The Rule of Rescue

Social Science & Medicine 56 (2003) 2407–2419

The Rule of Rescue John McKie*, Jeff Richardson Centre for Health Program Evaluation, Monash University, West Heidelberg, Victoria 3081, Australia

Abstract Jonsen coined the term ‘‘Rule of Rescue’’(RR) to describe the imperative people feel to rescue identifiable individuals facing avoidable death. In this paper we attempt to draw a more detailed picture of the RR, identifying its conflict with cost-effectiveness analysis, the preference it entails for identifiable over statistical lives, the shock–horror response it elicits, the preference it entails for lifesaving over non-lifesaving measures, its extension to non-life-threatening conditions, and whether it is motivated by duty or sympathy. We also consider the measurement problems it raises, and argue that quantifying the RR would probably require a two-stage procedure. In the first stage the size of the individual utility gain from a health intervention would be assessed using a technique such as the Standard Gamble or the Time Trade-Off, and in the second the social benefits arising from the RR would be quantified employing the Person TradeOff. We also consider the normative status of the RR. We argue that it can be defended from a utilitarian point of view, on the ground that rescues increase well-being by reinforcing people’s belief that they live in a community that places great value upon life. However, utilitarianism has long been criticised for failing to take sufficient account of fairness, and the case is no different here: fairness requires that we do not discriminate between individuals on morally irrelevant grounds, whereas being ‘‘identifiable’’ does not seem to be a morally relevant ground for discrimination. r 2003 Elsevier Science Ltd. All rights reserved. Keywords: Rule of Rescue; Utilitarianism; Fairness; Cost-effectiveness analysis; QALY

Introduction Why do we mount expensive searches—for sailors lost at sea, for example—when there is little chance of finding those who are missing? (Creadon, 1997). If searching is expensive, and the chance of success is negligible, surely the money would be better spent in other ways? Why do we offer critically ill patients intensive care, when prognosis is terrible? (Osborne & Evans, 1994, p. 779). Surely these policies divert resources from other activities where the benefits would be greater? Why do some patients receive a second or third heart or liver transplant, when first-time recipients have a higher 1-year survival rate? (Ubel, Arnold, & Caplan, 1998, pp. 276–279). When organs are in short supply why not give priority to first-time transplants if they have a better chance of survival? These practices

*Corresponding author. E-mail address: [email protected] (J. McKie).

manifest a psychological imperative that is hard to resist: namely, the imperative to rescue identifiable individuals facing avoidable death, without giving too much thought to the opportunity cost of doing so. Jonsen dubbed this the ‘‘Rule of Rescue’’ (RR) (Jonsen, 1986, pp. 172–174). In this paper we address three main questions: (1) What is the nature of this imperative? (2) How might we go about measuring it? (3) Is it ethically defensible?

What is the RR? Opportunity costs In the health sector it has become common to use ‘‘Quality-Adjusted Life Years’’ or QALYs as the unit of effectiveness in cost-effectiveness analysis (CEA). In its simplest form the QALY represents a year of life that has been weighted, or discounted, by an index of the

0277-9536/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 0 2 ) 0 0 2 4 4 - 7

2408

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

quality of life. By convention, full health has a weighting of 1 and death has a weighting of 0. So, for example, if a year of life on hospital dialysis is considered to be worth only 57 per cent as much as a year of normal health (as reported by Torrance, 1987), then 20 years of life on dialysis would be equivalent to 20
0.57=11.4 QALYs. In conventional CEA only a change in the length and quality of life are important and QALYs combine these two dimensions. QALYs therefore provide a (conceptually) simple method for prioritising health care: all else equal, the lower the cost of a QALY the greater the value for money offered by a programme or treatment, and thus the higher a priority it should receive. However, the RR conflicts with this logic—the standard logic underlying CEA.1 Decisions influenced by the RR show a strong tendency to disregard costeffectiveness when the life of an identifiable individual is in danger. Allocative efficiency—maximizing healthrelated utility per unit cost—is simply not the only, or even the major, factor when an identifiable person’s life is visibly threatened. The RR does not, of course, always conflict with CEA. On a particular occasion the cost of saving an identifiable person from death (for example, by performing cardio-pulmonary resuscitation (CPR) or throwing a life preserver overboard) may be minimal, and the benefit (saving a life) so great that the course of action dictated by the RR is also the most cost effective. This being said, when the life of an identifiable individual is visibly threatened, the cost-effectiveness of the measures taken is normally not uppermost in people’s minds. If the measures taken are cost-effective they are so only per accidens. They are not taken because they are cost effective.2 According to Hadorn (1991a, b, 1996), the RR played an important role in the infamous Oregon prioritysetting exercise for allocating resources in the state’s Medicaid program. In 1990 the Oregon Health Services Commission produced a priority list of health services. It ranked some 1600 condition-treatment pairs, for example, ‘‘medical therapy for hypoglycemic coma’’, ‘‘septo1 The term ‘‘standard logic’’ refers here to the comparison of programme costs with programme benefits, whether benefits are measured by Quality-Adjusted Life Years, disability-adjusted life years (DALYs) (see Murray & Lopez, 1996), or healthyyear equivalents (HYEs) (see Mehrez & Gafni, 1989). 2 It is worth noting that there is legal backing for the RR. ‘‘The mining company which refuses to spend money on saving the trapped men may be liable not only civilly but also criminally for a wrongful omission to save one towards whom it stands in a special relation of duty’’ (Fried, 1969, p. 1418). More generally, we suspect that there would be little or no sympathy either in court or by the general public for a legal defence which argued that a corporation had allowed deaths to occur because the cost of lifesaving (to the corporation) somewhat exceeded the benefits of lifesaving (primarily to the victim and his/her family).

plasty/repair/control of hemorrhage for life-threatening epistaxis’’, ‘‘stabilization for open fracture of the ribs and sternum’’. Rankings were based on the expected outcomes with and without treatment, the duration of the treatment effect, the expected quality of life of patients assessed on a scale calibrated by Oregon residents, and the cost of the treatment. The initial list was never forwarded to the legislature because of the counterintuitive ordering that resulted, especially concerning lifesaving treatments. For example, dental caps for pulp or near pulp exposure were assigned a higher priority than surgical treatment for ectopic pregnancy (salpingectomy/salpingoophorectomy), and splints for temporomandibular joint disorder were ranked higher than appendectomies for appendicitis (Hadorn, 1991a; Dixon & Welch, 1991). A revised list was produced a year later that saw all treatment of life-threatening conditions placed in a separate high-priority category. Hadorn argued that this occurred because, ‘‘any plan to distribute health care services must take human nature into account if the plan is to be acceptable to society. In this regard there is a fact about the human psyche that will inevitably trump the utilitarian rationality that is implicit in cost-effectiveness analysis: people cannot stand idly by when an identified person’s life is visibly threatened if rescue measures are available’’ (Hadorn, 1991a, p. 2219). Perhaps the most conspicuous feature of the RR then is the tendency to disregard opportunity costs when the life of an identifiable individual is visibly threatened. There is a tendency to ‘‘act first and ask questions later’’. Considerations about costs are pushed into the background. Identifiable individuals Another noticeable feature of the RR is the focus on identifiable individuals.3 This also contrasts with the impersonal perspective of conventional CEA, which assumes that the distribution of QALYs among the population is unimportant. In particular, measures that are expected to produce the same number of QALYs are considered equally valuable irrespective of the context (for example, emergency or common place illness) or the personal characteristics of the potential beneficiaries, including whether they are ‘‘identifiable’’ or ‘‘merely statistical’’. These assumptions make the simple QALY algorithm egalitarian, in the sense that an improvement 3

As Jenni and Loewenstein point out, animals can evoke the same imperative to rescue. ‘‘Identifiable victims need not be human: in 1988 a multi-national effort spent millions to rescue three grey whales trapped under the Arctic ice cap, while at the same time the Japanese whaling industry was spending millions to locate and harvest whales’’ (Jenni & Loewenstein, 1997, p. 236).

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

in A’s health is considered to be equivalent to a similar improvement in B’s health regardless of intelligence, gender, social class, occupation, race, and so on. Programmes and services are ranked without regard to the way the QALYs are distributed among particular individuals or groups. As Torrance states: ‘‘the method is egalitarian within the health domain; that is, each individual’s health is counted equally’’ (Torrance, 1986, p. 17).4 But the RR is incompatible with this sort of egalitarianism: it implicitly gives added importance to identifiable as opposed to statistical lives—it violates the assumption of ‘‘distributive neutrality’’. A lifesaving intervention takes on added value—or people act as though a lifesaving intervention takes on added value— if it benefits this person, where the demonstrative ‘‘this’’ refers to an individual who is, to those in a position to assist, present and identifiable. It is worth noting that the special value attached to identifiable individuals serves to distinguish the RR from the preference for helping the worst off. The latter preference underlies Rawls’s difference principle: social and economic inequalities are to be arranged so that they are to the greatest benefit of the least advantaged members of society (Rawls, 1971, pp. 302–303). Callahan expressed the same sentiment, noting, in particular, the cost of this preference for helping the worst off: ‘‘Our bias, I contend, should be to give priority to persons whose suffering and inability to function in ordinary life is most pronounced, even if the available treatment for them is comparatively less efficacious than for other conditions’’ (Callahan, 1994, p. 463). Several empirical studies have also revealed that people are often prepared to sacrifice health gains in order to ensure that the more severely ill are given priority over the less severely ill (Nord, Richardson, & Macarounas-Kirchmann, 1993; Nord, 1993, 1996, 1999; Nord, Richardson, Street, Kuhse, & Singer, 1995a; Ubel, Scanlon, Loewenstein, & Kamlet, 1996; Ubel, 2000; Pinto, 1997; Nord, Pinto, Richardson, Menzel, & Ubel, 1999). However, this preference for helping the worst off reflects a different sentiment from the RR, since the latter uniquely applies to identifiable individuals in immediate peril, rather than to an entire category of anonymous 4

Williams notes that: ‘‘Conventionally, being dead is regarded as of zero value to everybody, and being healthy is regarded as being worth one to everybody [on a quality of life scale]. This convention has important ethical implications because it represents a strong and quite specific egalitarian position’’ (Williams, 1996, pp. 1796–1797). In contrast with orthodox economics individuals may therefore be compared because the endpoints of the quality of life scale are set equal for every individual. People are treated as if their minimum and maximum strengths of preference are equal. As this assumption is clearly wrong, QALY-based cost-effectiveness analysis, in effect, assumes that these intensities should be treated as if they are equal.

2409

individuals (those who are worst off). Note that Rawls argues that the difference principle would be chosen by a rational egoist placed behind an imaginary veil of ignorance explicitly to prevent him/her from treating anyone as an identifiable individual. By contrast, it is precisely because the individual is identifiable, and in dire circumstances, that there is the powerful urge to rescue. Shock Typically, when the RR is invoked to explain behaviour, one individual (or small group of individuals) is shocked by the desperate circumstances of another individual (or small group of individuals). Often situations evoking the RR response are unexpected and dramatic: they might involve miners trapped after an explosion, a child who has fallen down a well, or even a patient in cardiac arrest in the emergency ward. Such situations are likely to produce shock and horror in witnesses, or at the very least a sense of urgency. Witnesses often have no time to prepare themselves psychologically for events unfolding in real time. It is not surprising then that shock and horror are common accompaniments of the RR.5 Once again, this is at odds with conventional CEA, in which one lifesaving intervention is deemed more valuable than another only if it has a lower cost/QALY ratio. No value is attached to the prevention of shock and horror per se. Theoretically, these could be taken into account if they diminished the quality of life of others who might witness an unexpected death. In the same way, the health-related quality of life of others might be taken into account when a mentally handicapped child or dependent elderly person is moved out of the family and into institutional care. This being said, shock and horror could, at most, have a negligible effect upon benefits. QALYs are time based—a QALY is a life year weighted for quality of life. An hour or even a day in shock, while terrible for the person concerned, would have a negligible effect upon the total number of QALYs. Furthermore, it is clear from Hadorn’s initial description of the RR—particularly the low importance attached to cost effectiveness—that this concession concerning the ability of the QALY to take account of shock and horror misses the point. The RR expresses the view that we cannot ignore certain situations because of cost and low QALY benefits. Rather, we have a duty to give assistance irrespective of the cost effectiveness ratio. 5 Even events reported on television and in newspapers (such as the terrorist attack on the Twin Towers in New York in 2001) are often accompanied by graphic pictures and commentary. These serve to make victims identifiable to the general public, and are a powerful means of evoking the shock–horror reaction characteristic of the RR.

2410

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

Avoidable death

Duty or sympathy?

The RR attaches special value to interventions aimed at averting death, and this is also at odds with CEA. In conventional CEA measures that save lives are interchangeable with measures that improve the quality of life, so long as the potential QALY gains are the same. As a result, preventing a certain number of people from contracting a non-fatal disease will be equivalent to preventing a smaller number of people from contracting a fatal disease. The numbers that lead to an equivalence depend entirely upon the number of life years and the quality of life. The RR conflicts with this logic by giving added weight to lifesaving measures compared with nonlifesaving measures. We do not have to concede that the RR gives added weight to all lifesaving measures. Shock and horror occur more in contexts where life is threatened than in others, but it does not necessarily apply in all cases of death—for example, amongst the very elderly, or those who have had cancer for an extended period and we have come to accept the inevitability of death. The RR gives added weight to lifesaving measures compared with non-lifesaving measures because the RR arises in contexts where witnesses and bystanders are likely to be traumatised, and these are likely to be when life itself is unexpectedly threatened.

Imminent death

The RR might be motivated by a sense of moral obligation, or, alternatively, by sympathy, compassion, pity, and so on. Kant famously argued that only actions performed for the sake of duty have moral worth. He thought that feelings provide too flimsy a basis for moral behaviour. ‘‘If any action is to be morally good, it is not enough that it should conform to the moral law—it must also be done for the sake of the moral law: where this is not so, the conformity is only too contingent and precarious, since the non-moral ground at work will now and then produce actions which accord with the moral law, but very often actions which transgress it’’ (Kant, 1991, pp. 56–57). People should keep their promises, pay their debts, tell the truth, and so forth, not because they have certain feelings—because it is their ‘‘inclination’’—but out of respect for the moral law itself. Bochner et al. endorse this Kantian interpretation of the RR when they describe it as ‘‘a perceived duty to save endangered life where possible’’ (Bochner, Martin, Burgess, Somogyi, & Garry, 1994, p. 901). Likewise, Nord et al. describe it as ‘‘the sense of immediate duty that people feel towards those who present themselves to a health service with a serious condition’’ (Nord, Richardson, Street, Kuhse, & Singer, 1995b, p. 90). Dougherty describes the RR more ambiguously as ‘‘an ethical imperative to save individual lives even when money might be more efficiently spent to prevent deaths in the larger population’’ (Dougherty, 1993, p. 1359). If by ‘‘ethical imperative’’ Dougherty means ‘‘categorical imperative’’, we once again have a Kantian interpretation of the RR. However, it would be a mistake to downplay the role played by feelings in the RR response. Right-thinking people when confronted with a fellow human being in immediate, life-threatening danger will feel (as well as shock and horror) pity, sympathy, compassion, and be spurred to action accordingly. Nor can the moral worth of actions arising from such feelings be automatically impugned. Rousseau was explicit on this point (Rousseau, 1994, p. 48): ‘‘It is pity that sends us unreflecting to the aid of those we see suffering’’.6 Although Kant thought that only actions performed for the sake of duty have moral worth, there are situations

Moreover, in situations where the RR response is most likely to occur, death is usually imminent, and the risk of it occurring substantial, allowing for some vagueness in the terms ‘‘imminent’’ and ‘‘substantial’’. That is, measures taken now to avert the slight risk of an identifiable person dying in 20 years time would not normally be thought to fall under the RR. The RR response typically arises in a context of some urgency, where there is a high likelihood that death will occur almost immediately unless rescue measures are taken.

6 Mooney (1992, p. 1222) observes that while sympathy is something we undergo passively, it also contains an active component, motivating us to render assistance to those who are suffering or in danger. ‘‘To be seen as ‘merely sympathetic,’ that is, as feeling sympathy without in the least being moved toward benevolent action, is typically a rebuke, implying that one is deficient in sympathy itself. Whatever the depth of our empathic response, then, if we are sympathetic, we will also attend to the well-being of the other and be disposed toward helpful action.’’

Non-life-threatening conditions While the RR implicitly attaches special value to interventions offered in the face of death, and especially to interventions that will avert death, it can also be a factor when life is not endangered (Hadorn, 1991a, p. 2219). For example, children with physical deformities or disfigurements are sometimes flown from poor countries to wealthier countries for treatment. Their plight evokes the same ‘‘shock–horror’’ reaction—the same surge of compassion—as someone whose life is in immediate peril, and the same tendency to disregard, or underestimate, the value sacrificed by not putting resources to the best alternative use.

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

where it would seem entirely appropriate to ascribe moral worth to actions that arise from feelings of sympathy, pity, and so on. Using an example of Stocker’s (1976, p. 462), visiting a friend in hospital or consoling the bereaved would be valued less by most people if this behaviour was known to arise purely from a sense of duty, rather than being motivated also, and perhaps primarily, by feelings of compassion and sympathy. The description by Osborne and Evans of the RR, which makes no mention of duty, is more compatible with this non-Kantian interpretation. They characterise the RR as ‘‘the powerful human proclivity to rescue a single identified endangered life, regardless of cost, at the expense of many nameless faces who will therefore be denied health care’’ (Osborne & Evans, 1994, p. 779). Realistically, the rescue of an identifiable individual in immediate danger will arise neither purely from an abstract sense of duty, nor purely from empathetic emotions, but will be a mixture of both: ‘‘It is quite common for us to have more than one motive for what we do, and even more than one motive that by itself would be sufficient to produce a particular action. Kant never explicitly discusses overdetermined moral cases, where an action is done from the motive of duty and from some other non-moral motive’’. (Herman, 1981, pp. 359–360). So long as the ‘‘overdetermination of actions’’ is kept in mind—how behaviour can be motivated both by a sense of duty and by feelings of sympathy, pity, compassion, and so on—there is little danger in describing the RR as arising from a ‘‘sense of immediate duty’’, ‘‘a perceived duty’’, ‘‘an ethical imperative,’’ or a ‘‘powerful human proclivity’’. In summary then, the RR is characterised by a cluster of features. At its heart it describes the psychological imperative to rescue an identifiable individual (or small number of individuals) facing avoidable death (or other serious harm). This imperative is often driven by a ‘‘shock–horror’’ reaction: witnesses often have no time to prepare themselves for very disturbing events unfolding in real time. Whether the threat is to life itself, or merely to the well-being of the victim, there is typically a high probability of it occurring (if rescue measures are not undertaken), and it will occur in the short-term rather than the long-term. Most conspicuously, the RR is at odds with conventional CEA: there is little time, and little inclination, to reflect on the cost of the rescue measures undertaken when an identifiable individual is facing a real threat of avoidable death (or other serious harm).

Measuring the RR Since the maximisation of QALYs also maximises health-related utility, and since the RR is at odds with

2411

the straightforward maximisation of QALYs, the RR appears to violate the presumption that utility should be maximised. It is therefore an interesting question whether the RR can be justified from a utilitarian point of view. In this respect it is important to note the different possible sources of utility arising from a health intervention. In general terms there are four such sources: (1) The utility gained by the individual from the health state improvement (measured, for example, in QALYs). (2) The utility gained by the individual from the knowledge that an attempt to help has been made. (3) Society’s utility from the health state improvement. (4) Society’s utility from the knowledge that an attempt to help has been made. The utility under (1) and (3) derives from the health state improvement itself. Source (1) is concerned with health-related utility: it includes the individual’s ability to live a longer life and/or enjoy a better quality of life. Source (3) is a conventional externality, and may extend beyond health-related utility. For example, the individual who is restored to better health, or saved from death, generally brings happiness to others: he/she can return to work, look after dependents, and so on. By contrast, the utility under (2) and (4) derives from the knowledge that an attempt has been made to help. This source of utility is closely associated with the RR. Of course, the RR can produce the utility under (1) and (3) also: when rescues are successful a life is usually saved, and that person can contribute to society. But the utility under (2) and (4) is not recognised in conventional CEA, and the RR is a conspicuous source of this type of utility. The utility gained by the individual from source (2) is likely to be comparatively small, firstly because only one individual is affected, and secondly because the individual will derive his/her primary benefit from their improvement in health. But, since the effect described by (4) may be experienced by the entire population, it is potentially significant. It includes the enhanced feeling of security derived from knowing that one lives in a compassionate society, where those in desperate need are not ignored. The importance of this feeling is acknowledged by Scarre: ‘‘Without relations of mutual respect and the sense of belonging to a community which cares for the needs of each constituent member, people might muddle along in some sort of social nexus, but hardly very happily’’ (Scarre, 1996, p. 154). There is, at first glance, an apparent paradox involved in including the utility under (2) and (4). Since this utility arises from the attempt to help—from the intervention per se—and not from any health gain or from the external effects of a health gain, it is not derived from the public applying utilitarian principles. That is,

2412

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

maximising utility may require decision-makers to factor in utility derived from the application of non-utilitarian reasoning. But this is not a paradox at all. The fact that individuals obtain utility is an empirical matter and not a normative judgement. If individuals derive utility from applying non-utilitarian principles, either explicitly or implicitly, then that utility should be taken into account. Under (4) we include the utility derived by society from not knowing that rescue attempts are not being made. The double negative is intentional! If we do not hear about people who are being abandoned, we can assume we live in a society that does not abandon people, and this can be a source of utility.7 Doctors quietly going about their job are contributing significantly to social welfare over and above the utility gains they secure for their patients and their patient’s families. For measurement purposes, however, we must concentrate on what people know rather than what they do not know. There is simply too much that people do not know. If we can measure the utility people derive from knowing that an attempt to rescue has been made, or the disutility they derive from knowing that an attempt has not been made, we may infer the utility derived from not knowing that rescue attempts are not being made. Measuring the value of a health intervention, if it is to include the RR, is likely to require information from two sources. First, there must be an assessment of the utility gained by the individual and society from the health state improvement itself (the utility under (1) and (3)). Second, there must be an assessment of the utility gained by the individual and society from the knowledge that an attempt to help has been made (the utility under (2) and (4)). In practice, this would probably require a twostage procedure, involving different measuring techniques. For example, the Standard Gamble and Time TradeOff are well-suited to measuring individual utility (though each has its strengths and weaknesses (Richardson, 1994)). With the orthodox Standard Gamble (SG) the subject is offered two alternatives. Alternative 1 is a treatment with two possible outcomes. With probability p the patient is returned to normal health and lives for the remainder of his or her life; with probability (1  p) the patient dies immediately. Alternative 2 is the certainty of living in the health state being evaluated for the remainder of the subject’s life. Probability p is varied until the respondent is indifferent between the

7

In societies with corrupt or oppressive governments it might be more difficult to assume that people are not being abandoned simply because no one speaks or writes about such things. However, in societies where the government is basically trustworthy and there is a responsible and free press, and so on, it is reasonable to assume that people are not being abandoned if there are no credible stories to the contrary.

two alternatives. The probability at this point is taken to represent the utility index. This technique places a value on different health states in relation to death, which is conventionally put at 0, and normal health, which is conventionally put at 1. Using the SG then, it is reasonable to conclude that the value of saving one healthy person from death for 10 years is 10
1=10 QALYs, that the value of saving two healthy people from death for ten years is 10
2=20 QALYs, and so on. However, this ignores the utility derived by one person from the rescue of another. The figure ‘‘1’’ represents the utility of being healthy to each individual himself or herself. Thus, inferring that the utility involved in saving a healthy person from death is 10=1 leaves something important out, if the RR is quantitatively significant. The SG would seem to be illsuited to measure the RR. It measures individual utility, not social value.8 Similarly, with the orthodox Time Trade-Off (TTO) the subject is offered two alternatives—alternative 1 (the health state, S; to be evaluated) for time t (often the life expectancy of an individual with the chronic condition) followed by death; and alternative 2, which is healthy life for a lesser period of time x followed by death. Time x is varied until the respondent is indifferent between the two alternatives at which point the utility value for state S is given by U ¼ x=t: For some, this is the preferred technique as it most directly exposes the subject to the trade-off between life and quality of life (Mooney & Olsen, 1994; Richardson, 1994). Again, however, the conventional TTO does not capture the utility derived by one person from the rescue of another. At best it measures the value of a lifesaving intervention, or an improvement in quality of life, to the individual whose life is saved or improved. The social benefits arising from the RR are ignored. The person trade-off (PTO) is different in kind from the SG and TTO in this respect. With the PTO respondents are asked to compare two options each involving health improvement for a different number of people. The number of people in one option is varied until the two options are considered to be equally desirable. The value of the health improvements can then be inferred from the size of the two groups. For example, preventing the death (U ¼ 0) of ten people who will then remain in a poor health state A; may be valued as highly as saving the life of x people and restoring them to full health (U ¼ 1). In this case the value of saving the life of someone who will remain in health state A; relative to the value of saving the life of someone and restoring him or her to full health, is x=10: In this case 10 ðUðAÞ  0Þ ¼ xð1  0Þ or UðAÞ ¼ x=10:

8 On this distinction see, for example, Richardson and Nord (1997) and Dolan and Green (1998).

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

2413

Nord, in particular, has argued for the use of this technique (Nord, 1996, 1999). Importantly, the PTO does not encapsulate an individual’s preferences about their own lives (‘‘How great a risk to your own life would you be prepared to accept for an improvement in health?’’ ‘‘How much of your own life would you be prepared to sacrifice for an improvement in health?’’) Rather, it encapsulates an individual’s views about the lives of others. (‘‘Which of these two groups should be the higher priority for help?’’) Because it involves an impersonal trade-off between two groups, the PTO does allow the consideration of issues of justice and fairness. By contrast, when an individual is asked about the value of their own life, as with the SG and TTO, issues of fairness and justice do not arise. It is important to distinguish between these two perspectives—the ‘‘self-interest perspective’’ and the ‘‘caring-for-others perspective’’ (Nord, 1999, pp. 8–12)—because empirical studies show that individuals do not necessarily judge health improvements to others as the same as health improvements to themselves (Richardson & Nord, 1997; Dolan & Green, 1998).9 When it is applied to the actual improvement in which we have an interest, therefore, the PTO has the potential to capture the utility of the RR, if the context is described accurately, and subjects are able to imagine the plight of individuals in life-threatening situations. For example, by using scenarios that give varying amounts of personal detail, or in which the context is made increasingly more dramatic, it may be possible to quantify the preferences expressed in the RR response. Using the PTO in this way, people might systematically prefer to save the life of someone injured in a helicopter crash rather than someone dying in hospital from cancer, even if the number of QALYs gained is expected to be the same. People might systematically distinguish between different lifesaving interventions depending on the context. Theoretically, the PTO can capture the significance of the context, whereas the SG and TTO cannot. The factors that result in the RR are quintessentially ‘‘psychological’’ in the sense that they are affected not only by the objective circumstances of a person’s health state—their physical and mental suffering—but are shaped by the circumstances in which these occur. These contextual factors result in an intensification of the affect of the suffering upon others who are observing it either directly or indirectly through media reports. Consequently, the strength of the utility associated with

the RR depends upon circumstances that are extraneous to the immediate health state: factors such as the location or timing of the event, the perceived unfairness or unluckiness of the circumstances, and the empathy that the personal characteristics of the patient or victim can evoke. The effects of these types of factors are not well understood. It is certainly not possible, at present, to include them in a generic, decomposed health state instrument such as the EQ5D, QWB, HUI, 15D or AQoL.10 This implies that such generic instruments are, at best, useful for measuring ‘‘stage one’’ individual utility. Further, because these determinants of the RR are extrinsic to the health state, the stage-two measurement of social value must include a careful statement, not only of the health state, but also of the relevant elements of the context of the proposed intervention.

9 Nord’s distinction between the ‘‘self-interest perspective’’ and the ‘‘caring-for-others perspective’’ is similar to Dworkin’s distinction between ‘‘personal preferences’’ and ‘‘external preferences’’ (Dworkin, 1977), and Barry’s distinction between ‘‘privately-oriented’’ and ‘‘publicly-oriented’’ judgements and wants (Barry, 1965).

10 These instruments are described and compared in Richardson, Hawthorne, and Day (2001). 11 It should be noted that Fried does not accept this ‘‘personalist argument’’, and instead supports the preference for saving lives in present peril by distinguishing between different ways of dying (1969, pp. 1433–1437).

The normative status of the RR We turn now to the question of whether the RR is ethically defensible. We noted that underlying the RR is the desire to help an identifiable individual. This is contrary to conventional CEA, which does not distinguish between ‘‘identified’’ and ‘‘unidentified’’ individuals, but simply aims to maximise health (measured in QALYs). However, it does not automatically follow that the RR is morally indefensible. First, from a psychological point of view, the RR response is understandable. To abandon an identifiable person to death when they could be saved appears a heartless decision, and to symbolise a lack of concern for human life. Most people would find it hard to ‘‘stand idly by when an identified person’s life is visibly threatened if rescue measures are available’’ (as Hadorn put it), and would find it equally distressing to witness others’ failure to render assistance in this way. Often it is appropriate to deal with people as abstractions—as taxpayers, voters, purchasers, and so on. But as Fried observes (1969, p. 1430): ‘‘often too we encounter people as actual persons, and there it seems we have the occasion, the opportunity to show our deeper humanity’’.11 Furthermore, there is considerable social value in reinforcing acts driven by compassion and sympathy, and/or a sense of moral duty, even if the costeffectiveness of such acts is only an incidental consideration. It is almost certainly true that people obtain benefit from the belief that they are living in a caring and humane society, and that the observation of attempts to

2414

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

save life, whether heroic or more mundane, reinforces this. Eddy points out that one of the main criteria for applying the term Rule of Rescue to an intervention is ‘‘high symbolic value’’ Eddy (1994, p. 1795). Although preserving life at all cost is incompatible with CEA, the lesson that lifesaving rescues convey—that life is precious, and worth great effort to preserve—can be a source of long-term social utility.12 As Mooney observes: ‘‘even the detached bureaucratic utilitarian can plausibly calculate that encouraging sympathy among members of a community and between conflicting communities will make the world a better place’’ (Mooney, 1992, p. 1223). This is a defence of the symbolic value of sympathetic acts. It goes beyond the utility under (1) and (3)—the utility gained by the individual and society from the health state improvement itself—and points rather to the utility under (2) and (4)—the utility gained by the individual and society from the knowledge that an attempt to help has been made. MacLean makes the same point in different terms. He notes that most societies have rituals that express ‘‘sacred values’’. These sacred values in turn have a role in strengthening social integration. MacLean counts the value of life as one such value in most societies, including secular societies. Thus: ‘‘we need to guard against treating human life as exchangeable for other commodities. Inefficient or even merely symbolic actions are one such guard. In appropriate contexts, they may be necessary to express the special value of human life in our culture’’ (MacLean, 1986, p. 87). In saying that life is sacred, MacLean does not mean that life is ‘‘infinitely valuable’’ or literally ‘‘priceless’’, but that respect for individual human life is ‘‘deeply rooted and more or less universal’’ (p. 86), that it is a ‘‘special’’ value that we should be reluctant to trade for other ‘‘valued commodities’’ (p. 90). In brief, such ‘‘ritualized behavior’’ as rescuing, or attempting to rescue, sailors lost at sea or hikers lost in the wilderness or second-chance transplant patients, even if they ‘‘defy economic or even riskminimizing sense’’ (p. 87), nevertheless gives powerful expression to the ‘‘sacred value’’ of human life, and in this way contributes to social cohesion. It is especially 12

A corollary of the symbolic value of rescues is that they provide hope. Nord et al. (1995b) and Menzel et al. (1999) point out that hope of treatment is itself something that people value. Studies reveal that people are often prepared to forego a more cost-effective use of resources in order to preserve the hope of treatment for everyone: ‘‘a limited number of treatments offered to patients in a health state creates a rational basis for hope: no matter what health-related event occurs respondents know there is a chance of treatment and hope is a rational basis for utility’’ (Nord et al., 1995b, p. 90). The RR ministers to the hope that, should serious misfortune fall, there is a chance of receiving help, even if it is not justified on a purely cost-effective basis.

difficult, psychologically, to resist such efforts when the threat is unexpected and the circumstances shocking. Fried is sceptical about the cogency of this argument—which he calls ‘‘the symbolic value argument’’ (1969, p. 1425). The problem is, the money for rescues that reach beyond conventional CEA must come from somewhere. If the total budget allocated to saving lives is not increased, then the money must be diverted from programs aimed at saving ‘‘merely statistical’’ lives. The result then will be that the greatest number of lives is not saved in the long run. If the total budget allocated to saving lives is increased, then the money must be diverted from programs aimed at improving quality of life. But since the rescues reach beyond conventional CEA, the result in this case will be that ‘‘more lives are saved, but less of value is accomplished’’ (Fried, 1969, p. 1425). As Fried is aware, the crucial question is whether the reduction in overall lives saved (on the first interpretation) or the reduction in quality of life (on the second), is outweighed by the ‘‘high symbolic value’’ of the rescues—that is, by the utility gained from the public demonstration that life is precious and worth great effort to preserve. It is quite probable that many people receive satisfaction from the belief that they live in a caring society and, as a consequence, the importance of the ‘‘symbolic value’’ of rescues may well add more to the quality of life of the entire society than the reduction in the quality of life of a limited number of patients deprived of some health services. This seems to be the conclusion MacLean reaches: ‘‘a refusal to rescue, which might be rational ex ante, could save more lives while undermining society’s ability to express concern for the value of human life’’ (MacLean, 1986, p. 89). It would be implausible to suggest that those whose behaviour conforms to the RR perform a mental calculation of the costs and benefits of a rescue. To a large extent it just happens automatically. The distinction between rule-following and rule-governed behaviour makes this point clearer (Quine, 1974, pp. 104– 107; Cooper, 1975; Devitt & Sterelny, 1987, pp. 134– 140). Rule-following behaviour involves consultative acts. Those who follow rules entertain the relevant rules, they reflect on them, and modify their behaviour accordingly. The more general notion of rule-conforming behaviour, on the other hand, merely characterises behaviour that conforms to the relevant rule, but which does not necessarily involve such consultative acts. For example, knowing how to ride a bicycle does not consist in consulting mental representations of the relevant laws of physics, judging their applicability, and modulating one’s behaviour accordingly. A child can learn to ride a bicycle without having acquired the concepts of gravity, velocity, acceleration, and so on. Riding a bicycle is a skill or ability; it does not consist in applying propositional knowledge. In the same way, it is unnecessary to assume that those whose behaviour conforms to the RR

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

have reflected upon the rule or are able to articulate it, even if the rule correctly describes their behaviour. Just because their behaviour conforms to the rule does not mean they are following the rule. It is particularly implausible to think that people in some way entertain the RR at the time their behaviour can be described as conforming to it. People simply act when another person’s life is visibly endangered. The time for reflection on abstract principles is afterwards, if at all.13 Discussion The arguments above provide an ethical rationale for the RR, especially from a utilitarian perspective. However, they must be weighted against the argument that the RR discriminates on morally dubious grounds. Of course, it is sometimes justifiable to discriminate between individuals and groups if the grounds for discrimination are morally relevant.14 The problem is, it 13 Hare distinguishes between two levels of moral thinking: the intuitive and the critical (Hare, 1981, Part I). At the intuitive level we rely on prima facie principles that have the virtues of generality and simplicity, such as the Bible’s Ten Commandments: ‘‘Thou shalt not kill,’’ ‘‘Thou shalt not steal,’’ and so on. Moral thinking cannot remain at this level, however, because such principles have exceptions, and may generate conflicting obligations (particularly when novel situations arise). Nevertheless, such principles are useful in emergency situations, when there is not enough time to make complex calculations, or when acting contrary to our own immediate self-interests may be difficult. In the same way, the RR might be justified as a valuable prima facie principle or ‘‘rule of thumb’’. A general disposition in the community to act on the principle, ‘‘If someone’s life is visibly threatened, render assistance (without giving too much thought to the cost),’’ may do more to increase social welfare in the long run than its absence. At any rate, it is likely to do so given the world as we find it, where selflessness, courage, and so on, are sometimes in short supply and, consequently, remaining acts of selflessness are of particular value both because of their scarcity and through their positive contribution to the preservation of a type of social environment that promotes long-run well-being. 14 In brief, discrimination between people may be justified or unjustified, depending on the grounds for discrimination. For example, Daniels asks under what circumstances several people in an office building might have ‘‘equal access’’ to a supply of coffee. Some considerations seem relatively unimportant. ‘‘Should we worry about the fact that not all offices are equidistant from the pot? Some are thirty feet away, some only teny. Should we worry that some colleagues use more calories in walking to the pot than others?’’ However, other considerations are clearly more weighty. ‘‘If the lounge is open only to male colleagues, then female colleagues can complain they do not have equal access to the coffee. If the lounge is up a flight of stairs and there is no wheelchair ramp, then my paraplegic colleague may have a ground for claiming unequal access to the coffee.’’ (Daniels, 1982, p. 53.). The former but not the latter are defensible examples of justifiable discrimination.

2415

is difficult to see the prima facie discrimination inherent in the RR as defensible in this way. As Hadorn notes, the RR typically comes into play when the life of an ‘‘identified’’ individual is ‘‘visibly threatened’’ and rescue measures are available. In the case of anonymous individuals, by contrast, there is no opportunity for ‘‘human nature’’ or the ‘‘human psyche’’ to override the ‘‘utilitarian rationality’’ of CEA. But being identified does not seem to be a morally relevant criterion for discrimination. After all, those anonymous individuals who quietly die from preventable cancers in hospital wards are no less real than the trapped miner or lost sailor whose life is ‘‘visibly threatened’’, and their families will grieve no less for their deaths. Arguably then, the RR can be criticised on the ground of injustice, because justice requires that we do not discriminate between people on morally irrelevant grounds, and being ‘‘identified’’ does not seem to be a morally relevant ground.15 Jenni and Loewenstein (1997) locate the essence of the ‘‘identifiable victim effect’’ in the public preference for— or greater toleration of—wide rather than narrow distributions of risk. For example, people are more concerned about the risks of a vaccination program if only 10 per cent of the population are susceptible to adverse side effects, than if the whole population is 15 Being ‘‘identified’’ is closely associated with the notion of being present, and thus with the moral significance of distance. Kamm supports the intuition that we have a greater obligation to assist those who are near to us (such as a drowning child), compared with those who are far away (for example, those affected by famine in a third world country). She argues that we are permitted to give greater weight to our own interests and projects compared with the interests and projects of others. In other words, nearness or proximity per se is only of derivative moral significance. Rather than having any intrinsic moral significance itself, it is an indicator of what is important to the agent: ‘‘yif one takes advantage of the option of giving greater weight to what one cares about, there is y a duty y to take care of what is associated with the agent, for example, the area near her or near her means. After all, we are locatable human beings, positioned at the center of our world in virtue of our agent-centered perspective’’ (Kamm, 2000, p. 676). She adds: ‘‘It is the mind set—focusing on what looms larger for the agent—that is supposed to help explain the duty to aid related to the agent’s location or the location of his means’’ (Kamm, 2000, p. 676). However, this leaves the agent-centered perspective itself in need of justification. Do we really have a moral prerogative to give greater weight to our own interests and projects rather than giving equal weight to oneself and to others? While it is probably true that ‘‘what looms larger for the agent’’ has a role to play in explaining the RR response, giving special consideration to ‘‘what looms larger for the agent’’ is itself in need of moral justification. Singer and Unger, for example, dismiss the tendency to divide the world into insiders to whom we owe more, and outsiders to whom we owe less (Singer, 1972; Unger, 1996).

2416

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

susceptible, even if the number of people who will die is, say, one thousand in either case (Ritov & Baron, 1990). Jenni and Loewenstein speculate that the ‘‘identifiable victim effect’’ may be explained by this antagonism towards concentrated rather than shared distributions of risk: ‘‘In effect, identifiable victims become their own reference group, creating a situation where n out of n people will die if action is not taken’’ (1997, p. 238). The crucial question is, of course, what is the morally relevant reference group? When people’s behaviour conforms to the RR the relevant reference group becomes the identifiable person. But why is it the identifiable person, rather than those anonymous individuals who could (individually and collectively) benefit, perhaps even more, from the expenditure of the same resources? As Jenni and Loewenstein conclude: ‘‘Given that reference group size is often a matter of framing—a reference group of arbitrary size can be specified for virtually any hazard—a blanket endorsement of a policy that treats fatalities differently based on what proportion of the reference group they compose is normatively dubious’’ (Jenni & Loewenstein, 1997, p. 240). If we consider two people facing a similar threat, and focus on the utility expected by each from an intervention, it may seem unfair to favour one over the other. However, from a societal perspective—that is, taking into consideration utility source (4) —these cases may not be identical. Treatment of one person more than the other may increase social utility by reinforcing the view that life is precious, and that citizens live in a caring society. The equivalent prognosis of two patients, for example, does not preclude the possibility that extra benefit may be derived from treating one rather than the other. The extra benefit may not arise from an improvement to the patient’s health per se, but from the circumstances or context. For example, in one case treatment may give rise to a reinforcement of social values whereas in the second there is no such social benefit. If this is so, is it really unfair to give priority where the greatest benefit is gained? Is the social benefit arising from the reinforcement of widely held community values different in kind from the benefit of increasing the number of QALYs gained? Is it different in kind from the ‘‘indirect benefits’’ that economists generally recommend for including in the evaluation of health services that allow patients to contribute to society? It is worth distinguishing between two types of cases here: those where the societal demand for rescue measures is contrived by media coverage, often of a dramatic nature, and those where it is not so contrived. When the media deals with matters in an accurate and responsible way, the desire for action on the part of the community may well be a result of widespread sympathy for those who are suddenly facing a shocking and

unanticipated crisis. The urge to rescue in such situations reflects an awareness that any citizen may face a similar crisis, and an understanding that the patient and the patient’s family have not had the normal time for psychological adaptation to the situation. In such cases the RR may justify a systemic response going well beyond what would be justified by a conventional measurement of costs and individual medical benefits. On the other hand, in some cases the media may have a more detrimental impact upon the decisions made. As Pence observes, the RR ‘‘states that the patient who manages to get ‘‘rescued’’ by media attention (such as a photogenic child in liver failure whose parents get him on television) is the patient who will get the scarce organ’’ (Pence, 1998, p. 270). In cases where the public demand for action is largely contrived by sensational media coverage the charge of unfairness seems more justified. Why should those who are lucky enough or manipulative enough to attract media attention be thought to have a special claim on resources? A number of problems would arise if allocation decisions were based on media priorities. For a start, the driving goal of health care delivery would change. The medical profession is expected to be primarily concerned with improving the health of the community, and national health schemes are created and financed primarily to achieve this objective. The media is primarily driven by the profit motive. Improving the health of the community, while carried out by individual health professionals and policy makers, is very much a communitarian goal. Securing the greatest profit for the owners or shareholders of a newspaper or television network is by contrast a sectional interest. It is safe to say that allowing the media to prioritise health care (based on the ‘‘newsworthiness’’ of treatments or individuals) would have an adverse affect upon the health of the community in the long run.16 Even here, however, it must be acknowledged that, if media attention has focused on a particular individual’s plight, the response to that situation may well affect society’s self-perception. As Hadorn points out, the media coverage given to an individual denied access to lifesaving treatment would result in ‘‘continual societal discomfort and shame’’ (Hadorn, 1991a, p. 2219). It would be wrong to ignore actions that will remove such obstacles to the achievement of well-being— that is, ‘‘continual societal discomfort and shame’’— 16

This view is put by Jenni and Loewenstein when discussing the ‘‘vividness criterion’’: ‘‘Although people might reasonably respond in a more emotional fashion to familiar or vivid victims, it is less reasonable to endorse a policy that gives higher priority to more familiar victims. This would amount to allowing media coverage to determine aid allocation’’ (Jenni & Loewenstein, 1997, p. 240).

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

particularly when this is achieved by the promotion of concern for others. Concerns about the manipulation of social opinions will be mitigated in part, but not in full, by the fact that if these contrived benefits occur too regularly—if the media over-exposes the population to sensational medical crises—then the very basis of the benefit will be offset as people realise the unfairness of media-based priorities in the health sector. Nevertheless, to the extent that this does not occur, it may well still be desirable to promote the self-perception of being a humane society by accepting the inevitability of some media-based priorities. Utilitarians can justify this by pointing out that in this way social welfare may be maximised overall. In sum, this argument implies that we cannot completely dismiss the case of contrived preferences. Nevertheless, benefits arising in this way would need to be compared very carefully with the harmful effects of a media-based allocation of resources, even if the volume of resources involved is very small. We suspect that a consideration of these harmful effects would result in a significant discounting of the benefits attributable to contrived demand. For those who consider fairness to be of paramount importance these benefits would be discounted entirely, as they would advantage some individuals and disadvantage others on morally arbitrary grounds imposed by the media. For others, who consider the reason for social opinions to be of no relevance, benefits would not be discounted at all. Partial discounting would represent a compromise solution between these polar positions.

2417

We do not here come down on one side or the other in this perennial dispute between conflicting ethical theories. Rather, we have attempted to show what sort of normative considerations might be raised both in support of and against the RR, and in particular to indicate how it might be justified from a utilitarian point of view. Our conclusion is two-fold. First, the Rule of Rescue describes an important motivation that has been ignored by orthodox economics. However, it is almost self-evidently true that by this or some other name the RR describes an almost universal response to an impending catastrophe. For this reason it should not, as at present, be overlooked when resources are being allocated to meet social objectives. Secondly, the RR is likely to remain contentious as it may be both supported and criticised by legitimate ethical arguments and, at present, there is no agreement about how to resolve issues that lead to such a head-on clash of values. Consequently there is probably no unambiguous advice that may be offered to health economists, health service researchers or policy makers except to remember that the evaluation of health services is not simply a technical matter but a quintessentially ethical endeavour, and that in complex societies with divergent values there may be a range of considerations that may ‘‘trump’’ the utilitarian rationality that is implicit in cost-effectiveness analysis. The RR represents one such challenge to conventional CEA.

References Conclusion Utilitarians are guided by the goal of maximising quality and quantity of life in the allocation of limited health care resources, as is enshrined—at least in principle—in conventional CEA. There will, however, be occasions—and probably numerous occasions— when total utility can be maximised by abandoning the logic of conventional CEA in favour of ‘‘the powerful human proclivity to rescue a single identified endangered life, regardless of cost, at the expense of many nameless faces who will therefore be denied health care.’’ It is possible to justify the RR from a utilitarian point of view then, if the utilitarian point of view is construed widely and not limited to conventional CEA. By contrast, certain non-utilitarians will reject this defense of the RR. In particular, the discrimination against anonymous individuals that the RR entails will not sit easily with those for whom the concept of fairness goes beyond the simple maximisation of aggregate utility. For those who think that favouring identifiable over statistical lives is wrong, regardless of the utilitarian consequences, the RR will remain ethically questionable.

Barry, B. (1965). Political argument. London: Routledge and Kegan Paul. Bochner, F., Martin, E. D., Burgess, N. G., Somogyi, A. A., & Garry, M. H., 1994. Controversies in treatment: How can hospitals ration drugs? British Medical Journal, 308 (6933), 901–905, 907–908. Callahan, D. (1994). Setting mental health priorities: Problems and possibilities. The Milbank Quarterly, 72(3), 451–470. Cooper, D. E. (1975). Knowledge of language. London: Prism Press. Creadon, M., 1997. The ocean denied. Time, Vol. 149 (3), January 20. Daniels, N. (1982). Equity of access to health care: Some conceptual and ethical issues. Milbank Memorial Fund Quarterly, 60(1), 51–81. Devitt, M., & Sterelny, K. (1987). Language and reality. Oxford: Basil Blackwell. Dixon, J., & Welch, H. G. (1991). Priority setting: Lessons from Oregon. Lancet, 337, 891–894. Dolan, P., & Green, C. (1998). Using the person trade-off approach to examine differences between individual and social values. Health Economics, 7, 307–312. Dougherty, C. J. (1993). Review of economics and ethics in health care: The hippocratic predicament: Affordability, access, and accountability in American medicine, by Robert

2418

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419

M. Caplan. The New England Journal of Medicine, 328(18), 1358–1359. Dworkin, R. (1977). Taking rights seriously. Cambridge, MA: Harvard University Press. Eddy, D. M. (1994). From theory to practice: Principles for making difficult decisions in difficult times. Journal of the American Medical Association, 271(22), 1792–1798. Fried, C. (1969). The value of life. Harvard Law Review, 82(7), 1415–1437. Hadorn, D. C. (1991a). Setting health care priorities in Oregon: Cost-effectiveness meets the rule of rescue. Journal of the American Medical Association, 265(17), 2218–2225. Hadorn, D.C. (1991b). The Oregon priority-setting exercise: Quality of life and public policy. Hastings center report, Vol. 21 (3), pp. 11–16. Hadorn, D. C. (1996). The Oregon priority-setting exercise: Cost-effectiveness and the rule of rescue, revisited. Medical Decision Making, 16(2), 117–119. Hare, R. M. (1981). Moral thinking: Its levels, method and point. Oxford: The Clarendon Press. Herman, B. (1981). On the value of acting from the motive of duty. The Philosophical Review, XC(3), 359–382. Jenni, K. E., & Loewenstein, G. (1997). Explaining the ‘‘identifiable victim effect’’. Journal of Risk and Uncertainty, 14, 235–257. Jonsen, A. R. (1986). Bentham in a box: Technology assessment and health care allocation. Law, Medicine and Health Care, 14, 172–174. Kamm, F. M. (2000). Does distance matter morally to the duty to rescue? Law and Philosophy, 19(6), 655–681. Kant, I. (1991). In H.J. Paton (Ed. and Transl.), Groundwork of the metaphysic of morals. London: Routledge. MacLean, D. (1986). Social values and the distribution of risk. In D. MacLean (Ed.), Values at risk. New Jersey: Rowman and Allanheld. Mehrez, A., & Gafni, A. (1989). Quality-adjusted life years, utility theory and healthy-years equivalents. Medical Decision Making, 9(2), 142–149. Menzel, P., Gold, M.R., Nord, E., Pinto-Prades, J.-L., Richardson, J., Ubel, P., 1999. Towards a broader view of values in cost-effectiveness analysis of health. Hastings Center Report, Vol. 29 (3) (pp. 7–15). Mooney, E. F. (1992). Sympathy. In L. C. Becker, & C. B. Becker (Eds.), Encyclopedia of ethics, Vol. II. New York, London: Garland Publishing. Mooney, G., & Olsen, J. A. (1994). QALYs: Where next? In A. McGuire, P. Fenn, & K. Mayhew (Eds.), Providing health care: The economics of finance and delivery. Oxford: Oxford University Press. Murray, C., & Lopez, A. (1996). The global burden of disease. Published by the Harvard School of Public Health on behalf of the WHO and the World Bank. Harvard: Harvard University Press. Nord, E. (1993). The trade-off between severity of illness and treatment effect in cost-value analysis of health care. Health Policy, 24, 227–238. Nord, E. (1996). Health state index models for use in resource allocation decisions. International Journal of Technology Assessment in Health Care, 12(1), 31–44. Nord, E. (1999). Cost-value analysis in health care. Cambridge: Cambridge University Press.

Nord, E., Pinto, J. L., Richardson, J., Menzel, P., & Ubel, P. (1999). Incorporating societal concerns for fairness in numerical valuations of health programmes. Health Economics, 8, 25–39. Nord, E., Richardson, J., & Macarounas-Kirchmann, K. (1993). Social evaluation of health care versus personal evaluation of health states: Evidence on the validity of four health state scaling instruments using Norwegian and Australian surveys. International Journal of Technology Assessment in Health Care, 9, 463–478. Nord, E., Richardson, J., Street, A., Kuhse, H., & Singer, P. (1995a). Maximising health benefits vs egalitarianism: An Australian survey of health issues. Social Science & Medicine, 41(10), 1429–1437. Nord, E., Richardson, J., Street, A., Kuhse, H., & Singer, P. (1995b). Who cares about cost? Does economic analysis impose or reflect social values? Health Policy, 34, 79–94. Osborne, M., & Evans, T. W. (1994). Allocation of resources in intensive care: A transatlantic perspective. The Lancet, 343, 778–780. Pence, G.E. (Ed.). (1998). Classic works in medical ethics. Boston, MA: McGraw-Hill. Pinto, J. L. (1997). Is the person-trade-off a valid method for allocating health care resources? Health Economics, 6, 71–81. Quine, W. V., 1974. Methodological reflections on current linguistic theory. In: Gilbert Harman (Ed.), On Noam Chomsky: Critical essays. New York: Anchor Books. Rawls, J. (1971). A theory of justice. Oxford: Oxford University Press. Richardson, J. (1994). Cost-utility analyses: What should be measured? Social Science & Medicine, 39, 7–21. Richardson, J., Hawthorne, G., & Day, N. A. (2001). A comparison of the assessment of quality of life (AQoL) with four other generic utility instruments. Annals of Medicine, 33(5), 358–370. Richardson, J., & Nord, E. (1997). The importance of perspective in the measurement of quality adjusted life years. Medical Decision Making, 17, 33–41. Ritov, I., & Baron, J. (1990). Reluctance to vaccinate: Omission bias and ambiguity. Journal of Behavioral Decision Making, 3, 263–277. Rousseau, J.-J. (1994). A discourse on the origin of equality (F. Philip, Transl.). Oxford: Oxford University Press. Scarre, G. (1996). Utilitarianism. London: Routledge. Singer, P. (1972). Famine, affluence, and morality. Philosophy and Public Affairs, 1(3), 229–243. Stocker, M. (1976). The Schizophrenia of modern ethical theories. The Journal of Philosophy, LXIII(4), 453–466. Torrance, G. W. (1986). Measurement of health state utilities for economic appraisal. Journal of Health Economics, 5(1), 1–30. Torrance, G. W (1987). Utility approach to measuring health related quality of life. Journal of Chronic Diseases, 40(6), 593–600. Ubel, P. (2000). Pricing life. England, Cambridge, MA, London: The MIT Press. Ubel, P. A., Arnold, R. M., & Caplan, A. L. (1998). Rationing failure: Ethical lessons of retransplantation. In G. E. Pence (Ed.), Classical works in medical ethics (pp. 270–284). Boston, MA: McGraw-Hill.

J. McKie, J. Richardson / Social Science & Medicine 56 (2003) 2407–2419 Ubel, P., Scanlon, D., Loewenstein, G., & Kamlet, M. (1996). Individual utilities are inconsistent with rationing choices: A partial explanation of why Oregon’s cost-effectiveness list failed. Medical Decision Making, 16, 108–119.

2419

Unger, P. (1996). Living high and letting die: our illusion of innocence. New York: Oxford University Press. Williams, A. (1996). QALYs and ethics: A health economist’s perspective. Social Science & Medicine, 43(12), 1795–1804.