As Dr. Achtelik points out, the 500 000 estimate in McClellan and Tunis's editorial (1) referred to the total number of Medicare beneficiaries who would become eligible under expanded criteria. Assuming a life expectancy of 9 years, he estimated that the annual incidence of newly eligible patients would be one ninth of 500 000 or 56 000. Because the Medicare population is older than the MADIT-II (Multicenter Automatic Defibrillator Implantation Trial-II) population, the assumed life expectancy of 9 years might be rather generous. If the actual life expectancy were only 5 years, then the corresponding annual incidence would be 100 000 among Medicare beneficiaries.

In the first several years of expanded eligibility, many of the prevalent cases might opt for an ICD. The economic burden of such a bolus would still be very large. If half of eligible patients opted for an ICD in each of the first 2 years, in addition to 56 000 to 100 000 incident cases, the cost in each of the first 2 years could approach $12 billion, an enormous burden for our medical economy. Once the initial bolus was past, the annual cost would be a mere $1.7 billion to $3.5 billion a year, a burden still big enough to require some reasonable selectivity.

Patients enrolled in the MADIT-II trial had a mean age of 65 years, so only half of them would likely be included in the estimate. An equal or greater number of patients younger than 65 years of age will likely seek—and perhaps expect their insurers (or, in reality, society) to provide—an ICD. Enrollees in the SCDHeFT trial had a mean age of 60 years, suggesting that an even higher proportion of patients would fall beyond the Medicare estimate. We believe that Dr. Achtelik's estimates of an annual burden of only $1.7 billion may be overly conservative. The issue remains controversial (2). Even in the United States and Germany, resources available for medical care are limited and perhaps are becoming more so. Unrestrained use of expensive technologies may also displace other uses for our health care dollar. Many "low-tech" interventions can save life-years at far more modest costs. Facing such tradeoffs, we believe that insurers and health care systems will need to turn to formal cost-effectiveness analyses more frequently. Such studies are useful not only for determining whether we can afford the benefits, but also for identifying which technologies should be applied in which populations.