Although the implantable cardioverter defibrillator (ICD) has been in clinical use for nearly 20 years, the appropriate indications for this device—and how we arrive at those indications—remain controversial. The controversy has nothing to do with how well the device works, since the ICD has long been known to be extremely effective in its goal: to automatically terminate life-threatening ventricular arrhythmias.
Instead, the controversy stems from three general factors: the relative expense and inconvenience of using the ICD as opposed to other available therapies; confusion as to the most appropriate definition of "efficacy" for this device; and until relatively recently, the widespread notion that insufficient data existed to document the efficacy of the ICD.
This chapter reviews the present and emerging indications for the ICD. Before discussing the indications, however, we must first address the three general factors that render these indications controversial.
WHY INDICATIONS FOR THE ICD HAVE BEEN CONTROVERSIAL The Expense and Inconvenience of ICDs
There would be little controversy over its clinical indications if the ICD were relatively cheap and easy to use. The ICD, however, is very expensive. The hardware itself costs between $15,000 and $30,000 (US), and the total cost of implanting the defibrillator system often exceeds $40,000.
From: Contemporary Cardiology: Management of Cardiac Arrhythmias Edited by: L. I. Ganz © Humana Press Inc., Totowa, NJ
Furthermore, despite improvements in implantation techniques over the past 15 years, neither the implantation of the device nor the long-term management of its recipients are simple endeavors. Device selection, lead placement, intra-operative and postoperative testing, optimal programming of the device, follow-up, and troubleshooting all remain very complex (much more than for permanent pacemakers, for instance), and yet critically important tasks. Physicians who have not dedicated themselves to the treatment of complex arrhythmias and the use of antitachycardia devices are unlikely to have favorable results in managing patients with ICDs.
Both the expense and the continued complexity of the ICD create a persistent inclination among policymakers and physicians alike to seek alternative therapies, even when a suitable alternative may not exist. Thus, it has been relatively difficult to establish and expand the formal indications for use of this device.
Confusion as to the Most Appropriate Definition of "Efficacy"
Progress in establishing firm indications for the ICD has been hindered by a disagreement among electrophysiologists as to what constitutes "efficacy" for this device.
When the ICD first came into clinical use in the early 1980s, its efficacy seemed entirely obvious to most physicians. Both in laboratory testing and in the clinical arena, the defibrillator quickly proved itself to be extremely efficient at automatically recognizing and terminating ventricular tachyarrhythmias. Most clinicians who implanted defibrillators in more than a handful of patients rapidly saw firsthand examples of the life-saving capabilities of these devices. The ability of the ICD to prevent sudden death became immediately and dramatically obvious.
For early proponents of the ICD, it came as something of a surprise when, at the end of the 1980s, the efficacy of this device was challenged by serious and well-respected electrophysiologists (1-2).
The challenge went something like this: Although the ICD indeed prevents sudden death by terminating ventricular arrhythmias, that does not necessarily mean it also prolongs overall survival. Until well-designed studies show that the ICD significantly improves overall survival and prevents sudden death, its efficacy remains an open question.
The confusion created by this new definition of efficacy assured that, for much of the next decade, the indications for the ICD remained controversial.
There is no doubt that prolongation of overall survival is the outcome that must concern us the most. After all, it does little good to prevent sudden death if doing so fails to yield an overall survival benefit. Yet, although this is true, prolongation of overall survival nonetheless does not define the efficacy of the ICD.
The confusion in defining efficacy arises from mixing the concept of efficacy with that of outcome. The outcome one achieves with any treatment in medicine is different from the efficacy of that treatment. Penicillin is effective in treating infections with many gram-positive cocci. Yet, giving penicillin to a patient who has pseudomonas sepsis—despite the efficacy of the drug when it is used appropriately—will yield a poor outcome. This poor outcome does not mean the drug itself is inherently "ineffective"—it merely means that we have used it badly. We have achieved a poor outcome through our inappropriate use of this effective drug. The question when using penicillin is not whether the drug is effective in people with infections, but in which people with infections will this effective drug give us the outcome we desire?
Similarly, the ICD is designed to terminate ventricular arrhythmias and thus to prevent sudden death. It is demonstrably effective in doing this. The fact that we also want to prolong overall survival (i.e., the fact that we desire a particular outcome) simply means that we must pay attention to how we use it.
The relationship of the defibrillator's ability to prevent sudden death to its ability to prolong overall survival (i.e., the relationship between efficacy and outcome) can be stated as an axiom, the Axiom of Overall Survival (3): In a given population of patients followed over time t, the implantable cardioverter defibrillator will measurably prolong overall survival as long as the risk of sudden death from ventricular tachyar-rhythmias is sufficiently greater than the risk of death from all other causes combined.
In other words, the ability of the ICD to prolong overall survival depends, completely and solely, on the population of patients in which it is used. Despite the proven efficacy of the device, the outcome will be unfavorable if either of two conditions is not met: that is, if the risk of sudden death is not high, or the risk of death from other causes is not low.
Fig. 1 illustrates how the outcome with the ICD depends entirely on the risk of sudden death (dark circles) and the risk of death from all other causes (white circles) in the population receiving the device.
Population 1 is analogous to the population at large. In this population, over time t, the risk of both sudden death and death from all other causes is low. Implanting defibrillators in such a population will not have a measurable effect on overall survival— survival will be good with or without the defibrillator.
Patients in Population 2 carry a high risk of sudden death (over time t) but a relatively low risk of death from other causes. This sort of risk profile may be seen, for instance, in patients who present with ventricular fibrillation (VF) in the face of normal or near-normal cardiac function. In such patients, eliminating the risk of sudden death with the ICD would significantly reduce overall mortality and improve overall survival over time t.
In Population 3, the risk of sudden death is also quite substantial. However, the risk of death from other causes is even higher—and furthermore, the risks overlap, so that death is likely over time t even if sudden death can be prevented. Although an ICD may prevent numerous episodes of sudden death, overall survival may not be measurably improved. Patients with severe end-stage cardiac disease may fit into such a risk profile.
Population 4 represents the group of patients frequently referred to electrophysiolo-gists, either for evaluation of arrhythmia risk or management of documented arrhythmia. This population has varying degrees of risk and varying securities of underlying structural heart disease and comorbid conditions. The effect of the defibrillator on overall survival depends specifically on which patients are selected for implantation in this heterogeneous population.
These examples graphically illustrate how the ability of the ICD to prolong overall survival (the outcome one obtains with the device) is intrinsically dependent on the population in which the device is used, and not on the device itself.
Therefore, if a clinical trial fails to show a benefit in overall survival with the ICD this result simply indicates a failure to apply the defibrillator to a population of patients who meet the two criteria given by the Axiom of Overall Survival. Conversely, a clinical trial that does show an overall survival benefit does not "prove" the efficacy of the ICD itself. Instead, it merely indicates that a population of patients who meet
Fig. 1. The dark circles represent the risk of sudden death, and the white circles represent the risk of death from all other causes, over time t. Time t is represented by the area of the rectangle.
In Population 1, the risks of both sudden death and death from all other causes over time t are relatively small. In Population 1, the ICD would not measurably improve overall survival. The risk profile of Population 1 is representative of the population at large.
In Population 2, the risk of sudden death over time t is relatively large, and is substantially larger than the risk of death from all other causes. The ICD would significantly improve overall survival in this population over time t. Population 2 represents survivors of cardiac arrest whose electrical disease predominates over their structural heart disease.
In Population 3, the risk of sudden death is also high (similar to Population 2). But here, the risk of non-sudden death is even higher, and the two risks largely overlap. The area in which the circles overlap represents patients at high risk for both sudden death and non-sudden death. While implantation of a defibrillator would undoubtedly prevent many instances of sudden death in Population 3, overall survival for the group would not be measurably improved. Population 3 represents many patients with end-stage cardiac disease.
Although Populations 1-3 represent idealized subsets of patients, Population 4 is representative of the heterogeneous group of patients commonly referred to electrophysiologists for documented or suspected ventricular tachyarrhythmias. These patients have varying degrees of arrhythmias (and thus a spectrum of risks for sudden death), and also varying degrees of underlying cardiac and other organic disease (and thus a spectrum of risks for non-sudden death). The effect of the ICD on overall survival in Population 4 would depend on which patients in this group were "selected" to receive the device.
the criteria given by the Axiom has been successfully identified. In other words, clinical trials with the ICD are not actually measuring the efficacy of the device at all (since that efficacy was established long ago). Instead, they are simply measuring the efficiency of patient selection—the clinical outcome achieved when the ICD is used in selected populations of patients.
Unfortunately, because of the confusion over the definition of efficacy, electrophysiol-ogists spent much of the 1990s trying to "prove" the efficacy of the defibrillator, instead of working to identify subsets of patients who would benefit from this remarkably effective device. Progress in firmly establishing indications for the ICD has thus been delayed.
Perceived Inadequacy of Data Documenting the Efficacy of the ICD
To those who entered the 1990s assuming the efficacy of the ICD had been firmly established by its unquestioned ability to prevent sudden death, the indications for using this device seemed relatively clear. Namely, (and in accordance with the Axiom), these individuals sought to offer the defibrillator to patients whose risk for arrhythmic sudden death was substantially increased, and whose risk for early death from other causes appeared low. For these clinicians, deciding whether an ICD was likely to be beneficial was simply a matter of estimating the relative risks for sudden death and death from other causes.
These clinicians were quite taken aback when several very prominent electrophysiolo-gists—well-known experts in the field—demurred on this point of view, publicly stating that there was no useful data demonstrating the efficacy of the ICD and furthermore, the only way to get such data would be to conduct randomized clinical trials. Only when randomized trials demonstrated that the defibrillator prolonged the overall survival of its recipients could the device be said to be effective (1-2).
We have already addressed the point that defining "efficacy" of the defibrillator in this way is incorrect. Now added to this mistaken premise was the assertion that to demonstrate such efficacy, only data from randomized clinical trials could be considered.
This assertion, since it claimed the scientific high ground, carried the day. Accordingly, randomized trials with the ICD were organized, and during the 1990s considerable efforts were devoted to conducting these trials with the full support of the government, industry, and professional organizations.
Some of these trials have added materially to our ability to use the ICD effectively. Others (in the author's opinion) served only to confuse clinicians, for much of the 1990s, in regard to acceptable indications for the ICD.
It is important to consider how randomized trials with the ICD may not always be useful. Consider a clinical trial in which all patients who have survived a cardiac arrest are randomized to receive either the ICD or drug therapy. Such an indiscriminately selected group of subjects is likely to include individuals from each of Populations 1, 2, and 3 (Fig. 1). Taken as a group, the risk profile of this composite population will probably resemble Population 4 (Fig. 1). These patients will have varying degrees of arrhythmic risk, and varying degrees of underlying cardiac disease (and thus varying risks of death from nonarrhythmic causes). Taken as a group, there will be a relatively high risk of sudden death, a relatively high risk of death from all other causes, and a relatively high overlap between the risks.
Will the ICD significantly prolong survival in such a clinical trial? The answer depends on the precise mixture of patients from the heterogeneous population eligible for this trial that is finally selected for enrollment. If enrolled patients include a fairly high proportion of individuals from Population 2, overall survival will be prolonged for the group. But if enrolled patients include a fairly high proportion of individuals from Population 3, overall survival may not be prolonged. A positive outcome will depend solely on whether a sufficient proportion of patients from Population 2 are enrolled by design, by chance, or by covert preselection.
Depending on the variables of study design and of patient selection, the results easily come out either way, and either way, the study will tell us nothing about the efficacy of the ICD. Worse, since what this trial would tell us is merely the effect of using the ICD indiscriminately, its outcome would be of little practical use.
Since randomized trials with the ICD only give us outcome data—they merely tell us how well we are using this highly effective tool—to be useful such trials should be designed to measure the efficacy of the defibrillator on reasonably well-defined and
Secondary Prevention Trials With the ICD
Secondary Prevention Trials With the ICD
Was this article helpful?