Cost-Effectiveness of Implantable Defibrillators

ABSTRACT & COMMENTARY

Synopsis: ICD implantation is cost-effective in patients who meet MADIT entry criteria.

Source: Mushlin AI, et al. Circulation 1998;97: 2129-2135.

Due to the expense of implantable cardioverter difibrillators (ICDs), a cost-effectiveness analysis of the prophylactic use of ICDs in patients enrolled in the Multicenter Automatic Implantable Cardioverter Defibrillator Implantation Trial (MADIT) was performed. Patients were eligible for enrollment in MADIT if they had documented nonsustained VT, a left ventricular ejection fraction less than 35%, and sustained VT induced at electrophysiologic study. Patients with induced VT were randomized to either immediate ICD implantation or "conventional" antiarrhythmic therapy that was not specified by the protocol. At the start of the trial, transthoracic ICD implants were used with transvenous lead systems introduced later during the course of the study. The study was stopped in 1996 after enrollment of 196 patients when improved survival was noted in the ICD group. Over a four-year period, the average survival in the ICD group was 3.66 years compared to 2.80 years in the control group.

For this study, Mushlin and associates collected data on health care service use on all of the patients in the trial. These data included hospitalizations, emergency room and office visits, diagnostic tests and procedures, community services, medical supplies, and prescription medications. Use data were collected and costs assigned based on standard Medicare and Medicaid rates and wholesale drug and equipment prices. Cost data were compared for specified time periods to eliminate effects of increased survival in the ICD group. Analysis was performed using an intention-to-treat principle. Five patients randomized to an ICD did not receive one, and 11 patients in the control group eventually had an ICD implanted.

As expected, the average costs for the initial hospitalizations in the ICD group were considerably higher due to costs for the device and the implantation procedure ($44,600 vs $18,900). Subsequently, the control group patients had somewhat higher costs due to crossovers to an ICD and medication costs. The survival benefit attributed to ICD use was 0.8 years out of four, and this was associated with an incremental cost of $21,580. This translates to an incremental cost-effectiveness rate of $27,000 per life-year saved.

Mushlin et al also did a sensitivity analysis to look at the general applicability of their data to current practice. Substantially better cost-effectiveness would be seen with transvenous device implantation. Four patients in the control group eventually either began chronic hemodialysis or underwent cardiac transplantation, expensive outcomes that would presumably not be prevented by an ICD. If data from these four patients were eliminated, the incremental cost-effectiveness rate rose to $39,600 per life-year saved. Mushlin et al conclude that ICD implantation is cost-effective in patients who meet MADIT entry criteria.

COMMENT BY JOHN P. DiMARCO, MD, PhD

Cost-effectiveness is an important issue for all practicing physicians. The ICD is an example of an expensive therapy that has been shown to be more effective for prolonging survival than drug therapy in several populations of patients at risk for arrhythmic death. If one confined ICD use to the traditional indication of survivors of cardiac arrest or hypotensive VT, the relatively low resuscitation rates for these events would restrict the economic effects of ICD therapy on society. However, prophylactic ICD use in high-risk populations before a clinical episode of sustained arrhythmia has a much greater potential for increasing total medical costs. In the analysis presented by Mushlin et al, the incremental cost ratio of $27,000 per life-year saved is within the range of medical procedures previously deemed to be acceptable by society.

There are some significant limitations to the approach used by Mushlin et al that must be considered. They collected cost data only after patients were enrolled in the trial. In order for patients to be eligible, each needed a measurement of ejection fraction, ECG documentation of nonsustained VT, and an electrophysiologic study demonstrating inducible sustained VT. The costs of these procedures are not included in this analysis, but screening populations who may have such findings would be expensive. Many patients thought to be at risk for sudden death will have some measurement of ejection fraction available, but outpatient monitoring to look for nonsustained VT and electrophysiologic studies are not routine in most patients. Should Holters be performed periodically over time? If so, at what intervals? Should all patients with nonsustained VT then go on to electrophysiologic study? What are the costs of these procedures? These questions are difficult to address but should be considered in any discussion of the economic implications of the MADIT data. Focusing only on costs incurred after the positive findings at electrophysiologic study have been made does not tell the entire story.

MADIT had high mortality rates in the control group, and reductions in both cardiac and noncardiac mortality were seen in the ICD group. Due to difficulties in classifying deaths, total mortality is the appropriate end point for this type of study, but one might ask what the cost-effectiveness data would look like if only the decrease in cardiac mortality was considered in a sensitivity analysis.

Cost data on ICD therapy are continually changing as new devices with different features are introduced. In many patients, a single chamber ICD can now be implanted and tested on a totally outpatient basis. This significantly decreases costs. To date, most manufacturers have concentrated on decreased size and additional diagnostic and therapeutic functions in new devices. The companies also provide extensive support for implantations and follow-up. Costs for prophylactic therapy would be much lower if limited function devices with enhanced longevity that required limited manufacturer support were available. These devices could be priced much lower than standard full-function devices yet would still be appropriate for prophylactic therapy in most high-risk groups. New ideas are needed if ICD therapy is to become not only cost-effective but cost attractive.