VT/VF population.22 Finally, we present clinical characteristics and survival for patients thought, at the time of trial design, to be at too low risk for arrhythmia recurrence to be included in the trial.
Screening and Registry enrollment was a 2-step process. A 1-page screening form was completed for all patients seen or referred to an AVID center who had experienced VF or VT, regardless of cause. In addition, patients who experienced unexplained out-of-hospital syncope and were found to have structural heart disease and symptomatic VT induced at electrophysiological study were screened. Most local Investigational Review Boards allowed verbal patient consent for the Registry (because no active patient participation or risk was involved), but a minority required written consent.
Screening classified the index arrhythmia to determine eligibility for the randomized trial (categories A, B, and C, Figure 1⇓). Screening also excluded certain patients from the registry (Figure 2⇓).
Generally, only patients with a reasonable likelihood of being eligible to receive antiarrhythmic drug therapy, ICD therapy, or both were included in the AVID Registry. However, patients who experienced out-of-hospital VT or VF thought to be of transient or correctable cause (eg, in conjunction with myocardial infarction, acute ischemia, drug overdose, or severe electrolyte imbalance) were included to evaluate whether their outcome was better than that of the patients with other arrhythmias, as traditionally believed.2
The second step was to complete the AVID Registry form (Appendix 1) for any patient without Registry exclusion. Demographics, baseline data, and trial eligibility (pages 1 and 2) were obtained during screening (trial exclusion criteria are listed under item 8). Page 3 was completed at hospital discharge and covered procedures done after the index arrhythmia, discharge medications, heart rate, blood pressure, creatine kinase MB isoenzyme measurements obtained within 3 days after the index arrhythmia, the date of the index arrhythmia, and the admitting and discharge dates for the screening hospitalization. These 3 pages were sent to the Data Coordinating Center.
The fourth page was used solely for obtaining follow-up information through the National Death Index Service (NDIS).23 Only page 4 related the Registry identification number to specific patient-identifying information, such as name and social security number. To ensure patient confidentiality, page 4 was kept at the local site and the data were not stored at the coordinating center.
Follow-up of the Registry patients was through the NDIS. Each AVID center entered page 4 data on a diskette; these were sent to the NDIS, which conducted a search based on name, social security number, and birth date for matches against routinely assembled vital statistics (ie, mortality). Using the Registry identification number as a link, the results of this search were incorporated into the Registry database. There were 144 registrants (3.1%) without patient identifiers, either because the site did not participate in the Registry follow-up (n=128) or one of the key identifiers was missing (n=16).
This article reports results based on vital status reports through December 31, 1996, and thus is restricted to the 4219 registry patients with patient identifiers whose index arrhythmic event occurred before January 1, 1997. The NDIS search missed 3 of the 162 deaths (1.85%) that occurred in trial patients.
Survival estimates are based on the methods of Kaplan and Meier.24 Comparisons among survival curves were made by the global log-rank test.25 No differences were noted between analyses of all sites and analyses restricted to centers with relatively large numbers of registrants; thus, analyses presented are of all enrolling centers and all Registry patients.
Of 5989 patients screened, 4595 (77%) were registered, of whom 1016 (22%) were randomized. Of the 4595 registered, 4451 provided identifiers enabling NDIS follow-up; of these, 996 were randomized. Reasons for exclusion from the trial are shown in Figure 3⇓.
Registry Demographics and Representativeness of Trial Patients
The 4595 registrants averaged 64 years of age, and 77% were male. The mean left ventricular ejection fraction was 35%, and 77% had evidence of coronary artery disease (CAD). Of the 3579 nonrandomized patients, 1509 were AVID trial–eligible (trial-eligible arrhythmia and no medical exclusions) and 2070 were noneligible.
AVID Trial–Eligible Patients
Reasons for Nonrandomization
The reasons for not randomizing eligible patients are given in Table 1⇓. In these qualifying patients, patient or family refusal, followed by physician preference, were the most frequent causes for nonrandomization.
Among trial-eligible patients, only a few statistically significant differences between those randomized and those not randomized were found. Randomized patients were slightly older (65.1 versus 63.4 years); more frequently were male (79.4% versus 74.5%); more frequently had CAD (81.5% versus 75.7%) or a past history of myocardial infarction (67.0% versus 58.3%), congestive heart failure (CHF) (46.8% versus 41.8%), syncope (13.1% versus 9.5%), hypertension (55.8% versus 47.9%), or diabetes (24.3% versus 20.6%); and less frequently had no evident structural disease (2.9% versus 5.2%). However, the magnitude of the differences was small (<15% proportionate or <5% absolute differences) and unlikely to have a significant impact on enrollment or treatment decisions.
Cardiac procedures before or after the index arrhythmia were similarly distributed by randomization status in eligible patients. Overall, 28% had previously undergone bypass surgery, and 9% had had a percutaneous coronary intervention without differences between groups. After the arrhythmic event, 7.4% underwent bypass surgery and 3.2% a percutaneous coronary intervention. Nonrandomized patients rarely underwent ablation procedures (2%).
Discharge Cardiac Therapies
Discharge cardiac (nonarrhythmic) therapies were similarly distributed in eligible patients in both groups, with 27% of patients prescribed β-blockers, 14% calcium antagonists, 34% nitrates, 50% diuretics, and 24% warfarin. Two exceptions were small differences in ACE inhibitor use (68.5% versus 57.6%, randomized versus nonrandomized) and aspirin use (59.1% versus 54.0%).
In each group of eligible patients, ICD and antiarrhythmic drug therapies were used with equal frequency at discharge, with each being used in ≈50% of patients (Table 2⇓). By study design, both combination therapy (4%) and neither therapy (1%) were to be avoided in randomized patients, whereas nonrandomized patients could receive both therapies (11%) or neither therapy (11%).
Registry Patients Followed Up by the 1996 NDIS
A total of 4219 patients qualified for and were entered into the Registry by December 31, 1996, and were followed for vital status through December 31, 1996 (via the NDIS). The following is based on these 4219 patients.
Table 3⇓ presents the main clinical characteristics of the Registry patients (including trial patients) grouped according to presenting arrhythmia. Mean ages ranged from 61 to 66 years. The prevalence of CAD ranged from 73% to 82%. The range of mean ejection fractions was 32% to 41%. There was a 2-fold difference in history of CHF (22% to 44%), but a history of diabetes was present in ≈20% of all groups.
Table 4⇓ shows the postevent index arrhythmia procedures and discharge therapies in the patients grouped by presenting arrhythmia. Coronary bypass surgery was performed in 42% of the transient VF/VT group, 17% of the VF group, and ≈10% in other groups. There were substantial differences in rates of arrhythmia ablation procedures (1% to 16%) and use of β-blockers (23% to 44%), digitalis (29% to 43%), ACE inhibitors (40% to 60%), and diuretics (34% to 51%). The rate of ICD implantation varied from 20% to 58% and the use of antiarrhythmic drugs from 30% to 60%.
A relatively high rate of mortality was noted for all 6 arrhythmia groupings, with 2-year survival rates ranging from 76% in patients presenting with syncopal VT to 84% in patients with unexplained syncope and inducible VT (Figure 4⇓) (P=0.007 among groups).
Survival was similar across arrhythmia groups for patients without severe ventricular dysfunction (2-year rates ranged from 82% to 86%, P=NS) but was more variable for patients with poor ejection fraction (<0.35), with 2-year rates ranging from 68% to 83% (P=0.021) (Table 5⇓).
Summary of the AVID Registry Experience
The AVID Registry provides unique, prospectively collected information on a large and broadly representative population of patients surviving life-threatening ventricular tachyarrhythmias. In this article, we present in detail the design of the Registry. We confirm the preliminary observations of Kim et al22 as to the representativeness of the trial patients. Interestingly, the use of antiarrhythmic drug and ICD therapies was almost equal in the nonrandomized patients. We also present baseline and postevent characteristics of the Registry patients subgrouped by presenting arrhythmia, as well as outcome data.
The AVID Registry constitutes the largest published cohort of patients with life-threatening ventricular arrhythmias of which we are aware.26 It consists of patients who were eligible for the AVID main trial and were randomized, those not randomized but qualifying for AVID, and those not randomized and not qualifying. Reasons for and frequency of not randomizing patients were recorded; nonrandomization most often was due to patient or family refusal, followed by physician preference. Interestingly, patient/physician preference (or bias) for drug and device treatments was virtually identical. A strong preference for an ICD over antiarrhythmic drug therapy (or vice versa) was not observed in this Registry population, suggesting the presence of little overall therapeutic selection bias in participating AVID centers.
The issue of generalizability of AVID’s efficacy results to the universe of patients with VT/VF is of prime importance. The knowledge that AVID results were generated from a large percentage of eligible patients (41%) and that nonrandomized but eligible patients were strikingly similar also suggests that results of the study may be applied with more confidence to AVID trial–eligible patients seen in clinical practice. In contrast, many other studies have enrolled a small percentage of patients or have failed to keep a Registry, so that the patient pool’s “denominator” is unknown and uncharacterized.17 Furthermore, several other VT/VF types previously thought to be low-risk were found to be at a risk similar to that of the AVID trial–eligible patients. The surprisingly high and similar mortality rates across the arrhythmia subgroups are being investigated extensively, including ongoing collection of additional baseline and follow-up data. Preliminary multivariate analyses did not explain the similarity of the high rates on the basis of well-known baseline risk predictors, such as ejection fraction (Table 5⇑) or the choice of antiarrhythmic therapy. The factors influencing treatment selection in these nonrandomized groups are unknown, but given the approximately equal use of ICD and antiarrhythmic drug therapy, as in the AVID main trial, this finding raises the possibility that the expense of an ICD may be justified by a potentially large but as yet unproven treatment benefit in additional groups of patients, especially when associated with a low ejection fraction (<0.35).
The lower and similar mortality observed for patients with left ventricular ejection fraction ≥0.35 raises the question, at least from a cost-effectiveness perspective, of the suitability of ICD implantation in patients with relatively preserved left ventricular function.
Transient/Correctable Cause of VT/VF
The high mortality rate for patients with VT/VF thought to be provoked by reversible factors deserves special comment: 42% of patients had in-hospital revascularization, 2% ablation, and 4% pacemakers; many others had adjustment in medical therapy (eg, potassium replacement). However, fewer than half (47%) were discharged on specific antiarrhythmic therapy. Their poor subsequent outcome (29% mortality at 3 years) suggests that clinical judgment regarding reversibility and recurrence risk is not very accurate and that more aggressive and specific antiarrhythmic therapy (including ICD) may be appropriate for many of these patients, especially if they have a low ejection fraction (<0.35).
Registry Caveats and Recommendations
Although the Registry form collected a substantial amount of information, several subsequently proposed analyses would have required that some data items be modified. These included details with regard to primary versus secondary health insurer admission; discharge dates of the first hospitalization related to the index arrhythmia, in addition to the admission and discharge dates for the screening hospitalization; and finally, performance and results of electrophysiological studies (including drug testing).
Although our process for follow-up through the NDIS may appear cumbersome, we found that it worked well. Total mortality (and death certificate codes) but not cause-specific mortality (ie, sudden death versus other causes) is available. Finally, characteristics of patients at sites that enrolled high and low numbers of patients were similar.
In conclusion, the AVID Registry design formed an important part of the AVID study program and has provided important information about the natural history of VT/VF, current practice patterns across a broad base of institutions, and a firm basis for generalizing the AVID study results to the general population of patients with AVID-type ventricular arrhythmias.
Four-Page Registry Case Report Form
The AVID Registry case report form is shown in Figure 5⇓.
University of Alabama/Birmingham: A.E. Epstein, R.A.S. Cooper, S.M. Dailey, G.N. Kay, V.J. Plumb, R.S. Bubien, S.M. Knotts, P.T. McKenna. Arkansas Cardiology Clinic, Little Rock: G.S. Greer, I.H. Santoro, J. Swaim, S. Whittle. Baylor College of Medicine, Houston, Tex: C. Rizo-Patron, K.M. Belco, P. Payne, D.J. Arnold, W.X. Zhu, C.M. Pratt. Baystate Medical Center, Springfield, Mass: J.R. Cook, G. Kabell, J.B. Kirchhoffer, D. Warwick, B. Burkott, D. Tomaszewski. Boston University Medical Center Hospital, Boston, Mass: P.J. Podrid, T.T. Fuchs, M. Mazur. Brigham and Women’s Hospital, Boston, Mass: P.L. Friedman, W.G. Stevenson, M.M. Swat, L.I. Ganz, M.O. Sweeney, J.B. Shea. University of California/Irvine Medical Center: M.A. Brodsky, B.J. Allen, S.S. Ehrlich, L.J. Wolff, M.M. Macari-Hinson. University of California/San Francisco: M. Scheinman, S. Eisenberg, L. Epstein, A. Fitzpatrick, G. Griffin, R. Lee, M. Lesh, M. Namekawa-Wong, M. Wong. Cleveland Clinic Foundation, Cleveland, Ohio: B.L Wilkoff, S.L. Pinski, J.A. Cross, J.M. Shewchik. University Hospitals of Cleveland, Cleveland, Ohio: M.D. Carlson, D.S. Rosenbaum, W.R. Lewis, L.A. Biblo, J.A. Mackall, A.L. Waldo. Columbia-Presbyterian Medical Center, New York, NY: J. Coromilas, J.T. Bigger, Jr, F.D. Livelli, Jr, J.A. Reiffel, K. Hickey. University of Connecticut Health Center, Framington: E. Berns, N. Lippman, M.B. Barry, H.M. Carney. Cooper Hospital/University Medical Center, Camden, NJ: A.M. Russo, H.L. Waxman, C.A. Stubin, J.M. Morrissey, D.H. Raspa. Deaconess/Sacred Heart Medical Centers, Spokane, Wash: D.A. Chilson, T.J. Lessmeier, W.T. Pochis, J.M. Baxter. Florida Arrhythmia Consultants, Ft. Lauderdale: R.M. Luceri, P. Zilo, D.N. Weiss, A. Jonas, L. Vardeman. Florida Heart Group, Orlando: K.M. Schwartz, C. Asbell. University of Florida, Gainesville: A.B. Curtis, J.B. Conti, C. Nelson. Good Samaritan Hospital, Los Angeles, Calif: D.S. Cannom, A.K. Bhandari, R.D. Lerman, B.R. Firth. Indiana University Medical Center, Indianapolis: D.P. Zipes, E.D. Engelstein, W.M. Miles, L. Foreman. University of Iowa Hospitals, Iowa City: J.B. Martins, H.C. Lee, J.R. Hopson, R. Hopson. Lankenau Hospital/Medical Research Center, Wynnewood, Pa: R.A. Marinchak, S.J. Rials, P.R. Kowey, R.A. Filart, M. Hernandez, D.L. Scher, L.S. Zukerman, S. Farrell, D. Kolk, H. Criner, B. Tait. Likoff Cardiovascular Institute/Hahnemann University, Philadelphia, Pa: S.P. Kutalek, S.E. Hessen, S. Wilbur, F. Samuels, C. Baessler, D. Soto, M. Schusteri. Loma Linda University Medical Center, Loma Linda, Calif: V.I. Torres, S. Pai, M.L. Platt, P.C.R. Timothy, V.L. Bishop. Maine Medical Center, Portland: J.C. Love, J.E. Cutler, S. Bosworth-Farrell. University of Maryland Medical Center, Baltimore: M.R. Gold, S. Shorofsky, R. Peters, N. Kavesh, D. Froman, H. Scott. Medical College of Wisconsin, Milwaukee: J.A. Roth, A.M. Gadhoke, S.K. Mauermann. Methodist Hospital/Houston, Tex: A. Pacifico, N. Nasir, Jr, K. Trainor. Methodist Hospital of Indiana, Indianapolis: P. Foster, B. Crevey, H. Genovely, J. Schutzman, K. Viater, C. Coyle. Midwest Heart Research Foundation, Lombard, Ill: M.O. Nora, M.F. O’Toole, R. Collura, E.L. Enger. Minneapolis Heart Institute, Minneapolis, Minn: A. Almquist, P. Baldwin, R. Hauser, S. Milstein, M. Pritzker, V. Schreiber. Montefiore Medical Center/Bronx, NY: S.G. Kim, J.D. Fisher, K.J. Ferrick, J.N. Gross, U. Ben-Zur, J. Durkin, A. Ferrick. Mount Sinai Medical Center, Miami Beach, Fla: J. Zebede, P. Samet, A.O. Tolentino, S. Rubin. Sentara Norfolk General Hospital, Norfolk, Va: J.M. Herre, R.C. Bernstein, L.R. Klevan. North Shore University Hospital, Manhasset, NY: R.L. Jadonath, B. Goldner, K. Merkatz, L. Chepurko. University of Oklahoma Health Science Center, Oklahoma City: K.J. Beckman, J.H. McClelland, M.D. Gonzalez, L.E. Widman, R. Lane. Oregon Health Science University/Collaborating Medical Centers, Portland: J.H. McAnulty, B.D. Halperin, J. Kron, G.C. Larsen, M.H. Raitt, R.D. Swenson, R.C. Florek, C. Marchant, M. Hamlin, G. Heywood. University of Rochester/Collaborating Hospitals, Rochester, NY: T. Akiyama, J.P. Daubert, C. Kim, D. Switzer, P. Pande, D. Flynn, M. Keller, C.M. Ocampo, K. Wahl, J. Vogt. St. Joseph’s Hospital/Medical Center, Patterson, NJ: N.G. Tullo, C.A. Irmiere, A. Henry. St Luke’s–Roosevelt Hospital Center/Collaborating Institutions, New York, NY: J.S. Steinberg, F. Ehlert, S. Zelenkofske, E. Menchavez-Tan, M. DeStefano, G. Brown. Stanford University Medical Center/Collaborating Medical Center, Stanford, Calif: R.J. Sung, L.B. Liem, C. Young, M.R. Lauer, J. Peterson, L. Ottoboni, P. Goold. Tacoma General Hospital, Tacoma, Wash: M. Rome, K. Kresge. Temple University School of Medicine, Philadelphia, Pa: A.E. Buxton, H.H. Hsia, J.M. Miller, S.A. Rothman, N. Adelizzi, M.E. Gastineau, L.M. Thome, D.M. Whitley. University of Texas Medical School–Houston/Baylor College of Medicine: H.T. Shih, A.H. Dougherty, S. Jalal, G.V. Naccarrelli, D. Wolbrette, D. Wilson. The University of Texas Southwestern Medical Center, Dallas, Tex: R.L. Page, M. Jessen, R.C. Canby, G. Erwin. University of Utah Medical School/LDS Hospital, Salt Lake City: J.L. Anderson, B.G. Crandall, L.A. Karagounis, J.S. Osborn, D.A. Rawling, K. Summers, M. Biggins. University of Utah/VA Medical Center, Salt Lake City: R.C. Klein, R. Freedman, M. Marks. Valley Heart Associates, Modesto, Calif: W. Chien, R. Stevenot, J. Merillat, M. McPherson. Vanderbilt University Medical Center, Nashville, Tenn: D.S. Echt, D.M. Roden, M.S. Wathen, J.T. Lee, K.T. Murray, J.H. Baker, D.M. Crawford. University of Virginia Medical Center, Charlottesville: S. Nath, D. Haines, S. Ackerman, L. Bowman. Wilson Regional Medical Center, Johnson City, NY: N.J. Stamato, D. Whiting. Clinical Trial Center, University of Washington, Seattle: A.P. Hallstrom, H.L. Greene. Clinical Trials Branch/National Heart, Lung, and Blood Institute, Bethesda, Md: E.B. Schron, M.J. Domanski, D. Follmann, Y. Rosenberg, C.A. Jennings.
This study was supported in part by a contract (N01-11C-25117) with the National Heart, Lung, and Blood Institute, Bethesda, Md.
1 AVID Investigators are listed in Appendix 2.
- Received July 29, 1998.
- Revision received December 8, 1998.
- Accepted December 29, 1998.
- Copyright © 1999 by American Heart Association
Nichol G, Hallstrom AP, Ornato JP, Riegel B, Stiell IG, Valenzuela T, Wells GA, White RD, Weisfeldt ML. Potential cost-effectiveness of public access defibrillation in the United States. Circulation. 1998;97:1315–1320.
Cobb LA, Baum RS, Alvarez H III, Schaffer WA. Resuscitation from out-of-hospital ventricular fibrillation: 4 years follow-up. Circulation. 1975;52:(suppl III):III-223–III-235.
Goldstein S, Landis JR, Leighton R, Ritter G, Vasu CM, Wolfe RA, Acheson A, Vanderburg-Medendorp S. Predictive survival models for resuscitated victims of out-of-hospital arrest with coronary heart disease. Circulation. 1985;71:873–880.
Julian DG, Jackson FL, Prescott RJ, Szekely P. Controlled trial of sotalol for one year after myocardial infarction. Lancet. 1982;i:1142–1150.
Maloney J, Masterson M, Khoury D, Trohman R, Wilkoff B, Simmons T, Morant V, Castle L. Clinical performance of the implantable cardioverter defibrillator: electrocardiographic documentation of 191 spontaneous discharges. Pacing Clin Electrophysiol. 1991;14:230–235.
Zipes DP, Roberts D. Results of the international study of the implantable pacemaker cardioverter-defibrillator: a comparison of epicardial and endocardial lead systems. Circulation. 1995;92:59–65.
Wever EF, Hauer RN, Schrijvers G, van Capelle FJ, Tijssen JG, Crijns HJ, Algra A, Ramanna H, Bakker PF, Robles de Medina EO. Tiered strategy in postinfarct sudden death survivors: a randomized study. Circulation. 1996;93:489–496.
Bigger JT Jr, for the CABG Patch Investigators. Prophylactic use of implanted cardiac defibrillators in patients at high risk for ventricular arrhythmias after coronary artery bypass graft surgery. N Engl J Med. 1997;337:1468–1475.
Kim SG, Hallstrom A, Love JC, Rosenberg Y, Powell J, Roth J, Brodsky M, Moore R, Wilkoff B, for the AVID Investigators. Comparison of clinical characteristics and frequency of implantable defibrillator use between randomized patients in the Antiarrhythmics Versus Implantable Defibrillators (AVID) trial and nonrandomized registry patients. Am J Cardiol. 1997;80:454–457.
Boyle CA, Decouflé T. National sources of vital status information: extent of coverage and possible selectivity in reporting. Am J Epidemiol. 1990;131:160–168.
Willems AR, Tijssen JG, vanCapelle FJ, Kingma JH, Hauer RN, Vermeulen FE, Brugada P, van Hoogenhuyze DC, Janse MJ, the Dutch Ventricular Tachycardia Study Group of the Inter-university Cardiology Institute of the Netherlands. Determinants of prognosis in symptomatic ventricular tachycardia or ventricular fibrillation late after myocardial infarction. J Am Coll Cardiol. 1990;16:521–530.