The Implantable Cardioverter-Deﬁbrillator Does Everybody Need One? Satish R. Raj and Robert S. Sheldon
The implantable cardioverter-deﬁbrillator (ICD) has emerged as an effective, but expensive, therapy for arrhythmic sudden cardiac death. ICD use has been increasing by 20% to 30% per year. Clinical trials have shown that the ICD can be effective for both the primary prevention and the secondary prevention of sudden cardiac death in selected populations. Despite the available trial evidence, several issues pertaining to ICD use remain unresolved, including the treatment of patients not represented in clinical trials, the optimal selection of patients who will beneﬁt from an ICD, the duration of beneﬁt from an ICD, the quality of life for patients with an ICD, and both the cost-effectiveness and the cost impact of the ICD. These considerations are discussed in this article. Copyright © 2001 by W.B. Saunders Company
udden cardiac death (SCD) accounts for over 300,000 deaths annually in the United States.1 Successful resuscitation is uncommon, with only 6% of victims discharged alive from the hospital.2,3 Those that survived their ﬁrst episode of SCD were once treated with antiarrhythmic drug therapy guided by arrhythmia suppression during serial electrophysiologic testing or serial ambulatory electrocardiographic (Holter) monitoring.4,5 However, this approach has been shown to be unsuccessful at accurately predicting reduction in mortality by drug therapy.6,7 The current approach to the management of such patients consists of using an implantable cardioverterdeﬁbrillator (ICD),8-10 empiric amiodarone,11 or electrophysiologic testing– guided therapy with sotalol in select populations.12 Michel Mirowski ﬁrst conceived of the ICD after the death of his mentor, Harry Heller, who succumbed following a spell of recurrent ventricular tachyarrhythmias. He envisioned an implantable device that could both monitor the cardiac
rhythm and deliver an appropriate deﬁbrillation shock.13 Early animal studies led to successful human results by 1980.14-16 The United States Food and Drug Administration approved commercial ICDs in 1985. Since then, the ICD has undergone a remarkable evolution. The devices have evolved from large generators and cans that required abdominal implantation to smaller, thinner, lighter generators implanted in the pectoral region; from requiring a thoracotomy to sew deﬁbrillation patches on the epicardium to nonthoracotomy, transvenous, endocardial deﬁbrillation coils; and from devices that could do little more than deliver a deﬁbrillation shock to highly programmable devices with antitachycardia pacing and bradycardia pacing, and sophisticated telemetric and diagnostic algorthims. ICD use has grown 20% to 30% annually through the 1990s. Nearly 100,000 worldwide implants are projected in 2001. This growth has been because of a combination of factors, including improvements in device technology, evidence from randomized clinical trials, and the intuitive appeal of an electrical life insurance policy. This article ﬁrst reviews clinical trial evidence for the prevention of sudden cardiac death with an
From the Cardiovascular Research Group, University of Calgary, Calgary, Alberta, Canada. Supported in part by grant GR13914 from the Canadian Institutes of Health Research, Ottawa, Canada. Dr Raj is a Research Fellow of the Heart and Stroke Foundation of Canada and the Canadian Institutes of Health Research. Address reprint requests to Satish Raj, MD, FRCP (C), Faculty of Medicine, University of Calgary, Health Sciences Centre, 3330 Hospital Dr NW, Calgary, Alberta T2N 4N1, Canada. Copyright © 2001 by W.B. Saunders Company 0033-0620/01/4403-0003$35.00/0 doi:10.1053/pcad.2001.29146
Progress in Cardiovascular Diseases, Vol. 44, No. 3, (November/December) 2001: pp 169-194
RAJ AND SHELDON
ICD and then raises some issues regarding the appropriate use of the ICD that are not yet completely answered.
Guidelines for Implantation of ICDs In 1998, the American Heart Association (AHA) and the American College of Cardiology (ACC) published joint guidelines for the uses of the ICD.17 These are summarized in Table 1. In the AHA/ACC guideline format, class I recommendations are those for which there is evidence and/or general agreement that a given treatment is beneﬁcial, class II recommendations are for those for which there is conﬂicting evidence and/or a divergence of opinion about the usefulness/efﬁcacy of a
treatment (broken into class IIa and class IIb recommendations based on the strength of evidence or opinion), and class III recommendations are contraindications due to inefﬁcacy or harm.17 What is the evidence for these recommendations, and what impact will they have on patients and the health care system?
Clinical Trials of ICDs The clinical trials of ICDs for the prevention of SCD are grouped according to whether the target population are already survivors of ventricular tachyarrhythmias (secondary prevention) or are at high risk of SCD without a previous
Table 1. AHA/ACC 1998 Guidelines for the Clinical Indications for Implantation of the ICD Class I
Indications for ICD Therapy Cardiac arrest caused by VF or VT, not caused by a transient or reversible cause Spontaneous sustained VT Syncope of undetermined origin with clinically relevant, hemodynamically signiﬁcant sustained VT or VF induced at EP study when drug therapy is ineffective, not tolerated, or not preferred NSVT with coronary artery disease, prior MI, LV dysfunction, and inducible VF or sustained VT at EP study that is not suppressible by a class I antiarrhythmic drug None Cardiac arrest presumed to be caused by VF when EP testing is precluded by other medical conditions Severe symptoms attributable to sustained ventricular tachyarrhythmias while awaiting cardiac transplantation Familial or inherited conditions with a high risk for life-threatening ventricular tachyarrhythmias such as long QT syndrome or hypertrophic cardiomyopathy NSVT with coronary artery disease, prior MI, and LV dysfunction, and inducible sustained VT or VF at EP study Recurrent syncope of undetermined cause in the presence of ventricular dysfunction and inducible ventricular arrhythmias at EP study when other causes of syncope have been excluded Syncope of undetermined cause in a patient without inducible ventricular tachyarrhythmias Incessant VT or VF VF or VT resulting from arrhythmias amenable to surgical or catheter ablation; eg, atrial arrhythmias associated with the Wolff-Parkinson-White syndrome, right ventricular outﬂow tract VT, idiopathic LV tachycardia, or fascicular VT Ventricular tachyarrhythmias caused by a transient or reversible disorder (eg, AMI, electrolyte imbalance, drugs, trauma) Signiﬁcant psychiatric illnesses that may be aggravated by device implantation or may preclude systematic follow-up Terminal illnesses with projected life expectancy ⱕ6 months Patients with coronary artery disease with LV dysfunction and prolonged QRS duration in the absence of spontaneous or inducible sustained or nonsustained VT who are undergoing coronary bypass surgery NYHA class IV drug-refractory congestive heart failure in patients who are not candidates for cardiac transplantation
NOTE. Class I—Conditions for which there is evidence and/or general agreement that a procedure or treatment is beneﬁcial, useful, and effective. Class II—Conditions for which there is conﬂicting evidence and/or a divergence of opinion about the usefulness/efﬁcacy of a procedure or treatment (IIa—weight of evidence/opinion is in favor of usefulness/ efﬁcacy; IIb— usefulness/efﬁcacy is less well established by evidence/opinion). Class III—Conditions for which there is evidence and/or general agreement that a procedure/treatment is not useful/effective and in some cases may be harmful. Abbreviations: VF, ventricular ﬁbrillation; VT, ventricular tachycardia; EP, electrophysiological; NSVT, nonsustained ventricular tachycardia; MI, myocardial infarction; LV, left ventricular; AMI, acute myocardial infarction; NYHA, New York Heart Association. Data from reference 17.
Table 2. Completed and Published Trials of the ICD Trial Secondary prevention The Dutch Study
Antiarrhythmics Versus Implantable Deﬁbrillator (AVID) Trial
Control Group Therapy Conventional therapy including guided IA, IC, and III AAD therapy; RF ablation; arrhythmia surgery or a late ICD Amiodarone or sotalol
Canadian Implantable Deﬁbrillator Study (CIDS)
Cardiac Arrest Study— Hamburg (CASH)
Propafenone (terminated early), metoprolol, or amiodarone
Primary prevention Multicenter Automatic Implantable Deﬁbrillator (MADIT) Trial Coronary Artery Bypass Graft-Implantable CardioverterDeﬁbrillator Study (CABG-Patch) *Multicenter Unsustained Tachycardia Trial (MUSTT)
Conventional AAD therapy
Survivor of cardiac arrest caused by VF or VT, at least 4 weeks post-MI who have inducible ventricular arrhythmias at EP study
Signiﬁcant reduction in total mortality with ICD compared with conventional therapy
Survivor of cardiac arrest or syncopal VT or symptomatic sustained VT with LVEF ⱕ40% Survivor of cardiac arrest or syncopal VT or symptomatic sustained VT with LVEF ⬍35% or syncope with inducible VT at EP study Survivor of cardiac arrest
Signiﬁcant reduction in total mortality with ICD compared with AAD therapy
Prior MI and LVEF ⱕ35% and NSVT and inducible VT at EP study, which is not suppressible with procainamide Coronary artery bypass surgery and LVEF ⬍36% and a positive signal averaged ECG
Signiﬁcant reduction in total mortality with ICD versus conventional therapy
Coronary artery disease and LVEF ⱕ40% and NSVT and inducible VT at EP study; ICD used only if 1 or more AAD failed to suppress inducible VT
Signiﬁcant reduction in primary endpoint (cardiac arrest or death) with EP guided therapy; entire beneﬁt restricted to subgroup that received ICD
Nonsigniﬁcant trend toward reduction in total mortality with ICD compared with amiodarone
Nonsigniﬁcant trend toward reduction in total mortality with ICD compared with metoprolol/ amiodarone
No difference in total mortality
*MUSTT is actually a randomized trial of EP study guided therapy versus no antiarrhythmic drug therapy, with the ICD used as 1 option in the EP study guided therapy group. Abbreviations: AAD, antiarrhythmic drug; RF, radiofrequency; VF, ventricular ﬁbrillation; VT, ventricular tachycardia; MI, myocardial infarction; EP, electrophysiologic; LVEF, left ventricular ejection fraction; NSVT, nonsustained ventricular tachycardia; ECG, electrocardiogram.
ventricular tachyarrhythmia (primary prevention).
Secondary Prevention of SCD Four randomized controlled trials have compared the efﬁcacy of ICDs versus antiarrhythmic drug therapy: the Dutch Study,18 the Antiarrhythmic Versus Deﬁbrillator (AVID) Trial,8 the Canadian Implantable Deﬁbrillator Study (CIDS),9 and the Cardiac Arrest Study Hamburg (CASH).10 These trials are summarized in Table 2.
The Dutch Study The subjects were patients with prior cardiac arrest caused by ventricular ﬁbrillation or ventricular tachycardia in the setting of a remote myocardial infarction.18 If ventricular arrhythmias were induced at an electrophysiologic study, the patients were randomized to receive either an early ICD or conventional therapy with antiarrhythmic drugs selected according to their ability to suppress ventricular dysrhythmias. This suppressive ability could be assessed
172 by using Holter monitoring, exercise testing, or electrophysiologic studies. Among serial drug nonresponders, radiofrequency catheter ablation or arrhythmia surgery was performed if a stable ventricular tachycardia was induced. An ICD was implanted in patients who had unstable arrhythmias or were still inducible and nonsuppressible after ablation or surgery. The main outcomes were total mortality, cardiac arrest, and heart transplant. Analysis was performed by intention-to-treat. Total study enrollment was only 60 patients, with 31 patients randomly allocated to conventional therapy and the remaining 29 patients to an early ICD. The groups were evenly matched for age (57 ⫾ 10 and 59 ⫾ 11 years), left ventricular ejection fraction (29 ⫾ 12% and 30 ⫾ 11%), prior myocardial infarction, and functional status. There was a 63% relative risk reduction in the total mortality rate in the ICD group compared with the conventional group (14% v 35% deaths). Of note, 20 patients (61%) assigned to the conventional therapy group failed serial drug testing, and a late ICD was implanted in 15 of those patients. However, this study has several important limitations. The study sample was small, and all-cause mortality was not the primary endpoint. The comparison with early ICD was not placebo but class I or class III antiarrhythmic medications. These drugs may increase mortality. Thus, it may be that the apparent beneﬁt of ICD seen here was actually caused by increased mortality because of treatment with antiarrhythmic drugs. Finally, the rate of revascularization was higher in the ICD group (26% v 10%), which may have reduced its mortality rate.
Antiarrhythmics Versus Implantable Deﬁbrillator (AVID) Trial AVID8,19 was a randomized clinical trial that assessed the relative beneﬁt of antiarrhythmic drugs versus ICDs on survival of patients with a history of life-threatening ventricular arrhythmias. Patients were recruited into AVID if they were resuscitated from ventricular ﬁbrillation, had ventricular tachycardia with syncope, or had hemodynamically signiﬁcant sustained ventricular tachycardia with a left ventricular ejection fraction ⱕ40%. Hemodynamic signiﬁcance was deﬁned by the presence of near-syncope, hypoten-
RAJ AND SHELDON
sion, angina, or heart failure. Patients were excluded from the trial if the ventricular tachyarrhythmia was believed to be caused by a reversible cause such as acute myocardial infarction, electrolyte imbalance, or drug toxicity; if there were contraindications to either therapy; or if there was a high likelihood of death from comorbidities. The primary endpoint was total mortality, with quality of life and cost-effectiveness as prespeciﬁed secondary endpoints. The analysis was performed by intention-to-treat. A sample size of 1,200 patients was chosen to detect a 30% ICD-related decrease in an expected 40% 4-year mortality in the drugtreated group (2-sided analysis; alpha ⫽ 0.05). However, the study was terminated prematurely when an interim analysis showed that the probability of beneﬁt from the ICD had crossed prespeciﬁed bounds for study termination. Of the 1,885 patients eligible for randomization, 1,016 patients were randomized to therapy with an ICD (n ⫽ 507) or antiarrhythmic drug therapy (n ⫽ 509). The subjects were 80% male with a mean age of 65 ⫾ 10 years. The 2 study groups had similar mean left ventricular ejection fractions (32% v 31%), and comparable proportions of patients underwent coronary artery revascularization (10% v 12%). The groups had a similar distribution of presenting arrhythmias. However, there was a slightly higher incidence of New York Heart Association (NYHA) class III heart failure in the drug group (12% v 7%). Of the 509 patients randomized to drug therapy, 496 received empiric amiodarone therapy and 13 received sotalol. At 2 years, 85% of the patients assigned to amiodarone were still taking the drug. In the ICD group, 93% received a nonthoractomy system, and only 2% did not have an ICD implanted. Over a mean follow-up of 18.2 ⫾ 12.2 months, the crude death rates (⫾ 2 SD) were 15.8 ⫾ 3.2% (n ⫽ 80) in the ICD group and 24 ⫾ 3.7% (n ⫽ 122) in the antiarrhythmic drug group. The ICD group had a greater survival rate than the antiarrhythmic drug group at 1 year (89.3% versus 82.3% survival; relative risk reduction [RRR] [⫾ 2 SD], 39 ⫾ 20%), at 2 years (81.6% v 74.7%; RRR, 27% ⫾ 21%), and at 3 years (75.4% v 64.1%; RRR, 31% ⫾ 21%). Therefore, treatment with an ICD extended life by a mean of 2.7 months in the ﬁrst 3 years. This beneﬁcial effect was seen in all prespeciﬁed subgroups (Fig 1). The point estimate of
Fig 1. A hazard plot showing the point estimates and 95% conﬁdence intervals for the overall group and prespeciﬁed subgroups from the AVID trial. A hazard ratio of less than 1.0 represents a survival advantage from the implantable cardioverter-deﬁbrillator. None of the subgroup point estimates was signiﬁcantly different from the overall point estimate. (Abbreviations: LVEF, left ventricular ejection fraction; CAD, coronary artery disease). (Reprinted with permission from the Antiarrhythmics Versus Implantable Deﬁbrillators (AVID) Investigators: A comparison of antiarrhythmicdrug therapy with implantable deﬁbrillators in patients resuscitated from near-fatal ventricular arrhythmias. N Engl J Med 337:1576-1583, 1997. Copyright © 1997 Massachusetts Medical Society. All rights reserved.)
the group with a relatively preserved ejection fraction (⬎35%) showed a less impressive beneﬁt of the ICD in this subgroup, but the wide conﬁdence intervals also suggest that either ICD or drug therapy might be superior in this subgroup. The conclusions of the study might be confounded by a signiﬁcant discrepancy in the use of ␤-blockers. Patients in the ICD group were almost 3 times as likely to be on ␤-blockers as were the patients in the antiarrhythmic drug group (42.3% v 16.5%). This might be important, given that ␤-blockers improve the survival of patients with myocardial infarction and arrhythmias.20,21 Metaanalysis of the amiodarone primary prevention
trials showed that much of the beneﬁt of amiodarone in decreasing arrhythmic death was restricted to patients on concurrent nonrandomly assigned ␤-blockers.22 However, the AVID investigators performed a Cox regression analysis and found that adjustment for ␤-blocker use reduced the estimated survival beneﬁt in the ICD group compared with the drug group from 0.62 to an adjusted hazard ratio of 0.67.19 Therefore, even after adjustment for ␤-blocker use, the beneﬁt to the ICD group was signiﬁcant. The AVID Trial did not have a placebo group, but instead the ICD group was compared with an antiarrhythmic drug group, effectively amiodarone. This was because it was believed to be unethical to not treat a high-risk group of ventricular arrhythmia survivors.19 Although it is possible that the apparent ICD beneﬁt was actually caused by a detrimental effect of amiodarone, this is very unlikely in light of the consistently neutral to positive effect of amiodarone on total mortality seen in other randomized trials.11 Although the results of AVID are statistically signiﬁcant, the unadjusted prolongation of life by ICD therapy was only 2.7 months over the ﬁrst 3 years. Although this absolute effect is far from inconsequential, it is not a striking effect.23 Further, this beneﬁt came at a cost of greater rehospitalization in the ICD group (74.8%) than in the drug group (64.7%) at 2 years.
Canadian Implantable Deﬁbrillator Study (CIDS) CIDS9 enrolled patients believed to be at high risk of sudden deaths. Patients were eligible for randomization to either amiodarone or an ICD with the following criterion: if they had documented ventricular ﬁbrillation, had out an out-of-hospital cardiac arrest requiring electrical shock, sustained ventricular tachycardia causing syncope, sustained ventricular tachycardia ⱖ150 beats per minute in patients with left ventricular ejection fractions ⱕ35% associated with symptoms of hemodynamic compromise, or had unmonitored syncope with later documentation of spontaneous ventricular tachycardia or sustained ventricular tachycardia induced during an electrophysiologic study. The primary endpoint was total mortality, and the sample size was 650 patients. This provided adequate power to detect an expected 33%
174 relative risk reduction with the ICD from the expected 30% 3-year mortality with amiodarone. The initial intent of the authors was to perform 1-sided comparisons, seeking only a beneﬁt of the ICD over amiodarone and without concerns about any deleterious effects of the ICD. However, the ﬁnal publication reports 2-sided comparisons.9 All patients received follow-up for at least 1 year, with a cumulative follow-up of 48 months. A total of 659 patients were randomized, with 328 randomized to the ICD group. Most of these patients received the current standard of nonthoracotomy devices, with a thoracotomy performed in 33, and no ICD implantation in 18 patients. At 5 years, 85% of the patients assigned to amiodarone were still receiving this medication. Crossover was signiﬁcant with 28.1% of the patients in the ICD group also receiving amiodarone, and 21.4% of the amiodarone group received an ICD. There was a nonsigniﬁcant trend toward a reduction in total mortality in the ICD group (8.3% per year) compared with the amiodarone group (10.2% per year; RRR [95% conﬁdence interval (CI)], 19.7% [⫺7.7% to 40%]; P ⫽ .14). This was almost entirely caused by a nonsigniﬁcant reduction in the risk of arrhythmic death with the ICD compared with amiodarone (3.0% per year v 4.5% per year; RRR, 32.8% [⫺7.2% to 57.8%]; P ⫽ .09). The cumulative total mortality risk in the ICD group and amiodarone group at 1 year was 9.46% versus 11.18% (RRR, 15.4%), at 2 years it was 14.75% versus 20.97% (RRR, 29.7%); and at 3 years it was 23.32% versus 27.03% (RRR, 13.7%). Note that the annualized mortality reduction from the ICD is not consistent over time. Although there was an almost 30% risk reduction at 2 years, the risk reduction was only 15% at 3 years. There were over 150 patients remaining in each study arm at 3 years, so this ﬁnding is not likely merely because of small numbers. Rather, it may represent a true diminution of the beneﬁt of the ICD over antiarrhythmic drug therapy after 2 years after the index arrhythmia. It is worth recalling that AVID followed-up on only 104 patients in both groups combined. The CIDS data might be more accurate at this longer time horizon. The reasons for the lack of statistical signiﬁcance in the CIDS trial as compared with the AVID trial are not entirely clear. The CIDS trial was likely underpowered. It was originally designed for a 1-sided analysis but was published with a 2-sided analysis.
RAJ AND SHELDON
The AVID trial was designed with the same statistical assumptions as were used in the ﬁnal publication and thus was more likely to be adequately powered. There was a lower 2-year mortality in the antiarrhythmic drug group in CIDS (20%) compared with AVID (25%), as well as a lower relative risk reduction in mortality with an ICD (20% v 27%). Many observers, including the CIDS authors,9 have concluded that these results simply represent a slightly underpowered but concordant result with those of the AVID study, as opposed to disputing the superiority of ICDs.
Cardiac Arrest Study Hamburg (CASH) The CASH10 trial was an open-label comparison of antiarrhythmic drug therapy versus ICD for survivors of cardiac arrest caused by a sustained ventricular tachyarrhythmia. This was a more restrictive enrollment criterion than that used in either AVID or CIDS. The CASH trial began randomizing patients in 1987 equally into 1 of 4 treatment arms: metoprolol, propafenone, amiodarone, or ICD. The primary study endpoint was total mortality. Analysis was performed on the intentionto-treat principle. Sample size was calculated based on a 1-sided alpha of 0.05, based on the presumption that the ICD would be better. The authors did not account for interim analyses or multiple looks at the data in planning their analysis. They anticipated that the ICD would show a 38% RRR in the expected 50% mortality in the antiarrhythmic drug arms. The propafenone arm of this study was prematurely terminated in March 1992 after an interim analysis showed that over an 11.3-month follow-up period, this group had a signiﬁcantly higher mortality rate (29.3%) compared with the ICD group (11.5%, P ⫽ .01). Enrollment continued in the other 3 arms until March 1996, and follow-up was continued for another 2 years, until March 1998. A total of 288 patients were recruited to the remaining 3 arms of the study: 99 to the ICD group, 92 to the amiodarone group, and 97 to the metoprolol group. For ﬁnal analysis, the amiodarone and metoprolol groups were combined into 1 antiarrhythmic drug group and compared with the ICD group.10 Because a pure ␤-blocker was part of the randomization, they were not used at all in either the amiodarone or the ICD group. The groups were well
matched on the whole. The mean age of the entire population was 58 ⫾ 11 years, 80% were men, and 73% had coronary artery disease. The mean ejection fraction was 46% in each group, and most patients were in NYHA class II heart failure. Coronary artery revascularization was performed in 19% to 21% of the patients. Over a mean follow-up of 57 ⫾ 34 months, the crude death rate in the antiarrhythmic drug group (44.4% [95% CI, 37.2% to 51.8%]) was nonsigniﬁcantly greater than in the ICD group (36.4% [95% CI, 26.9% to 46.6%]; hazard ratio, 0.766; 1P ⫽ .08). Over the ﬁrst 9 years of study followup, the relative reductions in mortality rates were 41.9%, 39.3%, 28.4%, 27.7%, 22.8%, 11.4%, 9.1%, 10.6%, and 24.7% for each of those years. The crude death rates were not apparently different between the amiodarone and the metoprolol groups. These data would seem to indicate that the incremental beneﬁt of the ICD over antiarrhythmic therapy (in this case, amiodarone or metoprolol) diminishes beyond the ﬁrst 3 years after the index event. Follow-up included 100 patients total in year 6, suggesting that this time dependence to the incremental beneﬁt of the ICD is not merely a spurious ﬁnding because of small patient numbers late in follow-up. This study has several weaknesses. First, it was not blinded, and potential bias in patient treatment might have thus affected the outcome.24 The statistical approach might have been methodologically stronger if an analysis of variance of the effects among the 3 groups was performed. Unless the authors were conﬁdent that the implantable deﬁbrillator could not increase mortality, a 2-sided statistical approach would have been more appropriate. These more rigorous statistical approaches might have increased the neutrality of the CASH results. It is possible that the more neutral result was related to the higher ejection fraction in CASH (46%) than in AVID (32%). The beneﬁcial effects of the ICD over antiarrhythmic drug therapy extended only weakly to patients with an ejection fraction greater than 35% in AVID,25 and this group may have been disproportionately represented in the CASH patient population.
Meta-Analysis of the ICD in Secondary Prevention Trials Before the completion of their respective studies, the steering committees of AVID, CIDS, and
CASH trials had agreed to combine their results in the form of a meta-analysis.26 The purpose of the meta-analysis was to pool the data to allow for the most accurate available point estimate of the relative beneﬁt of ICD over antiarrhythmic drug therapy in survivors of ventricular tachyarrhythmias. Although the AVID trial had many more patients than either the CIDS or CASH trials, it was stopped early.8 Thus, the CIDS and CASH trials may provide more accurate information regarding beneﬁts over the longer term. In fact, the total patient-years of follow-up were greater in the CIDS trial than in the other studies.26 The metoprolol and propafenone groups from the CASH trial were excluded from the meta-analysis to allow for a purer comparison between amiodarone and the ICD. Because of the similarities in the patient population and trial design, the authors were able to pool individual patient data for analysis and perform a more traditional ﬁxed effects meta-analysis. The ﬁxed effects approach showed that the ICD beneﬁt was consistent through the 3 studies (P heterogeneity ⫽ 0.31).26 This supports the notion that the CIDS ﬁndings were consistent with the AVID results, despite the nonsigniﬁcant P value.9 The summary risk reduction for total mortality with the ICD over amiodarone was 28% (95% CI, 13% to 40%; P ⫽ .0006). For arrhythmic death, the RRR was 50% (95% CI, 33% to 63%; P ⬍ .0001). Over a follow-up period of 6 years, the mean survival was extended by 4.4 months with the ICD (Fig 2). Finally, patients treated with epicardial deﬁbrillators beneﬁted less than those treated with nonthoracotomy devices. This might have contributed to the lack of ICD beneﬁt observed in the CASH trials, in which many of the ICDs were implanted in the epicardial era.
Primary Prevention of SCD The survival rate of an out-of-hospital cardiac arrest is about 6% to 20%.2 Given the dismal prognosis after such an event, we need to identify highrisk individuals and/or prevent sudden cardiac death in them. Various methods have been used to predict SCD among patients with coronary artery disease. These methods include invasive electrophysiologic testing7,27 as well as noninvasive tests such as signal averaged electrocardiograms (SAECG),28 baroreceptor sensitivity,29 low heart rate variability,30,31 nonsustained ventricular
RAJ AND SHELDON
Image Unavailable. Please See Print Journal.
Fig 2. Cumulative risk of overall death and arrhythmic death for amiodarone (Amio) and the ICD treatment arms from the pooled meta-analysis of AVID, CIDS, and CASH. (Reprinted with permission.26)
tachycardia29,32 and a low ejection fraction.33 Many of these tests play an important role in patient selection for trials of the primary prevention of sudden cardiac death. There is a paradox in the identiﬁcation of the patient at risk34-36 (Fig 3). The very well-screened high-risk patients account for a small proportion of subsequent sudden cardiac death events. In contrast, the larger subgroups with lower risk populations account for a much larger number of sudden cardiac deaths. Thus, with exquisite screening, the majority of the patients most likely to die suddenly will be missed. There is no current solution to this problem. There have been 4 primary prevention trials of the ICD that have been completed at the present time: MADIT,37 CABG-Patch,38 MUSTT,39 and AMIOVIRT. The results of the former 3 trials have been published (Table 2), whereas AMIOVIRT has only been reported in abstract form (Table 3).
Multicenter Automatic Deﬁbrillator Trial (MADIT) The rationale for MADIT37,40 was based on studies that showed that patients with poor left ventricular ejection fraction33 and nonsustained ventricular tachycardia29,32 have a high risk of sudden cardiac death. Patients were assigned randomly to
receive (or not receive) an ICD in addition to a background of conventional antiarrhythmic therapy at the discretion of their attending physicians. Patients were eligible for further screening if they had a prior myocardial infarction, a documented
Image Unavailable. Please See Print Journal.
Fig 3. Population impact of published and emerging implantable deﬁbrillator trials. Estimates of incidence and absolute burden of sudden cardiac death among 6 populations are shown. The arrows indicate that the published ICD trials (MADIT, AVID, CASH) affect a small proportion of the total sudden cardiac deaths. Sudden Cardiac Death-Heart Failure Trial (SCD-HeFT) will target a larger fraction of the population at risk for sudden cardiac death. (Reprinted with permission from Myerburg RJ, Mitrani R, Interian, AJ: Interpretation of outcomes of antiarrhythmic clinical design features and population impact. Circulation 97:15141521, 1998.)
Table 3. Ongoing and Publication-Pending Trials of the ICD Trial Secondary Prevention Deﬁbrillator versus ␤-blockers for Unexplained Death in Thailand (DEBUT) Primary Prevention Amiodarone Versus Implantable CardioverterDeﬁbrillator Randomized Trial (AMIOVIRT) Second Multicenter Automatic Implantable Deﬁbrillator Trial (MADIT II) Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT) Deﬁbrillators in Nonischemic cardiomyopathy Treatment Evaluation (DEFINITE) Deﬁbrillator in Acute Myocardial Infarction (DINAMIT) The NORDIC Study
*␤-blocker Strategy plus Implantable CardioverterDeﬁbrillator (BEST-ICD)
Control Group Therapy
Male survivors of sudden death or with characteristic ECG pattern of ST segment elevation in right precordial leads
Signiﬁcant reduction in total mortality with ICD compared with ␤-blockers
Nonischemic cardiomyopathy and LVEF ⬍35% and asymptomatic NSVT
Prior MI and LVEF ⱕ30% but not meeting MADIT inclusion criteria
No reduction in total mortality with ICD compared with amiodarone Ongoing
Amiodarone or placebo
LVEF ⬍35% and NYHA class II or III heart failure
Nonischemic cardiomyopathy and LVEF ⬍35% and symptomatic heart failure Recent MI and LVEF ⬍35% and either mean 24-hour heart rate ⬎80 beats/ min or low HRV Recent MI and LVEF ⫽ 10%-36% and 1 of: low HRV or NSVT or atrial ﬁbrillation or QT dispersion ⬎110 ms Recent MI and LVEF ⬍36% and 1 of: low HRV or late potentials on signal averaged ECG; ICD used only if VT is induced at EP study that cannot be suppressed with AAD
*BEST-ICD is actually a randomized trial of EP study guided therapy versus no antiarrhythmic drug therapy, with the ICD used as 1 option in the EP study guided therapy group. Abbreviations: ECG, electrocardiogram; LVEF, left ventricular ejection fraction; NSVT, nonsustained ventricular tachycardia; MI, myocardial infarction; NYHA, New York Heart Association; HRV, heart rate variability; VT, ventricular tachycardia; EP study, electrophysiological study; AAD, antiarrhythmic drug.
episode of nonsustained ventricular tachycardia within 3 months of enrollment, an ejection fraction of ⱕ35%, and NYHA class I–III heart failure. They were eligible for randomization if they subsequently had inducible ventricular tachycardia during an electrophysiological study that could not be suppressed with procainamide. Patients were excluded if they had a prior cardiac arrest or sustained ventricular arrhythmias, recent myocardial infarction, recent or planned coronary artery revascularization, cardiogenic shock, symptomatic hypotension or NYHA class IV heart failure, or a low likelihood of surviving the duration of the trial because of a major comorbidity. This was clearly a highly selected population with numerous exclusion criteria. MADIT was designed to detect a 46% relative reduction in total mortality
in the ICD group compared with an expected 30% 2-year mortality in the non-ICD group. A 2-sided analysis was planned on an intention-to-treat principle. This allowed for the possibility that the ICD could be deleterious. A total of 196 patients were enrolled in MADIT, 95 in the ICD group and 101 in the non-ICD group. The mean age was 63 ⫾ 9 years, 92% of the patients were men, and median ejection fraction was 26%. Over 60% had NYHA class II–III heart failure, and almost 50% had previous coronary artery bypass grafting. The groups were evenly matched. Over a mean follow-up per patient of 27 months, there were 15 deaths in the ICD group compared with 39 deaths in the non-ICD group, for a RRR of 54% (95% CI, 18% to 74%; P ⫽ .009) with an ICD. Within the ﬁrst year after ICD im-
178 plantation, 40% of patients received a shock, with this number increasing to 90% by 5 years.40 It was not possible to determine whether these shocks were appropriate. Although the beneﬁt was apparently great, there are some concerns about the MADIT trial. The population was very highly selected; indeed, less than 200 patients were enrolled over a 5-year period at 32 institutions. This suggests the real possibility of an entry bias.41 There was no registry of patients who were not included in the study, which limits the generalizability of the study. There was a much higher rate of ␤-blocker use in the ICD group than in the non-ICD group, and ␤-blockers themselves may have an antiarrhythmic role.20,21 Another anomaly concerns the modes of death in the MADIT trial. There was a striking reduction in all-cause mortality in MADIT, which was because of both a decrease in cardiac causes of death as well as a decrease in noncardiac deaths and unknown deaths. There is no apparent biologically plausible mechanism for the ICD to reduce noncardiac death. One possibility is that some deaths were misclassiﬁed, or alternatively, at least some of their ﬁndings may be owed to chance alone. Despite this, MADIT provided the ﬁrst wellrandomized evidence that an ICD could decrease mortality in a high-risk population. ICD implantation rates increased and spiked shortly after the publication of this landmark study.42
The Coronary Artery Bypass Graft (CABG) Patch Trial The CABG Patch Trial was the second randomized clinical trial of an ICD for primary prevention of SCD to be published.38,43 The results of CABG Patch contrast with MADIT results. The CABG Patch trial enrolled patients with coronary artery disease undergoing elective bypass surgery (CABG) who had a left ventricular ejection fraction of less than 36% and an abnormal signal averaged electrocardiogram. Patients were randomized to therapy with an ICD or no speciﬁc antiarrhythmic therapy in addition to the CABG surgery. After a pilot study to determine feasibility, a full-scale mortality trial was started. It was estimated that 800 patients would be required to show a 26% decrease with the ICD of an estimated 40-month control mortality of 36.5%. Early in the
RAJ AND SHELDON
trial, the sample size was increased to 900 in an attempt to offset a lower than expected mortality. Randomization was stratiﬁed both by clinical center and by whether the ejection fraction was greater than 20%. To enrich the arrhythmogenic potential of the population, a signal averaged electrocardiogram (ECG) was used to exclude very low-risk patients from the trial. In the CABG Patch Trial, any 1 of 3 abnormal parameters was considered sufﬁcient for a positive signal averaged ECG. In contrast, traditional criteria often require 2 of the 3 parameters to be abnormal for a positive signal averaged ECG.28 A total of 900 patients were randomized, 446 to the control group and 454 to the ICD group. The mean age was 63 years, and 85% of the patients were men. Almost 25% of the patients were in NYHA class III–IV heart failure, and 55% of the patients had 3-vessel coronary artery disease. The mean left ventricular ejection fraction was 27% ⫾ 6%. ␤-blocker use was similar in the 2 groups. The CABG Patch Trial was terminated prematurely at the fourth interim analysis when it was determined that it was unlikely that any beneﬁt of the prophylactic ICD would be found. During an average follow-up of 32 ⫾ 16 months, there were 101 deaths in the control group and 95 in the control group, and the survival curves superimposable. The hazard ratio for the ICD was 1.07 (95% CI, 0.81 to 1.42; P ⫽ NS). Why were the results of MADIT and the CABG Patch Trial so different? At baseline, the groups appear similar, with comparable mean age, gender, severity of heart failure, and left ventricular ejection fraction. Despite these similarities, the 4-year mortality in the CABG Patch trial was 24% in contrast to the 49% seen in MADIT.44,45 It is possible that the CABG Patch trial screened and enrolled a lower-risk population by using signal averaged ECG instead of the rigid electrophysiologic study criteria used in the MADIT trial. However, it is also possible that the universal revascularization in CABG Patch served to alter the arrhythmic milieu, resulting in a lower rate of signiﬁcant arrhythmias. Further analysis in CABG Patch found that ventricular arrhythmias were inducible in too great a percentage of CABG Patch patients to argue that the cohort was of inherently low risk. Rather, the low rate of sudden cardiac death was likely caused by an uncoupling of the
arrhythmic substrate from the arrhythmia by the revascularization procedure.46
Multicenter Unsustained Tachycardia Trial (MUSTT) The MUSTT study39 was not designed as a primary prevention study of ICD versus medical therapy. Rather, it was a randomized control trial of electrophysiologically guided antiarrhythmic therapy versus no therapy in a high-risk population of patients with coronary artery disease, a left ventricular ejection fraction ⱕ40%, asymptomatic nonsustained ventricular tachycardia, and inducible sustained ventricular arrhythmias during an electrophysiologic study. Importantly, patients who did not respond acutely to antiarrhythmic drugs could receive an ICD. A total of 704 patients with inducible ventricular tachyarrhythmias were randomly assigned to the 2 treatment groups. The 5-year estimates for the primary endpoint of cardiac arrest or death (not total mortality) were 32% for the untreated group and 25% for the drug-treated group (27% relative risk reduction [95% CI, 1% to 47%]). For overall mortality, the 5-year estimates were 48% for the untreated group, and 42% for the electrophysiologically guided therapy group (20% RRR [95% CI, 1% to 36%]). When the groups were subjected to an on-treatment analysis, the entire beneﬁt was restricted to the patients who received an ICD. There was a remarkable 76% relative risk reduction for the primary endpoint associated with ICD use (95% CI, 55% to 87%; P ⬍ .001). The subgroup with ICDs accounted for the entire beneﬁt seen in the group randomized to receive antiarrhythmic therapy. These results must be interpreted with caution because ICD therapy was not randomly allocated. It is possible that the ICD group beneﬁted from a favorable selection bias. This is unlikely, however, because one could only receive an ICD after failing 1 or more drug trials. Given the extremely beneﬁcial response to ICDs in this study, it has almost universally been accepted as a de facto SCD primary prevention of trial.
AMIOVIRT AMIOVIRT is a randomized trial of amiodarone versus an implantable deﬁbrillator that was presented in abstract form at the American Heart As-
sociation Scientiﬁc Sessions in November 2000.47 The results have not yet been published. Patients were eligible if they had a nonischemic dilated cardiomyopathy with an ejection fraction greater than 35%, asymptomatic nonsustained ventricular tachycardia, and NYHA class I–III heart failure. A total of 103 patients were randomized to either ICD (n ⫽ 51) or amiodarone (n ⫽ 52). The mean ejection fraction was 22% to 23%, with similar clinical characteristics in both groups. After a mean follow-up of 21 months, there was no statistically signiﬁcant mortality difference between the two groups, and the study was terminated because of futility. There were 13 deaths in the ICD group, and 8 deaths in the amiodarone group, with 3-year survivals of 79% and 85%, respectively. AMIOVIRT was not a trial of either amiodarone or an ICD versus optimal medical therapy. The more deﬁnitive trial in this regard will be the Sudden Cardiac Death-Heart Failure Trial (SCDHeFT) (see below), which compares ICD to both amiodarone and placebo in addition to optimal medical therapy, and includes patients with nonischemic dilated cardiomyopathy.
All of the Questions Are Not Yet Answered Despite the signiﬁcant amount of clinical trial evidence available regarding the use of the ICD, many issues remain unresolved (Table 4). Some of these remaining issues have begun to be addressed, some are the subjects of ongoing clinical trials, and others may never be completely resolved.
Competing Modes of Death Patients with ventricular tachyarrhythmias typically have other cardiac diseases in addition to their arrhythmias. They often have coronary artery disease, myocardial ischemia, and left ventricular dysfunction. They may suffer from nonarrhythmic cardiac causes and noncardiac causes of death in addition to arrhythmic death. It is not surprising that the ICD does not decrease the numbers of patients dying from nonarrhythmic causes.26 The AHA/ACC recognized the competing modes of death when deriving their ICD implantation guidelines.17 Contraindications to ICD implantation include both terminal illnesses with
180 projected life expectancy less than 6 months and NYHA class IV drug-refractory congestive heart failure in patients who are not candidates for cardiac transplantation. This situation of competing mortality is not unique to the ICD setting. ␤-blocking drugs recently emerged as important treatment for patients with congestive heart failure. However, very few of the patients in the studies that afﬁrmed the beneﬁts of ␤-blockade were unequivocally in NYHA class IV heart failure.48,49 In fact, when speciﬁcally assessed in sicker patients with NYHA III and IV heart failure,50 the ␤-blockade did not result in a signiﬁcant survival beneﬁt. In the meta-analysis of the ICD secondary prevention trials, only 9% of the ICD-treated patients were in NYHA class III or IV,26 and most of these were in NYHA class III. Thus, we have no evidence about the beneﬁts of the ICD in the patient with class IV heart failure.
Groups Excluded From the Studies Several groups were excluded from the various clinical trials of the ICD. A few of these will be discussed next (Table 4). These include patients with hemodynamically stable ventricular tachy-
Table 4. Unresolved Issues Concerning the ICD Competing modes of death Groups excluded from the studies Stable ventricular tachycardia VT/VF caused by reversible causes Syncope in nonischemic dilated cardiomyopathy Rare hereditary disorders associated with ventricular arrhythmias Brugada syndrome Long QT syndrome Hypertrophic cardiomyopathy Arrhythmogenic right ventricular dysplasia Can we expand the number of patients who beneﬁt? SCD-HeFT MADIT II DINAMIT NORDIC BEST-ICD Can we predict who will beneﬁt? Ejection fraction CIDS score AVID low-risk sextile Duration of beneﬁt? Quality of life Cost Cost-effectiveness Cost to society
RAJ AND SHELDON
cardia (especially if the left ventricular ejection fraction is not ⱕ40%), ventricular tachyarrhythmias caused by a transient or reversible cause, and syncope in the setting of dilated cardiomyopathy (DCM).
Stable Ventricular Tachycardias In the AHA/ACC guidelines, spontaneous ventricular tachycardia is a class I indication for an ICD, without regard to ejection fraction. However, there is no properly conducted clinical trial evidence that shows that these patients will beneﬁt from an ICD, although a recent report from the AVID Registry suggests that an ICD may be appropriate.51 The AVID Registry performed follow-up on the ineligible patients and determined mortality data through the National Death Index. Patients with hemodynamically stable ventricular tachycardia had a mortality rate similar to those enrolled patients with hemodynamically unstable cardiac rhythms. Although these data are interesting, they are hypothesis generating and do not prove beneﬁt.
Ventricular Tachyarrhythmia Caused by a Transient or Reversible Cause The deﬁbrillator studies excluded patients with ventricular tachyarrhythmias caused by transient or reversible causes, and this criterion is currently a contraindication to ICD implantation.17 Examples of this are ventricular arrhythmias in the setting of an acute myocardial infarction and in the setting of signiﬁcant electrolyte disturbances. It was assumed that once the transient or reversible event has resolved, the substrate for the ventricular tachyarrhythmia disappeared. This assumption has recently been questioned. The AVID Registry51 patients with transient or correctable causes had a mortality of 17.8%, compared with 17.0% for a ventricular ﬁbrillation cardiac arrest and 15.8% for symptomatic ventricular tachycardia. There may remain a poorly diagnosed and poorly understood substrate for ventricular tachyarrhythmias in these patients.
Syncope With DCM Among patients with nonischemic DCM, syncope is associated with a high risk of sudden cardiac death.52 Electrophysiologic studies have a poor
sensitivity for ventricular arrhythmias in this population.53 Fruhwald et al52 prospectively performed follow-up on 23 patients with syncope and DCM and on 201 patients with DCM but no syncope. The mean left ventricular ejection fraction was 30% in each group, and the patients received follow-up for a mean of 2.5 years. The patients with syncope had a higher proportion of sudden death (5 of 6 deaths) than the control group (13 of 41 deaths, P ⬍ .025). However, there was no difference in overall mortality rates between the syncope group (26%) versus the nonsyncope group (20%, P ⫽ NS). Fonarow et al54 compared survival among patients with DCM referred for heart transplant who had syncope and were treated with an ICD (n ⫽ 25) with those patients treated without an ICD (n ⫽ 122). Actuarial survival at 2 years was 84.9% with ICD therapy and 66.9% without an ICD therapy (P ⫽ .04). These and other55 nonrandomized data suggest that therapy with an ICD might be beneﬁcial among patients with a DCM and syncope.
Rare Hereditary Disorders Associated With Ventricular Arrhythmias Randomized clinical trials provide the highest standard of evidence available to aid in clinical decision making. However, large numbers of patients are often required to provide reliable answers to a question. There are several groups of patients with relatively rare hereditary disorders associated with ventricular arrhythmias for whom a proper randomized trial might not be feasible. Some of these patients may suffer from high rates of arrhythmias and sudden death. In primarily cardiac electrical disorders, there may not be a signiﬁcant risk of a competing mode of death in the predominantly young patient population. Randomized data from large trials may never be available for these types of patients, and clinical decisions will have to be made with the more limited data available. These groups of patients have a class IIb indication for implantation of an ICD.17
Sudden Unexplained Death Syndrome (Brugada Syndrome) Sudden unexplained death syndrome is believed to be a genetic disorder that predominantly affects Southeast Asian patients with structurally normal hearts. The majority of deaths occur while the
patients are asleep. This disorder has a characteristic electrocardiographic pattern and is very similar to the syndrome described by Brugada et al.56-59 The Deﬁbrillator Versus ␤-Blockers for Unexplained Death in Thailand (DEBUT) randomized 61 men who had either survived an episode of SCD or had the characteristic ECG pattern to receive either ␤-blockers or an implantable deﬁbrillator (Table 3). The results have not been published. Over a mean follow-up of 24 months, 11 patients experienced an arrhythmia characteristic of sudden death. In the ␤-blocker arm, 4 people died, whereas no one died in the ICD arm. However, 7 people in the ICD arm received 1 or more appropriate therapies for lifethreatening ventricular arrhythmias. The study was terminated prematurely owing to the excess mortality in the ␤-blocker group. Interpreting the results is difﬁcult because a quite heterogeneous group was studied. Recent clinical trials have already determined that ICDs confer a survival advantage among those patients who have been resuscitated from ventricular ﬁbrillation.26 It is not clear how large a proportion of the DEBUT study population had idiopathic ventricular ﬁbrillation, and how many were enrolled based on an ECG suggestive of the sudden unexplained death syndrome, but without prior arrhythmia. We must await publication of the trial results before more deﬁnitive conclusions can be drawn.
Long QT Syndrome The congenital long QT syndrome (LQTS) is a genetic ion channel disorder that is associated with torsades de pointes ventricular tachycardia and ventricular ﬁbrillation. Despite therapy with ␤-blockers, 30% of patients continue to have syncope and death.60 It would be difﬁcult to conduct a randomized controlled trial in this population because of the rarity of this disorder. Accordingly, the University of Rochester has launched a registry of all patients with long QT syndrome implanted with an ICD and is tracking appropriate ICD discharges as a surrogate endpoint for avoided SCD.
Hypertrophic Cardiomyopathy Hypertrophic cardiomyopathy is an uncommon genetic myocardial disorder that affects 1 in 500 people.61 Sudden death, often caused by ventric-
182 ular arrhythmia, occurs in the overall hypertrophic cardiomyopathy population and is more common in high-risk subgroups.62,63 Because of the small number of high-risk patients, a randomized control trial of ICD therapy is unlikely.64 Maron et al64 published a retrospective study of ICD discharges in a cohort of patients with hypertrophic cardiomyopathy. Among patients without prior ventricular arrhythmias, the authors reported a 5% yearly rate of appropriate ICD discharge. Despite the fact that only some of those appropriate discharges might have been for lethal arrhythmias, these data suggest that therapy with an ICD be considered in this high-risk group. There is no higher level evidence.
Arrhythmogenic Right Ventricular Dysplasia Arrhythmogenic right ventricular dysplasia (ARVD) is an uncommon disorder associated with ﬁbrofatty inﬁltration and replacement of the myocardium.65 Patients with this disorder are prone to ventricular arrhythmias. Antiarrhythmic drug therapy has proven mostly ineffective in this group of patients.66 Implantable deﬁbrillator therapy is an attractive alternative in these patients, despite the lack of any randomized trial data. There are currently attempts to form a registry67 that may provide useful information in the future about discharge rate.
Can We Increase the Numbers of Patients Who Beneﬁt From the ICD? ICD trials select high-risk populations for inclusion because patients who are not at risk of lethal ventricular tachyarrhythmias are not as likely to beneﬁt from this expensive and invasive treatment for ventricular tachyarrhythmias. Dr A.J. Moss has stated, “the sickest patients beneﬁt the most.”68 Although this strategy has improved the power and internal validity of the studies, it has limited our ability to generalize the data to the majority of patients who might beneﬁt. The populations at very high risk of sudden cardiac death (measured in annual incidence) account for a small percentage of the total number of people who suffer a sudden cardiac death event.36 In contrast, most of the sudden deaths occur in populations that have lower average risks. As shown in Fig 3, the published trials of the ICD have addressed only the
RAJ AND SHELDON
very high-risk groups and thus only a small proportion of the overall burden of sudden cardiac death. Two ongoing trials of the ICD, SCD-HeFT and Second Multicenter Automatic Implantable Deﬁbrillator Trial (MADIT II) (Table 3), are assessing whether the beneﬁts of the ICD extend to a broader population.
SCD-HeFT The SCD-HeFT69 aims to determine whether either amiodarone or an ICD will improve survival compared with placebo in patients with NYHA class II and class III heart failure and reduced left ventricular ejection fraction (ⱕ35%). This study includes patients with coronary artery disease or nonischemic dilated cardiomyopathy. The primary outcome is all-cause mortality. Optimal contemporary heart failure treatment, including angiotensin-coverting enzyme inhibitors, is mandated. Therapy with ␤-blockers is recommended but ultimately left to the discretion of the treating physicians. The anticipated sample size is 2,500 patients, based on an expected 25% mortality at 2.5 years. The study is powered to detect a 25% relative reduction in mortality.
MADIT II The MADIT II is a streamlined follow-up to MADIT.70,71 The enrollment criteria include a depressed ejection fraction (⬍30%) at least 1 month after a myocardial infarction and at least 3 months after coronary artery revascularization. Two hundred patients will be randomized 3:2 to an ICD or no ICD. The primary outcome is overall mortality. The control group is expected to have a 19% 2-year mortality rate (9.5% sudden death rate), and the authors anticipate that the ICD will decrease the event rate by 80%.
The Deﬁbrillator in Acute Myocardial Infarction Trial (DINAMIT) The DINAMIT72 is a study of ICD versus no ICD in high-risk survivors of myocardial infarction. Entry criteria are a combination of a low ejection fraction (ⱕ35%) and either depressed heart rate variability or a mean 24-hour heart rate of ⱖ80 beats per minute.
The Nordic Study The Nordic study is a Danish study testing the efﬁcacy of the ICD in patients with acute myocardial infarction who are at high risk of sudden death because of a low ejection fraction (10% to 36%) and at least 1 of the following: reduced heart rate variability, nonsustained ventricular tachycardia on Holter monitor atrial ﬁbrillation, or QT interval dispersion greater than 110 ms. A total of 400 patients will be randomized equally to the 2 groups, with a primary endpoint of all-cause mortality. 70
␤-Blocker Strategy-Implantable CardioverterDeﬁbrillator (BEST-ICD) Trial The BEST-ICD70,73 is a trial comparing electrophysiologic study guided therapy (with either a drug or an ICD) with conventional therapy, which is contemporary cardiac therapy without speciﬁc antiarrhythmic drug therapy. Eligible patients include those with a recent myocardial infarction, an ejection fraction less than 36%, and another high-risk factor (reduced heart rate variability or the presence of ventricular late potentials on signal averaged ECGs). More than 1,500 patients will be randomized, with 60% receiving electrophysiologic guided therapy. If drugs are effective at arrhythmia suppression then they will be prescribed, otherwise the patients will receive an ICD. The primary endpoint is total mortality.
Can We Predict Which Patients Will Beneﬁt From an ICD? Although some patients derive great beneﬁt from an ICD, this beneﬁt does not extend to everyone. Given that ICD therapy is expensive74 and is not free of risks,75 it would be ideal to provide ICDs mainly to those patients who will derive beneﬁt from them. Risk stratiﬁcation has been attempted both from retrospective analyses (some of which were prespeciﬁed) of the published ICD trials and from ongoing ICD trials.
Ejection Fraction Left ventricular dysfunction is a strong predictor of a poor prognosis after an episode of syncope or ventricular arrhythmia.76,77 However, because of the risks of competing mortality from nonsudden
causes of death, some questioned whether an ICD would be of beneﬁt in patients with reduced left ventricular function. However, subgroup analysis from both CIDS78 and AVID25 showed no mortality beneﬁt from the ICD compared with amiodarone in patients with a left ventricular ejection fraction greater than 35%. Domanski et al25 divided the AVID population into 3 groups based on left ventricular ejection fraction. A statistically signiﬁcant beneﬁt was seen in the group with an ejection fraction between 20% to 35%. A similar risk reduction was seen in the group with an ejection fraction less than 20%, but the CIs were wide because of small numbers in this group. Recently, a subanalysis from MADIT68 showed that the survival advantage from an implantable deﬁbrillator over conventional therapy was particularly marked in the group with an ejection fraction less than 26% (Fig 4). The survival advantage conferred by the ICD was not statistically signiﬁcant in the group with an ejection fraction between 26% to 35%. All of these analyses are from post hoc subgroups and as such truly serve only to generate hypotheses. However, the dependence of beneﬁt on ejection fraction has been a consistent and reproducible ﬁnding across multiple studies, and it passes the test of biologic plausibility. The purist could argue that a trial is needed to test the beneﬁt (or lack thereof) of ICDs over amiodarone among patients with an ejection fraction greater than 35%. However, among the patients with aquired ventricular arrhythmias (in contrast with primary genetic arrhythmic disorders), it seems likely that the beneﬁts of the ICD are mostly limited to those patients with poor left ventricular ejection fractions.
CIDS Score We79 conducted a subgroup analysis of the CIDS study9 with the intent of predicting risk of death based on entry clinical criteria. These were advanced age (ⱖ70 years), poor left ventricular ejection fraction (ⱕ35%), and poor functional status (at least NYHA class III). This predicted a smooth and highly signiﬁcant gradation in risk of death. To assess the beneﬁt of the ICD in the sickest patients, we divided the population into risk quartiles. In the highest risk quartile, the ICD provided a 50% RRR, whereas there was not a signiﬁcant
RAJ AND SHELDON
Image Unavailable. Please See Print Journal.
98% speciﬁcity in CIDS, and 79% sensitivity and 91% speciﬁcity in AVID.80 The beneﬁt of the ICD was concentrated in 25% of patients in CIDS and in 31% of patients in AVID. Patients with 2 or more risk factors in both the CIDS and AVID studies had approximately 50% reductions in the likelihood of death with ICD therapy, whereas the remaining patients were very unlikely to beneﬁt from ICD therapy over medical therapy. Thus, this simple clinical tool can select most of the patients who beneﬁt most, but cannot completely exclude the possibility of beneﬁt in the lower-risk group. The use of the tool should be tempered by clinical judgment. Neither CIDS nor AVID had large populations of octogenarians or patients with terminal class IV heart failure, and the competing modes of death in these patients may simply overwhelm the beneﬁt of the ICD.
AVID Low-Risk Group Rather than seek to identify patients most likely to beneﬁt from the ICD, could we identify patients least likely to beneﬁt? In a prespeciﬁed analysis, the AVID investigators divided the entire cohort into risk sextiles.81 There was no beneﬁt of ICD
Fig 4. The impact of baseline ejection fraction on the beneﬁt of an ICD in MADIT. The ICD conferred a signiﬁcant survival advantage in the group of patients with an ejection fraction less than 26% (top curves), but this difference was not statistically signiﬁcant for the group with an ejection fraction between 26% to 35% (bottom curves). (Reprinted with permission from Moss AJ: Implantable cardioverter deﬁbrillator therapy: The sickest patients beneﬁt the most. Circulation 101:1633-1640, 2000.)
beneﬁt of the ICD in the remaining 3 lower risk quartiles (Fig 5). Therefore, in this study, the sickest patients did beneﬁt the most. From this, we created a simple score that predicted beneﬁt from the ICD compared with amiodarone therapy. Patients received 1 point for each of the following: age greater than 70 years, left ventricular ejection fraction less than 35%, and NYHA class III or IV. Thirteen of the ﬁfteen deaths prevented by the ICD occurred among patients with 2 or more risk factors. Use of a dichotomous score of 2 or more risk factors identiﬁed the patients who would beneﬁt from ICD therapy with 87% sensitivity and
Image Unavailable. Please See Print Journal.
Fig 5. Risk stratiﬁcation in CIDS. The ICD decreased the risk of death, compared with amiodarone, only in the quartile of patients with the highest risk (Q4). No beneﬁt was seen in the 3 lower risk quartiles (Q123). (Reprinted with permission from Sheldon R, Conrolly S, Krahn A, et al: Identiﬁcation of patients most likely to beneﬁt from implantable cardiverter-deﬁbrillator therapy: The Canadian Implantable Deﬁbrillator Study. Circulation 101:1660-1664, 2000.)
over antiarrhythmic drug therapy in the lowest risk sextile (0.03 ⫾ 0.12 year beneﬁt for antiarrhythmic drug therapy), whereas there was a signiﬁcant beneﬁt of the ICD in each of the 5 highest risk sextiles. By using a hazards model, they found that patients presenting with an isolated episode of ventricular ﬁbrillation in the absence of cerebrovascular disease and in the absence of prior arrhythmia who have undergone revascularization or have a left ventricular ejection fraction greater than 27% are not likely to beneﬁt from ICD therapy compared with amiodarone therapy. Although all the risk stratiﬁcation analyses are interesting, they require prospective validation before they can be judged to be rock-solid underpinnings for clinical decisions or health services provision.
Does the Beneﬁt of the ICD Persist? For how long do the beneﬁts of the ICD last? The trial data provide information for a follow-up period ranging 3 to 6 years,74,82,83 but we all hope that our patients will live longer than 6 years. What happens to the beneﬁts of the ICD over a long period? Unfortunately, owing to the early termination of some studies,8 we do not know the answer. Caution must be exercised in assuming that the beneﬁts of the ICD persist in long-term follow-up. It is not clear if the survival curves will continue to diverge, remain parallel, or converge given a longer follow-up. The populations that were studied have other competing modes of death (arrhythmic death versus nonarrhythmic death), and although the ICD might prevent an arrhythmic death, it is unlikely to decrease nonarrhythmic death.26 A cautionary example may be drawn from the Veterans Affairs Cooperative Study of Coronary Artery Bypass Surgery, which randomized high-risk patients with stable angina to bypass surgery or medical therapy.84,85 The investigators found that over a short time horizon (5 years), coronary artery bypass surgery afforded a signiﬁcant mortality advantage over medical therapy (Fig 6A).84,85 Note that in Figure 6A the survival curves continue to diverge, suggesting a consistent RRR. However, with longer follow-up, the survival curves converged, and the survival advantage disappeared completely after 14 years (Fig 6B). This is a graphic illustration of the peril in extrapolating the results of relatively brief stud-
Fig 6. Survival curves for the patients with high angiographic risk from the Veterans Affair Cooperative Study of Coronary Artery Bypass Surgery randomized to bypass surgery (Surgical) or medical therapy (Medical). (A) A marked survival advantage in the Surgical group when the groups were followed-up out to 7 years. (B) With longer follow-up, the survival advantage for the surgical group disappears, and the survival curves cross at 14 years of follow-up. Data from reference 85.
ies to much longer horizons. It is not reasonable to simply extrapolate the study data out to any desired time horizon. Similar problems, we suggest, are likely to occur in long-term studies of ICD beneﬁt. Consider the AVID and CASH results. The length of follow-up in the AVID study was short (mean follow-up was 2.6 years) because of early study termination. The survival curves are still divergent through the length of follow-up (Fig 7A). In contrast, the follow-up in the CASH study was much longer, with survival estimated out of 9 years in their Kaplan-Meier curves (Fig 7B). In the CASH study, the relative beneﬁt of the ICD over medical therapy diminished after the ﬁrst 5 years of followup.10 Note the similarities between the AVID results and the short-term bypass surgery results, and the CASH results and the longer-term bypass surgery results. The cumulative follow-up in the
RAJ AND SHELDON
Image Unavailable. Please See Print Journal.
Fig 7. Survival curves from 2 secondary prevention ICD trials, AVID and CASH. (A) Survival curves from AVID. At 3 years of follow-up, a divergence in the curves is seen indicating ongoing beneﬁt from the deﬁbrillator. (Reprinted with permission from the Antiarrhythmics Versus Implantable Deﬁbrillators (AVID) Investigators: A comparison of antiarrhythmic-drug therapy with implantable deﬁbrillators in patients resuscitated from near-fatal ventricular arrhythmias. N Engl J Med 337:1576-1583, 1997. Copyright © 1997 Massachusetts Medical Society. All rights reserved. (B) Survival curves from CASH. Initially, there is a survival advantage for the ICD over the combined amiodarone and metoprolol group (AMIO/METO), but this survival advantage diminishes after 5 years of follow-up. (Reprinted with permission from Kuck KH, Cappato R, Siebels J, et al: Randomized comparison of antiarrhythmic drug therapy with implantable deﬁbrillators in patients resuscitated from cardiac arrest: The Cardiac Arrest Study Hamburg (CASH). Circulation 102:748-754, 2000.)
various ICD trials is not long enough to determine whether the results from the CASH study represent the true long-term beneﬁts of the ICD or whether they are a statistical anomaly. However, they do illustrate some of the difﬁculties that arise in interpreting the trial data as a result of early study termination.
Quality Of Life Most of the current literature in support of the ICD has focused on all-cause mortality. This is the hardest endpoint, and one that has become the
standard outcome measure of many cardiovascular intervention trials. However, it is not the only important outcome. Increasingly, society is focusing not only on the quantity of life extension but also on the quality of that life. There are only limited data regarding quality of life among patients with ICDs or among patients randomized to ICD or medical therapy. One prospective study from Kaiser Permanente of quality of life among patients with life-threatening ventricular arrhythmias found that there was greater improvement in physical functioning and symptom scores in those patients initially treated with
an ICD than amiodarone.86 An AVID substudy reported that a better baseline quality of life (measured using 3 different tools) predicted a lower mortality in the overall trial, without regard to treatment group.87 This second study also suggests that the Kaiser study may have had a selection bias owing to the nonrandomized nature of ICD distribution. The more pertinent question is whether therapy with an ICD improves a patient’s quality of life. Despite the insurance policy qualities of the ICD, not all patients feel well after ICD implantation. There have been published reports of both ﬂorid psychopathology88 and affective distress requiring therapy89 among ICD recipients receiving shocks. Patients who received an ICD in the CABG-Patch study reported lower levels of psychological well-being at 6 months than did the control patients.90 This diminished sense of wellbeing was most marked among those patients who received shocks. This ﬁnding is similar to results from other studies.91 A recent abstract from AVID investigators showed that certain quality-of-life questions were associated with a higher risk of arrhythmic mortality and that treatment with an ICD reduced that risk.92 However, there has not yet been published information regarding the effect of an ICD on follow-up quality of life in the randomized ICD trials.
Cost-Effectiveness Analysis ICD therapy has shown mortality beneﬁts in several groups of patients.26,37 However, because ICD therapy is expensive, the question of whether society can afford such a therapy has been raised.93 If one assumes that there is a limit to health care resources, then there comes a zero-sum game.94 Given the various beneﬁcial health care initiatives competing for limited health care resources, does the increased availability of ICD therapy provide society optimal use of its resources?95
Basic Principles A cost-effectiveness analysis involves the quantiﬁcation of cost and health outcome for various treatments for a given disorder.95 The primary health outcomes in these studies are differences in life expectancy. This is an incremental cost-effectiveness analysis (iCEA), which is the difference in costs (C)
between the two therapies divided by the difference in life expectancy (LE) between the two therapies (iCEA⫽[C1-C2]/[LE1-LE2]). The results are expressed in dollars per life-year saved.95 By convention, therapies are considered cost-effective if the cost per life-year saved ($LYS), in United States dollars, is between $20,001 to $40,000; are considered borderline cost-effective if the $LYS is between $40,001 to $60,000; and are considered expensive or unattractive if the $LYS is greater than $60,000.96 Therapies are considered highly cost-effective if the $LYS is ⱕ$20,000. The ideal, but unfortunately rare, situation occurs when a therapy both saves money and prolongs life, in which case it is referred to as a dominant therapy.
Early Decision Analysis Models In the 1990s, several cost-effectiveness analyses of ICD therapy were performed by using decision analysis models.97-100 These models considered the cost-effectiveness of ICD among patients who were survivors of ventricular arrhythmias. Therapy with an ICD was compared with drug therapy, which was amiodarone in the majority of the cases. Unless otherwise stated, all costs per life-year gained ﬁgures are presented in United States dollars. Kuppermann et al97 compared costs for 203 patients treated with an ICD with historic control subjects from the pre-ICD era treated with conventional drug therapy. Mortality data were extracted from observational data in the contemporary published literature. Over a projected lifetime time horizon, the incremental cost-effectiveness was $7,400 per year of life saved. This was based on an estimated gain in life of 1.9 years. Larsen et al98 compared ICD therapy with amiodarone over a lifetime time horizon. Costs were based on just 64 patients from the New England Medical Center (21 ICD-treated patients and 43 amiodaronetreated patients). The estimated gain in life from an ICD over amiodarone was 2.22 years, with a cost-effectiveness ratio of $13,800 per year of life saved. This model uses a very small sample to determine cost information and, like the Kuppermann analysis, survival information was based on observational data from the literature. Kupersmith et al99 created a model comparing ICD therapy with electrophysiologically guided drug therapy. Cost data were extracted from a Medicare database. Control mortality was deter-
188 mined based on a time to ﬁrst event analysis (ﬁrst appropriate shock or death). This method is biased in favor of overestimating the control mortality since not all ﬁrst events would have resulted in death. They estimated a gain of 1.72 years over a 6-year time horizon. The incremental cost-effectiveness ratio was calculated to be $31,100 per year of life saved. Owens et al100 used a decision analysis model of ICD versus amiodarone assessed over a lifetime time horizon. Estimates of effectiveness were taken from preliminary clinical trial and ICD registry data. They found that their cost-effectiveness estimates were very sensitive to assumptions about the incremental beneﬁt of the ICD over amiodarone. If the ICD reduced the mortality rate by 40%, then the incremental cost-effectiveness was $37,300 per life year saved. However, if the mortality rate reduction with an ICD was only 20%, then the incremental cost-effectiveness increased to $74,400 per life-year saved. With the exception of the Owens model (with assumptions of a 20% 1-year mortality rate reduction with an ICD), the early decision analysis models all put estimates of the incremental cost-effectiveness of an ICD over amiodarone (or drug therapy) in the cost-effective range. These studies are all limited by their effectiveness information, which is based on observational data.95 These trials greatly overestimated ICD effectiveness, with estimates of life-years gained around 2 years. The Owens model,100 in assuming a mortality reduction of only 20%, may be closer to contemporary clinical reality.
Cost-Effectiveness From Randomized Clinical Trial Data Given the limitations of using retrospective and observational data in the aforementioned models, cost-effectiveness analyses were imbedded into the recent randomized ICD clinical trials.
The Dutch Study The Dutch study randomly assigned 60 postinfarct survivors of cardiac arrest to either an early ICD or electrophysiologic study guided therapy (antiarrhythmic drugs followed, if necessary, by a deﬁbrillator).101 By the end of the study, over 50% of the patients in the electrophysiologic-guided therapy group had received an ICD and 6 of the
RAJ AND SHELDON
patients underwent ventricular tachycardia surgery. The ICD was found to be a dominant therapy because it was associated with both a cost-saving and a survival advantage. This information has limited applicability to contemporary practice. Electrophysiologic-guided antiarrhythmic drug therapy is not widely used because of the results of the ESVEM study.6,12 The very high rate of ICD use in the electrophysiologic-guided therapy group combined with the use of ventricular tachycardia surgery may have inﬂated the costs in the drug treatment group. Thus, these data do not provide guidance about the contemporary costeffectiveness of ICDs.
MADIT The MADIT cost-effectiveness analysis looked at the beneﬁt of ICDs over conventional therapy in the primary prevention of sudden death in a highrisk population.82 The authors used actual costs from both inpatient and outpatient resource utilization in the MADIT study37 using a time horizon of 4 years and a 3% discounting of both the costs and the beneﬁts. The net costs for the ICD group were $21,580 (US dollars) greater than for the conventional therapy group. The mean survival was 0.86 years greater in the ICD group over the 4-year time horizon. The resulting incremental cost-effectiveness ratio was $27,000 per life-year saved. The mean survival beneﬁt was much greater in MADIT82 than in the secondary prevention studies.8-10 This large survival beneﬁt is largely responsible for the very favorable cost-effectiveness ratio.
AVID The cost-effectiveness analysis from the AVID study has yet to be published but has been presented in abstract form.83 The investigators assessed charges instead of cost (as in MADIT). The abstract is based on analysis of the 87% of the data available. Over a 3-year time horizon, the ICD group had a mean 0.24-year survival advantage over the antiarrhythmic drug group, but also a mean incremental charge of $27,580. The charges were $114,917 per year of life saved by the ICD over antiarrhythmic drug therapy. Although we await the ﬁnal published results, these preliminary data suggest that the beneﬁts of the ICD over
amiodarone were unattractive from a cost-effectiveness perspective.96
CIDS The CIDS economic analysis74 was conducted on the ﬁrst 430 patients enrolled in the CIDS,9 but was extrapolated to the entire study population. The CIDS analysis had the longest time horizon (6 years) of the clinical trial-based economic analyses. The ICD prolonged life by a mean of only 0.23 years over this time span. Costs were calculated from a single payer (provincial government) viewpoint, with all patients assumed to have a nonthoracotomy ICD for the purpose of the cost calculation.74 Costs were reported in 1999 Canadian dollars (C$1 ⬇ US$0.65). The mean cost per patient was C$49,115 greater in the ICD group than in the amiodarone group. The incremental costeffectiveness of ICD over amiodarone was C$213,543 per life-year gained. These clearly fall into an unattractive cost-effectiveness range.96 If analysis were restricted to patients with a left ventricular ejection fraction less than 35%, the cost-effectiveness becomes more attractive at C$108,484 per life-year saved because of a greater beneﬁt from the ICD over amiodarone.74 If patients had 2 or more CIDS risk factors79 (age ⱖ 70 years, left ventricular ejection fraction ⱕ35%, and NYHA class III or IV), the incremental cost of the ICD decreases to C$65,195 per year of life saved, which would be cost-effective.102 O’Brien et al74 also performed cost analysis with extrapolations out to 12 years (without assuming that the ICD generator would require replacement). The incremental cost-effectiveness ratios ranged from C$99,420 to C$149,710 per year of life saved depending on whether the authors assumed that the survival curves continue to diverge or converge beyond the available data.74 The ICD did not become attractive from a cost-effectiveness point of view even if it was assumed that the length of inpatient stay for an ICD implant was reduced to only 1 day.
Summary of Cost-Effectiveness Trial Data The early decision analysis models used overly optimistic projections of the survival beneﬁt of ICD over antiarrhythmic drug therapy. This consequently led to incremental cost-effectiveness estimates that were signiﬁcantly more favorable
than has been borne out from recent clinical trials. The cost-effectiveness of the primary prevention of sudden cardiac death (MADIT82) falls into a cost-effective range,96 whereas the secondary prevention of sudden cardiac death (AVID,83 CIDS74) has proven unattractive from a cost-effectiveness perspective. The difference in cost-effectiveness among MADIT, AVID, and CIDS is because of the much larger mortality beneﬁt seen in MADIT37 than in the other 2 studies.8,9 More accurate patient selection for ICD therapy could make that therapy available at a favorable cost-effectiveness ratio.
Cost Implications of ICDs Although the cost-effectiveness of a therapy is important to determine its relative value, another important issue, from a societal perspective, is the overall impact on the health care system. The widespread use of the ICD may provide a beneﬁt or a reasonable cost per life saved but may be so collectively expensive that it interferes with other societal priorities.103 What is the impact of the ICD trials? Pathmanathan et al104 used a registry to determine the impact of strict adherence to the AVID criteria to the Midlands region of the United Kingdom. During a 5-month follow-up, 69 patients (of the 132 patients entered into the registry) fulﬁlled the AVID criteria. Extrapolation of these data suggests the implantation of at least 166 new ICDs over a 12-month period compared with 23 implants in the prior year. Strict adherence to the AVID criteria would increase the ICD implant rate in the United Kingdom from 5 per million to at least 18 per million population. This was estimated to cost £24.1 million per year.104 The ICD implantation rates vary markedly throughout the world. The United States had an implant rate of greater than 175 implants per million population in 1999, which was much greater than the European Union implant rate of 34 per million population.105 Even in the European Union, the difference in implant rates range from a high of over 60 per million population in Germany to 12 to 13 per million population in France and the United Kingdom. There are likely multiple reasons for these differences. These might include a more litigious environment in the United States than in Europe, or the current trial evidence may have more greatly swayed American physi-
RAJ AND SHELDON
cians than the European physicians. However, it is also likely that the very different health care structure and the varied health care payers also contribute to the apparent differences. There is also the difﬁculty of indication extrapolation.103 Clinicians are often faced with difﬁcult patient scenarios that either do not ﬁt into available clinical trials or ﬁt only on the periphery. In these situations, the intuitive appeal of the ICD as a life insurance policy is hard to ignore. However, this approach might be prohibitively expensive from a societal perspective. Some have urged us to practice family-based medicine.106 That is, our only focus should be on treating patients as we would treat our relatives. However, we cannot lose sight of the societal impact of our decisions. Resources committed to 1 particular therapy or disease are ultimately not available to commit to other therapies or diseases. These are difﬁcult times, and we face ethical and moral issues that our predecessors have been spared. We are custodians of the health of the population as well as the health of single patients.
Summary The ICD represents a signiﬁcant advance in the treatment of patients at high risk of ventricular tachyarrhythmias. Several clinical trials have provided clinicians with guidance as to which patients will beneﬁt, and there are several other important trials still under way. However, the beneﬁts of this therapy do not appear to be evenly distributed. Rather, it may be that the sickest patients beneﬁt the most.68 Our challenges include developing highly sensitive and speciﬁc tools for identifying patients at risk for life-threatening ventricular arrhythmias and similarly developing simple ICDs that are sufﬁciently inexpensive so that we can use them widely to prevent sudden death caused by these arrhythmias.
References 1. Gillum RF. Sudden coronary death in the United States: 1980 –1985. Circulation 79:756-765, 1989 2. de Vreede-Swagemakers JJ, Gorgels AP, DuboisArbouw WI, et al: Out-of-hospital cardiac arrest in the 1990s: A population-based study in the Maastricht area on incidence, characteristics and survival. J Am Coll Cardiol 30:1500-1505, 1997 3. Holmberg M, Holmberg S, Herlitz J: The problem of out-of-hospital cardiac-arrest prevalence of sudden
death in Europe today. Am J Cardiol 83:88D-90D, 1999 Mitchell LB, Duff HJ, Manyari DE, et al: A randomized clinical trial of the noninvasive and invasive approaches to drug therapy of ventricular tachycardia. N Engl J Med 317:1681-1687, 1987 Mitchell LB, Duff HJ, Gillis AM, et al: A randomized clinical trial of the noninvasive and invasive approaches to drug therapy for ventricular tachycardia: long-term follow-up of the Calgary trial. Prog Cardiovasc Dis 38:377-384, 1996 Mason JW: A comparison of electrophysiologic testing with Holter monitoring to predict antiarrhythmicdrug efﬁcacy for ventricular tachyarrhythmias. Electrophysiologic Study versus Electrocardiographic Monitoring Investigators. N Engl J Med 329:445451, 1993 Buxton AE, Lee KL, DiCarlo L, et al: Electrophysiologic testing to identify patients with coronary artery disease who are at risk for sudden death. Multicenter Unsustained Tachycardia Trial Investigators. N Engl J Med 342:1937-1945, 2000 The Antiarrhythmics Versus Implantable Deﬁbrillators (AVID) Investigators: A comparison of antiarrhythmic-drug therapy with implantable deﬁbrillators in patients resuscitated from near-fatal ventricular arrhythmias. N Engl J Med 337:1576-1583, 1997 Connolly SJ, Gent M, Roberts RS, et al: Canadian implantable deﬁbrillator study (CIDS): A randomized trial of the implantable cardioverter deﬁbrillator against amiodarone. Circulation 101:1297-1302, 2000 Kuck KH, Cappato R, Siebels J, et al: Randomized comparison of antiarrhythmic drug therapy with implantable deﬁbrillators in patients resuscitated from cardiac arrest: The Cardiac Arrest Study Hamburg (CASH). Circulation 102:748-754, 2000 Anonymous: Effect of prophylactic amiodarone on mortality after acute myocardial infarction and in congestive heart failure: Meta-analysis of individual data from 6500 patients in randomised trials. Amiodarone Trials Meta-Analysis Investigators. Lancet 350:1417-1424, 1997 Mason JW: A comparison of seven antiarrhythmic drugs in patients with ventricular tachyarrhythmias. Electrophysiologic Study versus Electrocardiographic Monitoring Investigators. N Engl J Med 329: 452-458, 1993 Mirowski M, Mower MM, Staewen WS, et al: Standby automatic deﬁbrillator. An approach to prevention of sudden coronary death. Arch Intern Med 126:158-161, 1970 Mirowski M, Mower MM, Langer A, et al: A chronically implanted system for automatic deﬁbrillation in active conscious dogs. Experimental model for treatment of sudden death from ventricular ﬁbrillation. Circulation 58:90-94, 1978 Mirowski M, Reid PR, Mower MM, et al: Termination of malignant ventricular arrhythmias with an implanted automatic deﬁbrillator in human beings. N Engl J Med 303:322-324, 1980
IMPLANTABLE CARDIOVERTER-DEFIBRILLATOR 16. Mirowski M, Mower MM, Reid PR: The automatic implantable deﬁbrillator. Am Heart J 100:1089-1092, 1980 17. Gregoratos G, Cheitlin MD, Conill A, et al: ACC/AHA guidelines for implantation of cardiac pacemakers and antiarrhythmia devices: A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on Pacemaker Implantation). J Am Coll Cardiol 31:11751209, 1998 18. Wever EF, Hauer RN, van Capelle FL, et al: Randomized study of implantable deﬁbrillator as ﬁrst-choice therapy versus conventional strategy in postinfarct sudden death survivors. Circulation 91:2195-2203, 1995 19. Zipes DP: The antiarrhythmics versus implantable deﬁbrillator (AVID) trial, in Woosley RL, Singh SN (eds): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial Evidence. New York, NY, Marcel Dekker, 2000, pp 171-184 20. Burkart F, Pﬁsterer M, Kiowski W, et al: Effect of antiarrhythmic therapy on mortality in survivors of myocardial infarction with asymptomatic complex ventricular arrhythmias: Basel Antiarrhythmic Study of Infarct Survival (BASIS). J Am Coll Cardiol 16: 1711-1718, 1990 21. Teo KK, Yusuf S, Furberg CD: Effects of prophylactic antiarrhythmic drug therapy in acute myocardial infarction. An overview of results from randomized controlled trials. JAMA 270:1589-1595, 1993 22. Boutitie F, Boissel JP, Connolly SJ, et al: Amiodarone interaction with beta-blockers: Analysis of the merged EMIAT (European Myocardial Infarct Amiodarone Trial) and CAMIAT (Canadian Amiodarone Myocardial Infarction Trial) databases. The EMIAT and CAMIAT Investigators. Circulation 99:22682275, 1999 23. Singh BN: AVID: Critique, in Woosley RL, Singh SN (eds): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial Evidence. New York, NY, Marcel Dekker, 2000, pp 185-189 24. Singh SN, Woosley RL: CASH: Critique, in Woosley RL, Singh SN (eds): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial Evidence. New York, NY, Marcel Dekker, 2000, pp 289-290 25. Domanski MJ, Sakseena S, Epstein AE, et al: Relative effectiveness of the implantable cardioverterdeﬁbrillator and antiarrhythmic drugs in patients with varying degrees of left ventricular dysfunction who have survived malignant ventricular arrhythmias. AVID Investigators. Antiarrhythmics Versus Implantable Deﬁbrillators. J Am Coll Cardiol 34: 1090-1095, 1999 26. Connolly SJ, Hallstrom AP, Cappato R, et al: Metaanalysis of the implantable cardioverter deﬁbrillator secondary prevention trials. AVID, CASH and CIDS studies. Antiarrhythmics versus Implantable Deﬁbrillator Study. Cardiac Arrest Study Hamburg. Canadian Implantable Deﬁbrillator Study. Eur Heart J 21:2071-2078, 2000 27. Roy D, Marchand E, Theroux P, et al: Programmed
ventricular stimulation in survivors of an acute myocardial infarction. Circulation 72:487-494, 1985 Gomes JA, Winters SL, Stewart D, et al: A new noninvasive index to predict sustained ventricular tachycardia and sudden death in the ﬁrst year after myocardial infarction: Based on signal-averaged electrocardiogram, radionuclide ejection fraction and Holter monitoring. J Am Coll Cardiol 10:349357, 1987 La Rovere MT, Pinna GD, Hohnloser SH, et al: Baroreﬂex sensitivity and heart rate variability in the identiﬁcation of patients at risk for life-threatening arrhythmias: Implications for clinical trials. Circulation 103:2072-2077, 2001 Kleiger RE, Miller JP, Bigger JT, et al: Decreased heart rate variability and its association with increased mortality after acute myocardial infarction. Am J Cardiol 59:256-262, 1987 Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology: Heart rate variability: Standards of measurement, physiological interpretation and clinical use. Circulation 93:1043-1065, 1996 Buxton AE, Marchlinski FE, Doherty JU, et al: Repetitive, monomorphic ventricular tachycardia: Clinical and electrophysiologic characteristics in patients with and patients without organic heart disease. Am J Cardiol 54:997-1002, 1984 Bigger JT, Fleiss JL, Kleiger R, et al: The relationships among ventricular arrhythmias, left ventricular dysfunction, and mortality in the 2 years after myocardial infarction. Circulation 69:250-258, 1984 Schlapfer J, Kappenberger L, Fromer M. What risk should justify implantable cardioverter deﬁbrillator therapy? Am J Cardiol 83:101D-103D, 1999 Myerburg RJ, Mitrani R, Interian A, et al: The interpretation and clinical application of data from trials on sudden cardiac death. J Interv Cardiol Electrophysiol 4:95-102, 2000 (suppl 1) Myerburg RJ, Mitrani R, Interian AJ, et al: Interpretation of outcomes of antiarrhythmic clinical trials: Design features and population impact. Circulation 97:1514-1521, 1998 Moss AJ, Hall WJ, Cannom DS, et al: Improved survival with an implanted deﬁbrillator in patients with coronary disease at high risk for ventricular arrhythmia. Multicenter Automatic Deﬁbrillator Implantation Trial Investigators. N Engl J Med 335: 1933-1940, 1996 Bigger JT: Prophylactic use of implanted cardiac deﬁbrillators in patients at high risk for ventricular arrhythmias after coronary-artery bypass graft surgery. Coronary Artery Bypass Graft (CABG) Patch Trial Investigators. N Engl J Med 337:1569-1575, 1997 Buxton AE, Lee KL, Fisher JD, et al: A randomized study of the prevention of sudden death in patients with coronary artery disease. Multicenter Unsustained Tachycardia Trial Investigators. N Engl J Med 341:1882-1890, 1999 Moss AJ: The multicenter automatic deﬁbrillator im-
plantation trial (MADIT), in Woosley RL, Singh SN (eds): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial Evidence. New York, NY, Marcel Dekker, 2000, pp 155-163 Prystowsky EN: MADIT: Critique, in Woosley RL, Singh SN (eds): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial Evidence. New York, Marcel Dekker, 2000, pp 165-169 Higgins SL: Impact of the Multicenter Automatic Deﬁbrillator Implantation Trial on implantable cardioverter deﬁbrillator indication trends. Am J Cardiol 83:79D-82D, 1999 Bigger JT, Bloomﬁeld DM: The CABG Patch trial: Prophylactic use of implanted cardiac deﬁbrillators in high-risk CABG surgery patients, in Woosley RL, Singh SN (eds): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial Evidence. New York, NY, Marcel Dekker, 2000, pp 191-216. Block M, Breithardt G: The implantable cardioverter deﬁbrillator and primary prevention of sudden death: The Multicenter Automatic Deﬁbrillator Implantation Trial and the Coronary Artery Bypass Graft (CABG)Patch Trial. Am J Cardiol 83:74D-78D, 1999 Zimetbaum PJ, Josephson ME: CABG Patch: Critique, in Woosley RL, Singh SN (eds): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial evidence. New York, NY, Marcel Dekker, 2000, pp 217-223 Bigger JT, Rottman JN, Whang W, et al: CABG surgery unlinked the arrhythmic substrate from arrhythmic outcomes in the CABG Patch trial. Circulation 100:I-366-I-367, 1999 (abstr) Strickberger SA: Multicenter randomized trial comparing amiodarone to implantable deﬁbrillator in patients with nonischemic cardiomyopathy and aymptomatic ventricular tachycardia: AMIOVIRT Trial. Circulation 102:2794, 2000 (abstr) Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure (MERIT-HF): Effect of metoprolol CR/XL in chronic heart failure. Lancet 353:2001-2007, 1999 CIBIS-II Investigators and Committees: The Cardiac Insufﬁciency Bisoprolol Study II (CIBIS-II): A randomised trial. Lancet 353:9-13, 1999 The Beta-Blocker Evaluation of Survival Trial Investigators: A trial of the beta-blocker bucindolol in patients with advanced chronic heart failure. N Engl J Med 344:1659-1667, 2001 Anderson JL, Hallstrom AP, Epstein AE, et al: Design and results of the antiarrhythmics vs implantable deﬁbrillators (AVID) registry. The AVID Investigators. Circulation 99:1692-1699, 1999 Fruhwald FM, Eber B, Schumacher M, et al: Syncope in dilated cardiomyopathy is a predictor of sudden cardiac death. Cardiology 87:177-180, 1996 Sager PT, Choudhary R, Leon C, et al: The longterm prognosis of patients with out-of-hospital cardiac arrest but no inducible ventricular tachycardia. Am Heart J 120:1334-1342, 1990 Fonarow GC, Feliciano Z, Boyle NG, et al: Improved survival in patients with nonischemic advanced
RAJ AND SHELDON
heart failure and syncope treated with an implantable cardioverter-deﬁbrillator. Am J Cardiol 85:981985, 2000 Knight BP, Goyal R, Pelosi F, et al: Outcome of patients with nonischemic dilated cardiomyopathy and unexplained syncope treated with an implantable deﬁbrillator. J Am Coll Cardiol 33:1964-1970, 1999 Gussak I, Bjerregaard P, Hammill SC: Clinical diagnosis and risk stratiﬁcation in patients with Brugada syndrome. J Am Coll Cardiol 37:1635-1638, 2001 Antzelevitch C. The Brugada syndrome: Diagnostic criteria and cellular mechanisms. Eur Heart J 22: 356-363, 2001 Fujiki A, Usui M, Nagasawa H, et al: ST segment elevation in the right precordial leads induced with class IC antiarrhythmic drugs: Insight into the mechanism of Brugada syndrome. J Cardiovasc Electrophysiol 10:214-218, 1999 Antzelevitch C, Yan GX. Cellular and ionic mechanisms responsible for the Brugada syndrome. J Electrocardiol 33:33-39, 2000 (suppl) Moss AJ, Zareba W, Hall WJ, et al: Effectiveness and limitations of beta-blocker therapy in congenital long-QT syndrome. Circulation 101:616-623, 2000 Maron BJ, Gardin JM, Flack JM, et al: HCM in the general population. Circulation 94:588-589, 1996 Maron BJ, Olivotto I, Spirito P, et al: Epidemiology of hypertrophic cardiomyopathy-related death: Revisited in a large non-referral-based patient population. Circulation 102:858-864, 2000 Spirito P, Bellone P, Harris KM, et al: Magnitude of left ventricular hypertrophy and risk of sudden death in hypertrophic cardiomyopathy. N Engl J Med 342: 1778-1785, 2000 Maron BJ, Shen WK, Link MS, et al: Efﬁcacy of implantable cardioverter-deﬁbrillators for the prevention of sudden death in patients with hypertrophic cardiomyopathy. N Engl J Med 342:365-373, 2000 McKenna WJ, Thiene G, Nava A, et al: Diagnosis of arrhythmogenic right ventricular dysplasia/cardiomyopathy. Task Force of the Working Group Myocardial and Pericardial Disease of the European Society of Cardiology and of the Scientiﬁc Council on Cardiomyopathies of the International Society and Federation of Cardiology. Br Heart J 71:215-218, 1994 Wichter T, Borggrefe M, Haverkamp W, et al: Efﬁcacy of antiarrhythmic drugs in patients with arrhythmogenic right ventricular disease. Results in patients with inducible and noninducible ventricular tachycardia. Circulation 86:29-37, 1992 Corrado D, Fontaine G, Marcus FI, et al: Arrhythmogenic right ventricular dysplasia/cardiomyopathy: Need for an international registry. European Society of Cardiology and the Scientiﬁc Council on Cardiomyopathies of the World Heart Federation. J Cardiovasc Electrophysiol 11:827-832, 2000 Moss AJ: Implantable cardioverter deﬁbrillator ther-
apy: The sickest patients beneﬁt the most. Circulation 101:1638-1640, 2000 Bardy GH: The sudden cardiac death-heart failure trial (SCD-HeFT), in Woosley RL, Singh SN (eds.): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial Evidence. New York, NY, Marcel Dekker, pp 323-342, 2000 Schron EB, Friedman LM, Greene HL: Future clinical trials, in Woosley RL, Singh SN (eds): Arrhythmia Treatment and Therapy: Evaluation of Clinical Trial Evidence. New York, NY, Marcel Dekker, pp 351360, 2000 Moss AJ, Cannom DS, Daubert JP, et al: Multicenter automatic deﬁbrillator implantation trial II (MADIT II): Design and clinical protocol. Ann Noninvasive Electrocardiol 4:83-91, 1999 Hohnloser SH, Connolly SJ, Kuck KH, et al: The deﬁbrillator in acute myocardial infarction trial (DINAMIT): Study protocol. Am Heart J 140:735-739, 2000 Raviele A, Bongiorni MG, Brignole M, et al: Which strategy is “best” after myocardial infarction? The Beta-blocker Strategy Plus Implantable Cardioverter Deﬁbrillator Trial: Rationale and study design. Am J Cardiol 83:104D-111D, 1999 O’Brien BJ, Connolly SJ, Goeree R, et al: Costeffectiveness of the implantable cardioverter-deﬁbrillator: Results from the Canadian Implantable Deﬁbrillator Study (CIDS). Circulation 103:1416-1421, 2001 Bardy GH, Yee R, Jung W: Multicenter experience with a pectoral unipolar implantable cardioverterdeﬁbrillator. Active Can Investigators. J Am Coll Cardiol 28:400-410, 1996 Middlekauff HR, Stevenson WG, Saxon LA: Prognosis after syncope: Impact of left ventricular function. Am Heart J 125:121-127, 1993 De Sutter J, Kazmierczak J, Fonteyne W, et al: Factors determining long-term outcomes and survival in patients with coronary artery disease and ventricular tachyarrhythmias: A single center experience. Pacing Clin Electrophysiol 23:1947-1952, 2000 Krahn A, Klein GJ, Yee R, et al: The effect of ejection fraction on the relative beneﬁt of the implantable deﬁbrillator in the Canadian Implantable Deﬁbrillator Study. Circulation 98:I-468, 1998 (abstr) Sheldon R, Connolly S, Krahn A, et al: Identiﬁcation of patients most likely to beneﬁt from implantable cardioverter-deﬁbrillator therapy: The Canadian Implantable Deﬁbrillator Study. Circulation 101:16601664, 2000 Exner DV, Sheldon RS, Pinski SL, et al: Do baseline characteristics accurately discriminate between patients likely versus unlikely to beneﬁt from implantable deﬁbrillator therapy? Evaluation of the Canadian Implantable Deﬁbrillator Study Implantable Cardioverter Deﬁbrillatory Efﬁcacy Score in the Antiarrhythmics versus Implantable Deﬁbrillators Trial. Am Heart J 141:99-104, 2001 Hallstrom AP, McAnulty JH, Wilkoff BL, et al: Pa-
tients at lower risk of arrhythmia recurrence: A subgroup in whom implantable deﬁbrillators may not offer beneﬁt. Antiarrhythmics Versus Implantable Deﬁbrillator (AVID) Trial Investigators. J Am Coll Cardiol 37:1093-1099, 2001 Mushlin AI, Hall WJ, Zwanziger J, et al: The costeffectiveness of automatic implantable cardiac deﬁbrillators: Results from MADIT. Multicenter Automatic Deﬁbrillator Implantation Trial. Circulation 97: 2129-2135, 1998 Larsen GC, McAnulty JH, Hallstrom A, et al: Hospitalization charges in the antiarrhythmics versus implantable deﬁbrillators (AVID) trial: The AVID economic analysis study. Circulation 96:I-77, 1997 (abstr) Peduzzi P, Kamina A, Detre K: Twenty-two-year follow-up in the VA Cooperative Study of Coronary Artery Bypass Surgery for Stable Angina. Am J Cardiol 81:1393-1399, 1998 The Veterans Affairs Coronary Artery Bypass Surgery Cooperative Study Group: Eighteen-year follow-up in the Veterans Affairs Cooperative Study of Coronary Artery Bypass Surgery for Stable Angina. Circulation 86:121-130, 1992 Paglione M, Selby JV, Uratsu CS, et al: Quality of life over time in patients with life-threatening ventricular arrhythmias. Circulation 100:I-570, 1999 (abstr) Steinberg JS, Vloka ME, Qing Y, et al: Relationship of baseline quality of life scores to long-term survival in the antiarrhythmics versus implantable deﬁbrillators (AVID) trial. Circulation 100:I-367, 1999 (abstr) Bourke JP, Turkington D, Thomas G, et al: Florid psychopathology in patients receiving shocks from implanted cardioverter-deﬁbrillators. Heart 78:581583, 1997 Sears SF, Conti JB, Curtis AB, et al: Affective distress and implantable cardioverter deﬁbrillators: Cases for psychological and behavioral interventions. Pacing Clin Electrophysiol 22:1831-1834, 1999 Namerow PB, Firth BR, Heywood GM, et al: Qualityof-life six months after CABG surgery in patients randomized to ICD versus no ICD therapy: Findings from the CABG Patch Trial. Pacing Clin Electrophysiol 22:1305-1313, 1999 Dunbar SB, Smith P, Zhang Z, et al: ICD activations and quality of life one year after implant. Pacing Clin Electrophysiol 24:579, 2001 (abstr) Steinberg JS, Paquette M, Dorian P, et al: Detailed analysis of quality of life responses by patients in AVID: Relationship to mortality. Pacing Clin Electrophysiol 23:568, 2000 (abstr) Zipes DP: Implantable cardioverter-deﬁbrillator: A Volkswagen or a Rolls Royce: How much will we pay to save a life? Circulation 103:1372-1374, 2001 Califf RM, Eisenstein EL: Critical concepts in costeffectiveness for cardiovascular specialists. Am Heart J 140:S143-S147, 2000 (suppl) O’Brien BJ: Chapter 3. Cost effectiveness of ICD therapy: A review of published evidence. Can J Cardiol 16:1307-1312, 2000
194 96. Goldman L, Gordon DJ, Rifkind BM, et al: Cost and health implications of cholesterol lowering. Circulation 85:1960-1968, 1992 97. Kuppermann M, Luce BR, McGovern B, et al: An analysis of the cost-effectiveness of the implantable deﬁbrillator. Circulation 81:91-100, 1990 98. Larsen GC, Manolis AS, Sonnenberg FA, et al: Costeffectiveness of the implantable cardioverter-deﬁbrillator: Effect of improved battery life and comparison with amiodarone therapy. J Am Coll Cardiol 19:1323-1334, 1992 99. Kupersmith J, Hogan A, Guerrero P, et al: Evaluating and improving the cost-effectiveness of the implantable cardioverter-deﬁbrillator. Am Heart J 130:507515, 1995 100. Owens DK, Sanders GD, Harris RA, et al: Costeffectiveness of implantable cardioverter deﬁbrillators relative to amiodarone for prevention of sudden cardiac death. Ann Intern Med 126:1-12, 1997 101. Weaver EF, Hauer RN, Schrijvers G, et al: Cost-effectiveness of implantable deﬁbrillator as ﬁrst-choice therapy versus electrophysiologically guided, tiered strategy in postinfarct sudden death survivors. A randomized study. Circulation 93:489-496, 1996
RAJ AND SHELDON 102. Sheldon R, O’Brien BJ, Blackhouse G, et al: Effect of clinical risk stratiﬁcation on cost-effectiveness of the implantable cardioverter-deﬁbrillator: The Canadian Implantable Deﬁbrillator Study. Circulation 104: 1622-1626, 2001 103. Simpson CS, Klein GJ, Hoffmaster B: Expensive medical technologies and “indication extrapolation”: The case of implantable cardioverter-deﬁbrillators. Am Heart J 140:419-422, 2000 104. Pathmanathan RK, Lau EW, Cooper J, et al: Potential impact of antiarrhythmic drugs versus implantable deﬁbrillators on the management of ventricular arrhythmias: The Midlands trial of empirical amiodarone versus electrophysiologically guided intervention and cardioverter implant registry data. Heart 80:68-70, 1998 105. Delacretaz E, Schlaepfer J, Metzger J, et al. Evidence rather than costs must guide use of the implantable cardioverter deﬁbrillator. Am J Cardiol 86:K52-K57, 2000 106. Prystowsky E. ICD replacement: A family-based medicine approach. Pacing Clin Electrophysiol 23: 2022-2023, 2000