2. WELFARE INTERDEPENDENCE

The World Health Organization uses disability-adjusted life years (DALYs) to report the burden of disease – that is, the harm resulting from premature mortality and disability from disease and injury for the population in different countries. The DALYs associated with a given condition are the sum of the years of life lost and the years lived with disability due to that condition. The conditions range from facial vitiligo through rheumatoid arthritis to quadriplegia.

The burden of premature mortality from a condition is measured on the basis of years of life lost due to that condition. The measure represents the gap between current mortality and an ideal standard of life expectancy. The burden of death at each age is partly determined by the expectation of life at that age. The standard life expectancy is set to 82.5 years at birth for females and 80 years for males; at 50, it falls just short of 34 years for females and 31 years for males. The differences between the sexes occur because of their ostensibly different survival potential.

Standard life expectancies only partly determine the years of life lost from premature death, since years of life lost are obtained by subjecting standard expected years of life to age-weighting and discounting. Thus, the death of a newborn results in 33.12 years of life lost if the child is female and 33.01 years of life lost if the child is male; for a 50-year old, the years of life lost are 17.69 and 17.12, depending on the person's sex.Footnote ⁷

The burden of morbidity from disease and injury is measured in terms of lost quality of life during the time the person is in a worse health state than perfect health. This component of DALYs is years lived with disability. Lost quality of life due to different conditions is represented by disability weights between 0 and 1, with 0 for perfect health and 1 for death. Years lived with disability are calculated by weighting the time spent with the disability with its disability weight. But years lived with disability are also subject to age-weighting and discounting. Age-weighting is applied both to the age of the onset of the disability and the duration of the time in diminished health.

DALYs are therefore a composite measure of the harms resulting both from lost quantity of life and lost quality of life. Since they are standardized, the burden of disease in different regions and populations are comparable and health trends can be discerned. For example, according to the projections of the 1996 global burden of disease study, the number of DALYs associated with HIV in both its treated and untreated forms will be just below 2.3 million in 2010 in all established market economies, while it will be well over 20 million in that year in Sub-Saharan Africa (with a slightly larger population). DALYs can also be used as an outcome measure for the purposes of cost-effectiveness analysis of health interventions and policies. The objective associated with the reduction of burden of disease is to minimize DALYs, reducing the gap between ideal population health and the actual population health.

Let us consider age-weighting in the context of disability-adjusted life years. Suppose that in my original example both patients are female. If we give the life-saving drug to the 20-year-old, the overall years of life lost, without age-weighting and discounting, will be 78.1 – the sum of the standard life expectancies at ages 30 and 60. If we give the drug to the 60-year-old, the overall years of life lost, again without age-weighting and discounting, will be 79.28 – the sum of the standard life expectancies at ages 20 and 70. The reason for the difference is that the decrease in life expectancy between the ages 20 and 30 is larger than between 60 and 70. For obvious reasons, life expectancies decrease at higher ages; but in the life expectancy table used for calculating years of life lost, they decrease at a decreasing rate after childhood.

What about years lived with disability in the example? Since the extra ten years, whomever they go to, will be spent in perfect health, there are no years lived with disability to consider. Thus, the overall DALYs are 78.1 and 79.28.

DALYs would therefore recommend that we give the drug to the younger patient. But this is not because we value years of life lived at different ages differently. We simply minimize the overall burden of premature mortality thereby.

Nevertheless, we might be tempted by the following argument. The difference between 78.1 and 79.28 years of life lost – around one year and two months – is very small, especially given that no matter what we do, there will be almost 80 lost years of life. We might be tempted to argue that such a small difference should not make the difference between life and death. Some philosophers have argued that there are “irrelevant utilities” – harms and benefits which are insufficient to justify a life-or-death choice.Footnote ⁸ Following their lead, we might come to the conclusion that one year and two months are not sufficient to justify giving the drug to the younger patient. We might instead, for example, prefer to allocate the life-saving drug by tossing a fair coin, giving an equal chance to both patients.

DALY calculations, however, normally do value years lived at different ages differently. The relative value of a year lived at 20 is around one and a half the relative value of a year lived at 60. If we introduce age-weighting in our example, giving the drug to the 20-year-old results in approximately 69.28 (non-discounted) DALYs; giving the drug to the 60-year-old results in slightly less than 77 (non-discounted) DALYs. The difference is over seven years of life lost. Even if we were attracted to giving an equal chance to both patients before, now the burden of premature mortality in the second case is substantially larger than in the first. Arguably, such a large difference should be sufficient to justify giving the drug to the 20-year-old. Thence age-weighting can make the difference between life and death.

Why does the WHO include age-weights in the measurement of the burden of premature mortality, disease and injury? Christopher Murray, one of the developers of DALYs, appeals to two principles that he claims rest on our ordinary notion of fairness. These are as follows:

• Principle 1. The burden calculated for like health outcomes should be the same.

• Principle 2. The non-health characteristics of the individual affected by a health outcome that should be considered in calculating the associated burden of disease should be restricted to age and sex.Footnote ⁹

These principles, taken together, establish information constraints on the calculation of DALYs. How bad a health outcome is evidently depends on a lot of variables. Some of these variables, like the etiology of the disability, are biomedical. These are relevant to the description of a health state. But DALYs do not merely describe health states; they measure their burden. According to the second principle, variables like income, level of education or the availability of social support are irrelevant for this exercise.

Murray (Reference Murray, Murray and Lopez1996: 7) uses the following example to illustrate the motivation for the principles. Suppose two patients who are identical in every respect except that one is rich and the other is poor arrive simultaneously at the emergency room, and there are insufficient medical supplies to treat both of them. He claims that it would be unfair to choose to save either of them on the basis of their wealth. Hence the principles should reflect this unfairness.

The problem with this example is that claims about how people's health should be evaluated cannot be inferred from claims about how they should be treated. We might agree with Murray that fairness requires that we should not discriminate between the patients on the basis of their wealth, thus they should have an equal chance to be saved. Nonetheless, we might also claim that more benefit would be produced by saving the rich patient (on the assumption that she has more well-being), or that the benefit produced by saving the poor patient should have more weight (on the assumption that benefits to the worse off have more value). Our view about how the patients ought to be treated should not distort the measurement of the value of the outcomes.

Some philosophers suggest that fairness is about mediating people's conflicting claims for some good – in this case, saving their lives.Footnote ¹⁰ If this view is correct, Murray's principles cannot rest on the notion of fairness. I want to suggest, however, that the principles rest instead on the notion of impartiality. Fairness concerns how people should be treated; impartiality concerns how their interests should be taken into account. Thus, one way a choice can be unfair is if it is not based on an impartial appraisal of the interests of the people affected. On this interpretation, the principles give shape to some ideas about how we should evaluate health benefits by setting constraints on which consequences of disease or injury are relevant. If two people are in the same health state, it would not be impartial to calculate the burden of their disability differently – especially on the basis of a non-health characteristic like income.

Separating the roles of fairness and impartiality in motivating the principles is also important for the following reason. Its developers recommend DALYs both as a measure of population health and as a tool for cost-effectiveness analysis on the basis of which resource allocation decisions can be made. Measuring the burdens of ill-health in an impartial way is important for both purposes. But to suggest that the underlying principles already incorporate our concern for fairness is to disregard fairness claims that may arise independently. In order to make just resource allocation decisions, an impartial appraisal of the burdens of benefits is certainly necessary. However, it may in itself be insufficient, for other considerations may need to be taken into account. Therefore, whether or not DALYs are appropriate as a measure of burdens, they cannot be assumed without further argument to be an appropriate tool for decision making.

How can the constraints articulated in Murray's principles be defended? In particular, how can we justify Principle 2 that allows age to be taken into account? (I will ignore sex.)

One argument starts from the fact that people's productivity varies at different ages. Young adults contribute more to economic output than children and the elderly do. On this argument, DALYs are interpreted as a measure of the social value of health in terms of its contribution to human capital and economic prosperity. From this perspective, however, many other variables besides age are relevant for the evaluation of health: income and level of education are obvious examples. Thus, this argument cannot defend Principle 2.

Murray also rejects this argument, but he does not reject the basic idea that age-weighting makes it possible to incorporate the social value of health into the measurement of the burdens of ill-health. His own argument begins by noting that people promote the well-being of others in many ways – by providing care, household services and so on. These contributions cannot be captured if we focus on economic output and human capital only. Nonetheless, welfare interdependence, as he calls it, has a crucial role in society. If we broaden our focus to take all sorts of welfare contributions into account, it follows that, on the one hand, it is not necessarily the case that people with higher income or level of education contribute more to the welfare of others. Income or education are therefore not relevant variables. On the other hand, young adults tend to contribute more to the well-being of others by providing for their children and taking care of the elderly. Some of this contribution is economic, but most of it is non-monetary. Consequently, age is the most suitable variable to capture welfare interdependence.

One may worry that Murray's conclusion is too strong. For one thing, even if we can exclude income and education, other variables may be relevant to capture welfare interdependence. Why should we not take into account, for instance, whether one is a parent? On the other hand, welfare interdependencies are highly variable across cultures. In some of them, older generations provide substantive family support, including helping parents to raise their children. Emphasizing the dependence of the elderly on their adult children may be indicative of a form of cultural bias.

Murray also argues that using age as a variable is not discriminatory in the way using income or education level would be. Most people live through different ages: with the passage of time, they proceed from childhood to adulthood to old age. Thus, age-weighting does not discriminate between individual lives, only between periods within lives. Even though there certainly are variations in the extent to which particular individuals provide for the well-being of others, by and large it is true that people at different ages contribute differently. Age-weighting remains impartial with respect to the characteristics of particular persons.

Thus, there are two claims embedded in the argument from welfare interdependence: that age is the appropriate variable to take welfare interdependence into account, and that using this variable is not discriminatory.

Consider the second claim first. It seems to me that age-weighting can be discriminatory, especially if, for instance, the relative weight of childhood is smaller than the relative weight of young adulthood, as it is in DALYs. If a child suffers from the same disease for the same period as a young adult, the burden associated with her condition will typically be smaller, and treating the adult may lead to a larger overall reduction of DALYs. One might believe that an older person cannot reasonably complain if her interests are given a smaller weight than the interests of a younger person, because the older person has already benefited from having her interests count for more when she was younger. But this reasoning is not available in the case of the child: she has not been compensated by having her interests count for more at some other time (and she might not ever be compensated if she dies prematurely). Thus, the child could reasonably complain that her interests are given lower weight than the interests of her older peers. Unless some further argument is given, impartiality is violated in her case. Treating her differently on this basis would be unfair.

Leaving aside the worries about age-discrimination, I will concentrate on the first claim in the rest of this section. If welfare interdependence is taken into account, DALYs provide more than a summary measure of the badness of premature mortality and the consequences of disease and injury. Because of age-weighting, they also represent the social value of health. The social value of the health of a 20-year-old is greater than that of a 60-year-old, since young adults contribute more to the well-being of others. Other things being equal, depression when you are 20 is not worse for you than depression when you are 60; rather, it tends to be worse for others. It tends to be more burdensome for society.

Whether this rationale is adequate turns on how we identify the consequences of disease and injury. And here we run into problems. Here's why.

The WHO distinguishes between impairment, disability and handicap. An impairment is a loss or abnormality of psychological, physiological or anatomical function, described in biomedical terms. It is the direct consequence of disease or injury. A disability is a loss or restriction of ability to carry out an activity which is considered normal for humans as a consequence of an impairment. A handicap, in contrast, is the disadvantage resulting from an impairment or disability which limits or prevents the individual to fulfil her role in her economic, social and cultural environment.Footnote ¹¹ For instance, suppose a chemical imbalance in the brain causes depression. This impairment may lead to a disability characterized by loss of motivation and inability to concentrate, which, in turn, may lead to unemployment and social isolation – a handicap.

It is not entirely clear how handicap and well-being differ from one another. The notion of handicap captures the disadvantage that results from a disability, which varies with the social role and circumstances of the person. Perhaps one could say that a handicap reflects a particular way a person's life goes worse for that person. Thus, a handicap represents the consequences of a given disability only, such that different disabilities result in different handicaps. But the disadvantage that results from different disabilities cannot be partitioned this way. The same handicap may be the consequence of several disabilities. Co-existent disabilities can amplify each other's impact, and the removal of one disability may not lead to a reduction of disadvantage. I doubt that a distinction between handicap and well-being can be maintained.

Disability-adjusted life years attempt to represent disability. Suppose, however, that we tried to interpret DALYs some other way. For instance, we might suppose that DALYs measure handicap. In this case, they would capture the full consequences of health for well-being. These consequences would vary greatly with age, income, level of education, availability of support services, and many other factors. One relevant factor would be the support of others. That is, the way young adults contribute to the well-being of others would already be taken into account in DALYs.

This is because, on this interpretation, different disability weights would have to be assigned not only to treated and untreated forms of diseases and injuries, but also to the same disease or injury depending on whether the person is rich or poor, young or old, educated or not. The effects of welfare interdependence would already be incorporated into these disability weights. Weighting them by age would be counting them twice over.

On the other hand, suppose that DALYs measure impairment. But different impairments have different consequences for a person's well-being. Thus, if DALYs were interpreted this way, focusing on the medical condition rather than the badness of the condition, they could not incorporate the consequences of health for well-being at all. They would not measure the value of health or the burdens of ill-health.

As a matter of fact, Murray et al. (Reference Murray, Salomon, Mathers and Lopez2002: 735–7) argue that disability weights are a measure of health states and not a measure of the value or utility of those health states. But few commentators agree that it is possible to order and compare health states on any basis other than their value – at least beyond trivial comparisons. On the one hand, a person in full health is definitely healthier than a person with a disability. On the other hand, it is difficult to see how you can compare a pair of complex health states involving different disabilities unless you are prepared to appeal to their value. One way of doing that is to compare the health states with respect to their impact upon the person's well-being – which depends, among other things, on the economic, social and cultural environment. In any case, if DALYs are intended to serve as a measure of health rather than a measure of the value of health, then welfare interdependence should have no place in them when they are presented as a summary measure of population health.

DALYs attempt to strike a middle ground between impairment and handicap by focusing on disability. Welfare interdependence, however, should be relevant only to handicaps. Because of age-weighting, it turns out that depression is less bad if you are 60 than if you are 20. But this is not because, other things being equal, it is less of a disability to be depressed when you are 60 than to be depressed when you are 20; it is because, other things being equal again, it is a less bad handicap to be depressed at 60 than at 20.

Consequently, the middle ground of disability is unstable ground on which to justify age-weighting. DALYs attempt to include one sort of factor in determining the burdens of ill-health and exclude others. This is arbitrary. Either DALYs represent handicap, in which case many other variables besides age should be taken into account, or they represent disability, in which case age should be excluded along other variables.

Murray is aware of this problem. At one point, he suggests that disability weights represent average handicap (Reference Murray, Murray and Lopez1996: 38–9). I am not sure how that “average” is to be understood. When disability weights were assigned for different conditions, assessors from different regions of the world were told to think of the average individual in their own society in their average social environment. But disability weights are uniform across all regions. If they do represent global average handicap, then it would seem that they conceal huge disparities between different parts of the world. To refer back to my earlier example: the overall consequences for well-being of living with HIV in Sub-Saharan Africa, even if one receives treatment, are very different than living with HIV in established market economies. If DALYs represent average global handicap, they underestimate the burden of HIV in the former.

DALYs attempt to measure the burden of disease by the impact of ill-health on people's well-being, taken in a limited sense as disability. The variations of the impact with age either has no place in such a measure or they should be taken into account along with other sources of disadvantage associated with ill-health – but in this case, they lead to the double-counting, which certainly violates impartiality. Justifying age-weighting as a way to account for welfare interdependence does not succeed.

3. FAIR INNINGS

Some may prefer giving the life-saving drug to the 20-year-old person in my example because they think this is what fairness requires; it would be unfair to give it to the older person or selecting the recipient randomly. Several philosophers have argued for this view, using the fair innings argument.

The argument is intended to establish the conclusion that if different persons compete for life-extending resources and they are identical in every relevant respect except for their age, then it is justified to give priority to the younger person even if the net benefit accruing to that person is not larger than the net benefit that would be received by the older person. Different formulations of the argument specify different relative weights to ages.Footnote ¹²

A crucial premise is that justice or fairness can be applied to quantities of life. As a person grows older, she approaches or reaches her “fair share” of life – her fair innings. Hence, at least relative to others who have had less of her share of life, her claim on resources is diminished or relinquished entirely. The rationale that arguments from fair innings provide for age-weighting is that standard evaluations of the benefits of healthcare resource allocation fail to take account of requirements of fairness. One way to take these requirements into account is to introduce age-weights.

The fair innings argument was introduced by John Harris (Reference Harris1985: 87–102; Reference Harris, Bell and Mendus1988), although he gives it a very limited role. Harris argues that since everyone values going on living equally, everyone suffers the same injustice if their lives are cut short. In deciding whom to save, what matters is not the magnitude of the benefits, but the magnitude of the injustice. This implies that age should not be taken into account. This is the anti-ageist argument, which gives support to the following principle:

• Anti-Ageist Principle. Each person should have an equal claim for a life-saving resource, even if some person's prospect with respect to the quality or length of her life is worse than the prospects of others.

There is one exception, though. Some people have had a reasonable life – for instance, they have lived three scores and ten years. When they reach 70, they had their fair innings, and the remaining years are “bonus time”. They cannot reasonably complain that an injustice is done to them if resources are diverted to save the life of those who are under the threshold. This is Harris's fair innings argument, which gives support to another principle:

• Harris's Fair Innings Principle. If it is impossible to save the lives of both a person who has had her fair innings and another person who has not, then the life of the person who has not had her fair innings should be saved.

Conjoined, the two principles determine the following values for evaluating alternatives in life-saving situations. If the people affected are all under the fair innings threshold, saving any of them has uniform value, since, by the Anti-Ageist Principle, we are prohibited to differentiate by prospective quality or length of life. But if a person has had her fair innings, saving her has strictly smaller value (perhaps none at all). This difference becomes important when we have to choose between people under and over the fair innings threshold.

There are many difficulties with Harris's view.Footnote ¹³ I want to concentrate on the claim that the scope of the Fair Innings Principle can be limited the way Harris intends. Note that he assumes that there is a clear-cut distinction between life-saving and other contexts. He claims that the principle applies if and only if it is “clearly impossible to postpone the deaths of all those who wanted to go on living” (Reference Harris1985: 94).

Suppose you have to choose between giving the only drug you have to an older person who has had her fair innings and a younger person who has not. If you give the drug to the younger person, she will certainly survive, while the older person will die with a 0.9 probability. If you give the drug to the older person, she will certainly survive, and the younger person will also survive with a 0.9 probability. Giving the drug to the older person makes it very likely that the deaths of both persons can be postponed, although there is a slim chance that the young person dies. The older person is saved for certain, although she already had her fair innings.

The situation is illustrated in Figure 1. The numbers in the cells indicate the probability of the different outcomes, depending on who gets the drug.

Figure 1.

I am not sure whether we should regard this situation as one in which it is “clearly impossible to postpone the deaths of all”. Suppose it is such a situation: it is clearly impossible to make sure that no-one dies, regardless of what you do. If you give the drug to the older person, it is highly likely that both persons will survive – but it is not certain. So Harris's Fair Innings Principle should apply.

But if it does, then it prohibits giving the drug to the older person. You should give the drug to the younger person. But if you do that, the older person is highly likely to die.

On the other hand, suppose that this is not a situation in which Harris's Fair Innings Principle applies. It is not clearly impossible to postpone the deaths of all, since if you give the drug to the older person, both of them, with a bit of luck, can survive. Hence your choice should be determined by the Anti-Ageist Principle only. It tells you to give an equal chance of getting the drug to both persons. But if you do that, there is a significant probability that the older person will die.

You might believe that you should aim to minimize the risk of anyone dying. You can do that easily by giving the drug to the older person, or at least giving her a higher chance of getting the drug.Footnote ¹⁴

These counter-intuitive results are not merely artifacts of the Anti-Ageist Principle. They are due to the assumptions that a fair innings is a fixed threshold, beyond which a person had a full, reasonable life, and that the years remaining after one reached her fair innings are a kind of bonus, giving rise to no obligations of justice on others.

Of course, as Harris notes, setting a fair innings at three scores and ten years is a simplification. Besides quantity, the quality of life also matters. Reasonable people might disagree whether someone had her fair innings. Hence Harris further limits the scope of his Fair Innings Principle by insisting that it only applies if reasonable people agree whether someone had her fair share of life. But whenever they do, there will be a cut-off point somewhere, and we run into these problems.

There is, however, another way of interpreting the notion of fair innings. On the view I have been discussing, a fair innings is a more or less fixed threshold of life years beyond which the person has no claims for having her life prolonged, given that she competes for the available resources with others who have not reached this threshold. On an alternative view, a person's claim on resources steadily diminishes the longer she lives: she has had more of a fair innings. Justice requires that when we allocate scarce resources we take into account the share of life that a person has had relative to others. We could formulate a fair innings principle for this view as follows:

• Life-Share Fair Innings Principle. It is not unjust to allocate life-extending resources on the basis that the younger a person is, the weightier, other things being equal, that person's claim is on available resources.Footnote ¹⁵

This view is difficult to reconcile with the rationale for age-weighting from the notion of a full, reasonable life. On the natural understanding of that notion, a person's life has either been full and reasonable or not. Thus, some other rationale must be given to motivate the view that the weight of a person's claim diminishes as she ages.

One rationale might be that claims on distributable resources are generated by need.Footnote ¹⁶ Thus, in the context of the allocation of life-extending resources, one factor that determines the strength of a person's need is age. Other things being equal, the older a person is, the less she needs the benefits of life-extending treatment such that the weight of her claim upon available resources steadily diminishes. On this view, age-weighting ensures that we remain sensitive to differences with respect to need.

It should be clear that “need” is not used in this argument in its ordinary sense. If a person is about to be killed, we do not think he needs to be saved less if she is older. If there are two patients with a headache, we are unlikely to believe that one needs a pain reliever more just because she is younger.

Frances Kamm (Reference Kamm1993: 234–5; Reference Kamm, Murray, Salomon, Mathers and Lopez2002: 692) suggests that when we allocate life-extending resources, we should distinguish between need and urgency. Need is determined by how badly a person's life will have gone if she is not saved. In contrast, the urgency of helping a person is determined by how badly a person's life will go if she is not saved. Note that “urgency” is not used in its ordinary sense here either, in which it refers to how soon the employment of the resource is necessary. On Kamm's definitions, need is based on an evaluation of the overall benefit a person enjoys throughout her life, depending on whether she gets the resource; urgency refers to the prospective benefit that the person enjoys from the time of allocating the life-extending resource, depending on whether she gets it. When we consider need, our evaluation concerns a person's whole life, but when we consider urgency, our evaluation looks only at the future. Need and urgency are two different perspectives with respect to which we can evaluate competing claims.

Kamm argues that sometimes fairness requires that we decide on the basis of need. This is true when the candidates for a resource do not differ with respect to urgency, but it might be true even if thereby we help the less urgent. Consider the case in which there is no difference in urgency. For illustration, my original example will once again do. It is equally urgent for A and B to get the life-saving drug, since if they do not, they will die. We might also consider need, in which case we should

If A does not get the drug, she will have lived only 20 years, while if B does not get the drug, she will still have lived 60. Hence A's need seems to provide a stronger claim on the life-saving drug. But this is not necessarily so. Need is based on an evaluation of how well overall a person's life goes relative to the lives of others. In order to decide whether A or B is needier, we need to know more. If the older person's life will have been going badly enough to be worse overall than the younger person's life, then she is needier than the younger person. If her life will have been going just well enough overall to be as good as the younger person's life, they are equally needy.

That is to say, it is not life as a good, but the goods in life that matter: if A is needier, that is not because she will have had less life, but because she will have had fewer goods in her life. The ceteris paribus clause in the Life-Share Fair Innings Principle cloaks this distinction, since it can be read in different ways. On one reading, other things are equal only if the candidates for the resource will have been – that is, would overall be – equally well off. Then, according to the principle, the claim of the younger person gets more weight. But if the candidates for the resource would overall be equally well off, they are equally needy. Their needs generate claims with the same weight. This is in conflict with the principle. Therefore, the concept of need cannot give a rationale for the Life-Share Fair Innings Principle. The principle must be motivated by other considerations.

On another reading, other things might be equal even if the candidates would not be overall equally well off. It might be that the younger person would be worse off – that is, if not given the resource, she will have had less good in her life (perhaps merely due to having lived less). Therefore, she is needier. Or it might be that the older person would be worse off overall, despite having lived longer. In that case, she is needier. Of course, normally younger people tend to have had fewer goods in their life – especially compared with substantially older people. So often we can take age as an indication of need. But then the benefit that we have to allocate is not valued differently according to age; it is valued differently according to need. The two are different, since we care about age only insofar as it can stand as an indicator for something else. What we ultimately care about, in this case, is need.

Kamm might not disagree. She says that “age can be given some weight in decision making in virtue of its role in identifying the neediest candidates” (Reference Kamm1993: 248). Thus, if people of different ages compete for resources, we should favour the young, since their need is likely to be greater. But if this is so, then the Life-Share Fair Innings Principle is redundant. Appealing to other considerations, perhaps together with using age as an indicator, are sufficient to do the work.

Similar problems arise for other proposals which may be employed to defend the Life-Share Fair Innings Principle. For instance, we might try to defend the principle by appealing to equality. Alan Williams (Reference Williams1997) argues that the fair innings argument can be quantified in order to determine the equity-efficiency trade-offs involved in any attempt to reduce health inequalities.

Williams assumes that a fair innings is determined by the mean quality-adjusted life expectancy (QALE) at birth. Here we consider actual life expectancy at birth in a given country. This contrasts with standard life expectancies used in DALYs, which reflect some ideal, biologically determined survival potential. Actual life expectancies vary with societies and populations. They are adjusted for quality, since a long life with poor quality deserves consideration just as much as a shorter life with higher quality. QALEs represent the “normal” span of life people should enjoy in their society.

Since the better off tend to have greater QALEs, health disparities can be reduced by selecting policies which equalize QALEs by sacrificing some of the overall health of the population. This objective applies to the distribution of healthcare resources in all settings, not only in the context of saving or extending lives. The degree to which people are willing to sacrifice overall health reflect their aversion to inequality. On this view, the objective is to equalize health outcomes throughout people's lives.

For example, suppose you have to allocate funds between pediatric and geriatric care services. You know that pediatric care services would be utilized mainly by poorer people, and they would contribute to longer and better quality lives for the worst off. But you also know that more better-off people will need geriatric care services, and the overall improvement in the health of the population would be larger if they were given priority. Based on fair innings considerations, you may nevertheless choose the former program even at the expense of lower population health in order to reduce disparities in quality-adjusted life expectancies.

This view focuses on the disparities among QALEs of social classes. Different weights are attached to additional life years gained by people according to the social class they belong to. People from the worst-off class who exceed their fair innings tend not to be penalized the way their better off contemporaries are. Hence, Williams's version of the fair innings argument is not a proposal to weight health benefits to reflect their differing value due to the age of the recipient. Rather, it takes age – more precisely, quality-adjusted life expectancies – as a proxy equalizandum and age-weighting as an instrument of equalization.

The more accurately quality-adjusted life expectancy at birth tracks how well off a person is going to be throughout her life, the more accurate instrument age-weighting might be to approximate equality of outcomes or to indicate differential need. Whether using it is appropriate is partly an empirical question. But, as we have seen in Section 2, the measurement of health is difficult and controversial. Thus, age-weighting as an instrument of equalization is bound to leave us uncertain as to whether we target the important inequalities and how successful we are in achieving our aim. It is also more difficult to influence health inequalities than, for instance, inequalities of income and wealth – hence age-weighting may be a grossly inefficient policy instrument. Furthermore, it is a controversial question whether information on inequalities in terms of QALEs, QALYs, or similar measures allows us to identify injustices in society which are morally important to rectify. This is a large and complicated issue beyond the scope of this paper.Footnote ¹⁷

One consideration against using age-weighting as an instrument in the service of equality or other objectives might be this. One of the ways in which this paper's opening example is unrepresentative of decisions for the allocation of life-extending resources is that whoever gets the life-saving drug will derive the same net benefit: both the 20-year-old and the 60-year-old would live for another ten years with the same quality of life. Often, however, younger people will derive more benefit from the same resource since their remaining life is longer or their quality of life is higher or both. Thus, if we take into account overall expected net benefits, the young will tend to get priority simply in virtue of having better prospects.

The magnitude of expected benefits will be one factor in the allocation of life-extending resources on any sensible view. This already, so to speak, stacks the cards against the elderly. So far, the only rationale we have found for age-weighting health benefits is that by age-weighting we can promote some other value with which age is arguably connected. If we are concerned with these other values, we should try to take them into account directly.

4. AGE-WEIGHTING AND JUSTICE

The familiar device of the veil of ignorance is often used in discussions on justice in health (see Daniels Reference Daniels1985, McKie et al. Reference McKie, Richardson, Singer and Kuhse1998). How would rational individuals choose to allocate healthcare resources from an impartial position? What role would they give to age?

In order to answer these questions, Norman Daniels argues that the interpersonal problem of just distribution across different age-groups can be reduced to the intrapersonal problem of the “prudential lifetime allocation” of healthcare resources (Reference Daniels1985, Reference Daniels1988). Problems of distribution of resources between age-groups can be addressed by regarding them as problems of distribution between life-stages.

In Daniels's view, we appeal to the prudential reasoning of hypothetical deliberators placed behind a veil of ignorance. The veil constrains the available information. Thus, the deliberators do not know their age, socioeconomic status and income, family situation, genetic history and health prospect, and their particular conception of the good. They do know, however, the age and disease profiles and the technological stage of their society. Daniels suggests that in this situation people would select institutions for the distribution of health resources which protect people's range of opportunities at each stage of life. The relevant notion of opportunity is age-relative: the range of opportunities and the particular opportunities which are important for people change as they move through different stages of life.

Behind the veil, people have no reason to assume that their lifespan differs from what is normal in their society. Thus, they have to balance the amount of resources which increases their chances of reaching the normal lifespan with the amount they save for interventions which extend their life once they have exceeded the normal lifespan. If the circumstances require, they may prudentially limit the resources available to marginal extensions of their lives. Thus, in some cases they may opt for rationing healthcare resources by age. In doing so, however, they do not value benefits differently merely because they come at different times in their lives. The prudential allocation of resources over a lifetime does not entail that later life years are discounted.Footnote ¹⁸

Perhaps one could argue for a stronger claim. Prudential deliberators behind the veil of ignorance will realize that the resources necessary to protect their age-relative opportunity range differ at various stages of life. This is partly because the opportunities typical for various life-stages are different, partly because people face different risks at different life-stages, and partly because maintaining and restoring a person's health at a level which is appropriate for her age-relative range of opportunities require different resource expenditures. No doubt the institutional arrangement prudential deliberators would select will reflect these facts. Were the decision makers behind the veil, however, to assign decreasing value to later years, they would distort the trade-offs they have to make between life-stages. Rather than confronting the problem of resource allocation across time head-on, they would prejudice their judgments in favour of earlier life-stages. Once the veil is lifted and they reach their old age, they would have to confront the fact that the difficulty of protecting their range of opportunity is exacerbated because they selected institutions which include age-weighting in addition to other forms of rationing.

Daniels would not disagree: on his view, neutrality regarding time is a requirement of prudence. With respect to the choice behind the veil of ignorance, it requires that we do not give lesser weight to benefits which we receive at later stages of our life. The deliberators behind the veil are concerned with their well-being over their whole life (Reference Daniels1988: 56–61). But even if prudence had no such requirement, the deliberators, it seems to me, would want to avoid distorting the trade-offs between stages of their lives by discounting later periods of life. Although a veil of ignorance argument does not exclude the possibility of an agreement on age-based rationing, it does not establish the case for weighting benefits by age.