Do arrhythmia patients improve survival by participating in randomized clinical trials?

Abstract

It is debatable whether patients benefit directly from participation in a randomized clinical trial. We attempt to address this question for participants in the Cardiac Arrhythmia Suppression Trial (CAST) and the Antiarrhythmics Versus Implantable Defibrillators (AVID) studies. Survival rates were compared between eligible patients who enrolled in the trials and eligible patients who did not enroll, adjusting for baseline covariates. In CAST, despite that the active therapy was found to confer an almost threefold increased risk of death, survival was similar between the 3163 enrolled and the 1363 nonenrolled eligible patients. However, when patients were under study management, their risk of death was approximately 20% lower than when they left study management. In AVID, overall survival was similar between the 1016 enrolled and the 1246 nonenrolled eligible patients. However, mortality was substantially higher among patients not enrolled because the referring physician mandated the type of therapy. Overall these observational analyses suggest a net improvement in survival for the participants in these two trials.

Introduction

If the relative benefits of alternative therapeutic strategies can best be evaluated through the use of the randomized clinical trial paradigm, then it follows that patients as a group benefit from the conduct of randomized clinical trials. This does not necessarily imply, however, that the patient participants in randomized trials benefit from that participation (http://hiru.mcmaster.ca/ebm/trout/bibliography.htm). Among the general population, randomization is poorly understood [1] and may often be viewed with suspicion. This suspicion comes particularly from those groups that have been abused by research in the past or are aware of research gone awry, such as the early experiments conducted on inmates of prisons and mental institutions [2], [3], and, more recently, examples of misconduct on the part of researchers, such as the data manipulation by one researcher in a breast cancer trial [4], the genetic therapy trial problems [5], and the death of a healthy volunteer in an asthma study [6].

Consequently, some may take the view that trial participants are essentially making a sacrifice (or being sacrificed) for a societal good. However, many individuals involved in the actual conduct of randomized clinical trials tend to believe that participants benefit. Objective reasons why this might be so include that randomized clinical trials are generally conducted under the direction and guidance of recognized experts; patients tend to receive a very thorough medical evaluation, frequently including state-of-the-art diagnostics not routinely available; medications are carefully monitored; and frequent follow-ups are systematic and thorough. Beyond this, when a trial is ended and the participants are surveyed about their participation, they tend to report substantial satisfaction, citing in particular the access to quality care that they received [7], [8], [9], [10], [11], [12], [13]. These responses tend to be the same, no matter which treatment arm, whether active, placebo, or other control, the patient was randomized to, suggesting a placebo effect of trial participation per se. However, it must be admitted that patients who became severely incapacitated or died during the course of the trial are unable to provide input to such surveys.

One review of over 21 trials concluded [14]:

While the evidence is not conclusive, it is more likely that clinical trials have a positive rather than a negative effect on the outcome of patients. In the limited data available, the effect seems to be larger in trials where an effective treatment already exists and is included in the trial protocol.

To truly test whether participation in a trial was beneficial or not, it would be necessary to do a trial (the trial of substance) within a trial (the trial of participation). First, patients would have to agree to be randomized to be or not to be solicited to participate in the trial of substance, and furthermore, to agree to allow the primary endpoint to be ascertained through some noninteractive monitoring, e.g., if the endpoint was death this could be accomplished by use of the National Death Index Service (NDIS) [15]. Those patients who were randomized to be solicited for the substantive trial would then be approached for consent to be randomized into the substantive trial. At the end, the mortality in all patients randomized to be approached (including those patients who didn't agree to participate in the substantive trial) would be compared with those patients who were randomized not to be approached. Theoretically, if trial participation was beneficial, the overall benefit in those randomized to be approached should be higher than in those randomized not to be approached, and conversely if the trial participation was actually harmful. We do not know whether such an inner-outer trial strategy has ever been used. There may be ethical concerns, but a compelling obstacle is the fact that the number of patients needed to conduct such a strategy is at least twice as large as the number needed to conduct the substantive trial, thus making the trial both more difficult and costly to conduct. In the absence of such a strategy, the fallback position seems to be to compare the patients approached who refused to participate with the patients who agreed to participate. This still requires a noninvasive follow-up procedure, but also necessitates the collection of substantial baseline characteristics, since nonparticipants may differ from participants on a number of factors [16], [17]. Thus, any comparison of benefit needs to adjust for known risk factors. Such an approach is sometimes, though far too seldom, attempted by establishing a registry, including extensive baseline risk factor data, on all eligible patients and following these patients through, e.g., hospital record review, claims data, or the use of a service such as the NDIS. Even this level of study requires informed consent, but because of the absence of any randomization of treatments and the “noninvolvement” beyond the initial contact at baseline, most patients are willing to consent to this level of participation. This is the approach that we used in two studies of antiarrhythmic therapy: the Cardiac Arrhythmia Suppression Trial (CAST) [17], [18], [19], which recruited into a registry approximately 15,000 patients post-myocardial infarction (MI) with evidence of ventricular arrhythmia, and the Antiarrhythmics Versus Implantable Defibrillators Trial (AVID) [17], [20], which recruited into a registry approximately 5000 patients with potentially lethal ventricular arrhythmia. In CAST, the active therapy was actually shown to be substantially harmful when compared to placebo therapy. In AVID, the new active therapy was shown to be substantially better than the active control therapy. Thus, in CAST one might have expected the participants to do less well than nonparticipants, since half of the participants were put on what turned out to be one of several very harmful therapies, which were either not yet approved by the Food and Drug Administration or approved for limited indications. Similarly, in AVID one might have expected participants to do worse than nonparticipants, since half of the participants were required to take the control therapy, while presumably most nonparticipants would receive the more recent and subsequently proved better therapy.