The Milton Friedman Institute for Research in Economics

Medical care at the end of life, estimated to contribute up to a quarter of US health care spending, often encounters skepticism from payers and policy makers who question its high cost and often minimal health benefits. It seems generally agreed upon that medical resources are being wasted on excessive care for end-of-life treatments that often only prolong minimally an already frail life. However, though many observers have claimed that such spending is often irrational and wasteful, little explicit and systematic analysis exists on the incentives that determine end of life health care spending. There exists no positive theory that attempts to explain the high degree of end-of life spending and why differences across individuals, populations, or time occur in such spending. This paper attempts to provide the first rational and systematic analysis of the incentives behind end of life care. The main argument we make is that existing theoretical and empirical analysis of the value of life do not apply, and often under-values, the value of life near its end and terminal care. We argue that several factors drive up the value of life near its end including the low opportunity cost of medical spending near ones death, the value of hope including living into new innovations, and potential positive effect of on the value of life from being frail. We calibrate the ex-post value of hope associated with treatments for HIV patients to be as much as 4 times as high as standard per-capita estimates of treatment effects and as many as two and a half times as high as aggregate values across all cohorts. 1 We are thankful to Yang Lu and Eric Sun for valuable research assistance and for comments from Anupam Jena, Darius Lakdawalla, and Eric Sun as well as seminar participants at The University of Chicago, Washington University, the 2007 International Health Economics Association (IHEA) Meetings in Copenhagen, and RAND. Financial support from National Institutes of Aging (Grant P30 AG 12857) is gratefully acknowledged. Section 1: Introduction Medical care at the end of life often encounters skepticism from payers and policy makers who question its high cost and often minimal health benefits. Indeed, many studies have found that about a quarter of overall life-time spending on medical care occurs in an individual’s last year of life, regardless of whether that care is privately or publicly financed (Hogan et al. 2000; Lubitz and Riley 1993). It seems generally agreed upon that medical resources are being wasted on excessive end-of-life treatments that often only minimally prolong an already frail life. This excessive care at the end of life affects the overall distribution of health care spending as it is highly skewed, driving up lifetime average spending levels many times. For example, it has estimated that close to half of the overall spending on old individuals in the US stems from the top 5 % of the spending distribution (Garber et al (1998)). From an economic standpoint, it may seem obvious that much of this extreme end-of-life spending is irrational since the value of a life year is often estimated to be in the range of $100 thousand, but overall spending to extend life a few months near death can be in the millions. It can be argued that this vast misallocation of resources, induced by excessive end of life health care, has important consequences for the overall economy since end-oflife care makes up a substantial share of overall healthcare spending which is approximately 16% of the economy. This alleged over-spending on terminal care also has important implications for the public programs that pay for much of this end of life care, (such as the US’s Medicare for the old and Medicaid for the poor), as well funds like Social Security, which must continue paying out over longer, but lower quality, lives. Even though many observers have claimed that such spending is often futile and wasteful, it persists and is growing in both the private and public sectors. This may indicate that there are some less understood benefits of end-of-life care, and presumably larger than the often discussed costs. Indeed, little explicit and systematic analysis exists on the incentives that determine end-of-life care. We argue that a positive analysis of why and when high levels of terminal care spending occur is a prerequisite before any normative claims can be made or any policy proposals aimed at limiting such care can be justified on an efficiency basis. In this paper, we attempt to provide a rational choice analysis of the incentives behind end of life care. The main argument we make is that existing theoretical and empirical analyses of the value of a life do not apply to the valuation of life near its end and, therefore, do not apply to the demand for terminal care. In particular, that several forces operate which makes the value of end of life care higher than previously argued. First, if resources have no value when dead, a self-interested individual would be willing to forego his entire wealth to extend his life when dying, even if only for a few months. A substantial amount of spending on futile care is rational when there is little-to-no value of leaving wealth behind. The desire to spend one’s wealth on terminal care is supported by existing evidence that a large share, up to a half, of personal bankruptcies are associated with unforeseen health care spending, often taking place when faced with life-threatening diseases (Himmelstein (2005)). We argue that living, like other goods, has diminishing marginal utility--the willingness to pay for an additional year of life falls with how many years one has to live. This is in contrast to how the value of a statistical life-year is taught and explained: it is often prefaced with claiming that it is not how much people are willing to pay to avoid having a gun put to their head (presumably one’s wealth). However, terminal care decisions are often exactly of that nature. Second, we argue that an important ignored component of spending on end-of-life care is the preservation of hope of living which then raises valuation of life. We define the value of hope explicitly as the current consumption of future survival. If a patient is given a death sentence in 6 months, he values those 6 months less than if he knew he would live after that. The fear of knowing that the end is near is a bad, as is often revealed by people preferring a quick accidental death. We derive how this value of hope raises the willingness to pay for what appears as futile treatments. Related to such a value of hope is the option value of living to utilize a newly discovered treatment before one dies. Indeed, many celebrities, e.g. Michael J. Fox and the late Christopher Reeve, have invested large shares of their own wealth into speeding up the discovery of a cure for this purpose. Third, the social value of terminal care is often greater than the private value of the same treatments. However, existing analysis of the value of a life year mostly consider only private valuation. If the extension of a given person’s life has positive external effects on others (family members, tax-payers who do not tolerate letting poor people die), we would estimate and observe more spending than what is privately optimal. Since private willingness to pay for life extension is limited by one’s wealth, the mere existence of Medicaid seems inconsistent with a private valuation approach being relevant, as it would be infeasible for those patients to pay for the end-of-life care that they receive. Indeed, most rich countries don’t tolerate poor people dying when existing technologies can save them. Fourth, we argue that the rational level of terminal care for frail patients is often larger than commonly believed. In particular, we show when the value of terminal care may be the same regardless of the “quality” of life experienced by the patient. Therefore, even though a person may be frail and in very poor health, it may still be rational for him to value life-extending terminal care as much as a perfectly healthy person. This differs from a vast health economic literature arguing that there is less value in prolonging a life of lower quality, the driving assumption of “quality-adjusted life-year” (QALY) analysis. Because of all of these factors, the value of terminal care may exceed the levels currently attributed. To empirically assess the importance of one of these factors, the value of hope, we calibrate the option value of new innovation associated with terminal care for HIV patients in the 1990s. We find that the ex-post value of hope associated with treatments for HIV patients to be as much as four times larger than standard estimates. The option value is largest when treatments enabled patients before the breakthrough of HAART in 1996 to live to see that breakthrough. Although it is clear that the tremendous breakthroughs in HIV may be atypical of medical progress across the board, they still serve as a useful case to illustrate how one can quantify this value of hope more systematically. The paper is organized as follows. Section 2 discusses the non-linearity of the value of life. Section 3 discusses how the value of hope raises spending. Section 4 discusses how altruism within and across families affects terminal care. Section 5 discusses the impact of quality of life on rational terminal care. Section 6 considers the unusual aspects of R&D into new terminal care technologies. Section 7 provides our calibrations for HIV. Lastly, section 8 concludes. Section 2: Rational Terminal Care and the Non-Linearity of the Value of Life Consider the indirect utility function V(Y,S) of an individual with lifetime wealth Y and survival function S. For example, this indirect utility function may be the one resulting from a canonical consumption problem of the type