DAVID L. KASERMAN [*] Quality of life has been measured in many different ways for patients with chronic medical conditions. What is unique about the approach used here is that it uses suicide rates as a relatively objective measure of quality of life within the population of dialysis patients. Using a Heckman selection model, we estimate the relative suicide rates across patients undergoing both hemodialysis and peritoneal dialysis. Our empirical results show that patients on hemodialysis have relatively lower suicide rates after controlling for other factors. Specifically our results indicate that 141 fewer suicides will occur for every 1,000 patients shifted from peritoneal to hemodialysis. Prior estimates of the higher costs of the latter modality yield an estimated expenditure of $42,043 per suicide avoided. (JEL 118, L84, 131) I. INTRODUCTION Cost considerations have always played an important role in policy debates in the medical industry, and that role promises to grow as funding agencies' budgets are subjected to increasing scrutiny. Costs alone, however, cannot be the sole determining factor in resource allocation decisions in this sector of the economy. Rational choice requires that other pertinent factors be weighed in the analysis of alternative uses of limited funds. Among these other factors, quality of life considerations stand out as one of the more important components of optimal funding decisions. [1] As with any other index of an individual's level of utility, the concept of quality of life is highly subjective. Despite such subjectivity, in most product markets (e.g., cars, houses, and clothes) willingness to pay metrics are estimable from observed purchase behavior. Objective measurement of willingness to pay, however, is particularly problematic in many health care markets because, in many cases, patients do not pay directly (or often even indirectly) for the services they receive. Most of the cost of treatment is paid by third parties. This feature, of course, is also true for many goods that are publicly provided, like recreation in national forests and use of publicly provided highways. In the case of recreation and highways, however, proxies can be used to approximate willingness to pay. To be sure, such proxies are imperfect, but they provide some relevant information. [2] No such proxies appear to be available, however, for specific medical treatments. [3] Furthermore, any attempt to solicit direct revelations of willingness to pay is subject to the well-known problem of strategic misrepresentation of preferences. [4] Here, we propose the use of relative suicide rates as an objective (though extreme) indicator of the quality of life of medical patients. [5] To demonstrate this proposed use, the authors focus on dialysis patients. This particular application is of some interest in its own right. Dialysis patients exhibit a much higher (100 times) suicide rate than the general population. [6] As a result, the subject of suicides among this population is an important area of inquiry in itself. Moreover, the two alternative treatment modalities available to these patients--hemodialysis and peritoneal dialysis--exhibit substantially different treatment regimens that are believed to affect patients' quality of life differently. These alternative modalities also exhibit markedly different costs. Due to these cost differences, considerable savings may be obtained by shifting patients from the more expensive to the less expensive form of treatment. Such shifting, however, may not be worth the resulting cost reduction if the affected patients suffer a significant deterioration in the quality of life achieved. Therefore, the impact, if any, of treatment modality on the quality of life of dialysis patients is an important public policy consideration. Moreover, the basic approach of using observed suicide rates as an objective indicator of quality of life may be applicable in a variety of other policy decisions in which this factor is an issue. …