In Canada, contingency planning for potential blood shortages has recently taken place. The National Plan for the Management of Shortages of Labile Blood Components (2009)Footnote 1 divides the availability of blood into three phases: green, amber, and red. In the green phase business is carried on as usual; in the amber, elective procedures are canceled; and in the red, life-and-death decisions between patients have to be made. The plan recommends that in the green phase guidelines should be written for what to do in the red, but it does not say how those guidelines should read. My aim in what follows is to fill this gap. To this end, I first develop a theory of allocative justice that derives from John Rawls’s A Theory of Justice. Footnote 2 I then use that theory to propose guidelines about how to allocate blood in the red phase.
A Rawlsian Theory of Allocative Justice
I say that the theory of justice I want to develop “derives from” John Rawls’s A Theory of Justice advisedly, because distributing scarce resources is not Rawls’s concern in that work. Rather, Rawls’s aim is to determine what principles should govern the basic institutions of society, specifically those relating to the liberties and rights of its citizens, and he distinguishes this conception of justice from another conception, which he calls “allocative justice.” Nonetheless, a theory of allocative justice can be extracted from Rawls’s account, and the machinery Rawls sets up is in many ways better suited to the latter than the former.
The bare outline of Rawls’s account of justice is as follows. Rawls analyzes justice as fairness, and the heart of his theory is a decision procedure to determine what is fair. This comes in the form of a thought experiment. We are to first suppose that there is a group of decisionmakers who are charged with selecting the principles governing the basic social institutions under which they will live. The thought experiment then proceeds by asking us to suppose two things: first, that the decisionmakers are self-interested in that their sole interest is to get the best that they can for themselves and those they care about and, second, that they are behind a “veil of ignorance.” The idea of the last, Rawls says, “is simply to make vivid to ourselves the restrictions that it seems reasonable to impose on arguments for principles of justice and therefore on these principles themselves.”Footnote 3 In particular, decisionmakers do not know if they have natural or social advantages or disadvantages, and thus things such as intelligence, talents, physical abilities, education, and social position are hidden from them. Nor do they know what their own inclinations, aspirations, and conception of the good are. Still again, they do not know what their particular situation in life is. They thus do not know their race, sex, and marital and financial status, or whether they have children or are sickly or healthy. Decisionmakers in this motivational and informational state are said to be in the original position, and the thought experiment ends by asking those decisionmakers (whom I will call prudent decisionmakers [PDMs]) to decide what principles of justice will define the society in which they will live. Rawls’s contention is that any principles selected under these conditions will be fair. In my extended application, they are to decide how to allocate goods in society, and my contention is that any policies that emerge will similarly be fair.
It is important to understand that this allocative theory applies only to questions about how to distribute goods where people have an equal claim on them, that is, public goods. It is not possible, as Rawls notes, to give an answer to the abstract question of how goods should be distributed among definite individuals, all of whom want them, for that will depend on how the goods came to be, who contributed what, what the expectations of contributing were, and so on. This restriction, however, leaves the theory with ample work for it to do, for questions of allocation that satisfy that limitation run throughout healthcare. Decisions have to be made about how much of a community’s budget should go to healthcare; how much of the healthcare budget should go to acute care, long-term care, and community care; how resources within those areas should be allocated to their component units; how units should translate those resources into patient services; and how tests and treatments should be rationed or waiting lists constructed when demand exceeds supply. It would be extremely useful to have a principled way of looking at these matters, and the theory of allocation sketched previously and developed subsequently is designed to provide just that. I here limit the application to the allocation of blood, which is the simplest kind of allocation problem in that blood is a single life-saving good. Allocation problems become more difficult when decisions have to be made about several goods, and difficulties are further multiplied when the goods have different opportunity costs, degrees of effectiveness, and durations of benefit and are a mix of life-saving and non-life-saving therapies. Nonetheless, the aim is the same, namely, to find a mode of distribution that would be chosen by PDMs, and thus the theory should likewise help with these more complex problems.
Once Rawls’s machinery is transferred to questions of allocation, two problems that bedevil his official application disappear. It has always been a problem for Rawls to say how PDMs, who are deprived of everything that makes them distinctive from one another, can construct a society that adequately takes into account the “plurality and distinctness” of individuals.Footnote 4 When we are distributing blood, however, there is no such difficulty, for now there is no difference in what the people to whom the policy applies want: they want blood, and as much of it as is necessary for life. Characterizing them as clones in the Original Position is thus exactly right. There can also be no controversy over the appropriate decision rule for PDMs to use. When the issue is the basic structure of society, there is a real and hard choice to be made between the maximin rule, in which decisionmakers aim to avoid the worst outcome, and standard decision theory, in which decisionmakers seek to maximize expected utility on the supposition that they could with equal probability be anyone.Footnote 5 But in the case of allocating blood (or healthcare generally), PDMs can have no motivation to do anything other than aim at a policy that will give everyone the best chance to minimize morbidity, mortality, and discomfort. Indeed, if we take the best outcome of an allocation policy to be the minimization of morbidity, mortality, and discomfort, and the worst to be anything less than this, there is no distinction between maximax and minimax strategies: to seek the best outcome is to avoid the worst, and vice versa. There is thus no reason to deviate from standard decision theory.Footnote 6
It would be a serious criticism of any allocation policy if anyone could indicate a mode of allocation that would be more attractive to PDMs. For then the policy must cater to special interests and thus could not be defended on grounds of justice, which is to say that it cannot be defended as a public policy at all. And conversely, if one agrees that PDMs would choose the policy, one cannot complain about its consequences, for such complaints could then likewise only be grounded in special interests, which again is to say that as a public policy it cannot be complained about at all. The theory thus provides us with a necessary and sufficient condition for the distribution of an important class of goods.
Furthermore, the theory brings together accounts that are normally thought to be in competition. Because allocations made on basis of what PDMs would choose will be those that give individuals the best chance of the best outcome under the constraint of impartiality, they will coincide with those dictated by utilitarianism. And if, as GibbardFootnote 7 has argued and seems plausible, any allocation policy that PDMs would choose will not be (as Scanlon puts it) reasonably rejectable, we have a happy point of coincidence between utilitarianism and the contractarian theories of Rawls and Scanlon.Footnote 8 I now turn to apply the theory to the case of allocating blood.
Allocating Blood
Any guideline about how to allocate blood in the red phase must address four questions:
1) What factors should be taken into account in deciding who should receive blood; that is, should the allocation be made on a first-come-first-served basis or are further criteria relevant, and if so, what are they and how can they be consolidated into judgments that would make it possible to compare and prioritize patients?
2) Do numbers count; that is, should blood be distributed to try to save as many lives as possible, and if so, is it better to give an optimal amount of blood to fewer people or a suboptimal (but still useful) amount to more?
3) Is there a morally significant difference between withholding and withdrawing blood; that is, if we can withhold blood from a patient to allocate it to benefit another patient more, or to benefit more patients, can we withdraw it to do the same?
4) Is there a morally significant difference between actual and potential patients; that is, can an actual patient who needs blood here and now ever be denied it to distribute that blood to potential patients?
Let us take these in order.
Question 1: Criteria
Our first order of business is to determine what criteria, if any, PDMs would want to take into account in allocating blood.
Social Value
One common candidate is social value. Whether this should be taken into account depends on what is meant by “social value.” The British Columbia Blood Contingency Plan states that “all patients in British Columbia should have equal access to the available blood on the basis of need.”Footnote 9 If “need” is understood to refer to the medical need of patients, this excludes social value in any sense. That, however, does not seem right. There may be circumstances in which the utility of having medical or other personnel such as firefighters, police, and government officials would be to everyone’s advantage. And if so, PDMs would give them priority in just those circumstances in which the risk of not getting life-sustaining treatment because it would be diverted to these people is outweighed by the risk of not having functioning people in those roles.
The standard items constituting social value discussed in the literature, however, do not typically have this importance. These items include things such as whether persons are married, have dependents, are employed (and if so, in what capacity), are involved in the community, and—as the Seattle ethics committee charged with allocating dialysis famously said—are members of a church or active in Scouts. A cursory glance at these kinds of factors shows that, even if we grant that they are all morally relevant, it would be extremely difficult to reliably attribute social value to individuals in a way that would allow for comparative judgments between them.
But even if the difficulties that stand in the way of doing this could be removed and suitable judgments of social value could be unproblematically arrived at, PDMs would not want to take them into account. If social value is construed in such a way that only a few (such as the rich, famous, and influential) have it, PDMs would not choose to have it taken into account because they would not likely benefit from that, and their chance of survival would be reduced by sometimes being in competition with those who have such value. Nor could PDMs want to have social value taken into account if it is construed so that most have it. For then they would typically be in competition with others who have social value, and so having it would not be generally advantageous, but lacking it would be typically disadvantageous. Finally, if social value is construed in a way that those who have it and those who do not are more or less equally divided, individuals would have a more or less equal chance of having an enhanced or diminished chance of getting blood, and thus PDMs would not have any incentive to take it into account.Footnote 10
Quality of Life
Even if we generally exclude social value from selection criteria, PDMs still would not want blood allocated solely on the basis of need. At least they would not if this means that all who will die without blood will get an equal chance of receiving it. To begin with, an individual may live in a state in which he or she is permanently not conscious, and PDMs would not want a guideline that gave blood to such patients at the cost of conscious patients. It is thus very plausible to say that blood will only be given to patients who will be able to experience the benefit.
This immediately raises the question of whether eligibility for blood should be limited by further quality-of-life considerations in the form of quality-adjusted life years (QALYs). According to this concept, different people’s years of expected life are given different value because of a difference in their quality of life. Thus the lives of those with severe mental or physical disabilities will be given less value than those of those without those disabilities. It is common for individuals to say that they would rather be dead than live in certain states, and it is hence plausible to suppose that there would be support for a policy that denied blood to such individuals if that would give others a better chance of life. PDMs, however, will have difficulty agreeing to this. What stands in the way is the fact that those with serious disabilities often rate the quality of their lives highly, sometimes as or more highly than those without such disabilities.Footnote 11 Prior to being in such a state, one may think that such a life is not worth living, but if the evidence is right, once in that state one very likely would not have that view. Thus for PDMs to make prudentially rational decisions, their preferences must be “corrected” by exposure to the evidence. And once that is done, they will arguably not choose to disadvantage the disabled to advantage those who are not. It would thus seem most agreeable to PDMs if guidelines for allocating blood in the red phase typically excluded patients in persistent vegetative states (or those who are otherwise uncontroversially permanently unaware of their selves or surroundings)Footnote 12 from the possibility of receiving blood but left other discriminations of quality of life out of the picture.
Quantity of Life: Survival to Discharge, Comorbid Conditions, and Age
There are further criteria that PDMs would want to take into account. Specifically, they would want a policy that gives priority to those who have a greater chance of survival to discharge, have fewer comorbid conditions that limit life extension, and are younger. The reason is straightforward. Although different people will benefit if these factors are taken into account than if they are not, no fewer will. Taking them into account, however, means that those who benefit will benefit more. Because PDMs suppose they could with equal probability be anyone, they are thus faced with the question of whether to choose a plan with greater or lesser benefits, when the probability of receiving those benefits is unchanged, and this choice is easy.
But even if it is appropriate to allocate blood on the basis of the aforementioned determinants, there is the problem of how to compare patients in terms of each. It is not always easy to determine which patients have better prospects to survive to discharge or to live longer beyond that, and not every difference in age can be plausibly taken into account. There is also the problem that these determinants can come into conflict. A patient with a low chance of surviving to discharge may have an excellent chance of life extension if she does; a patient with a high chance of surviving to discharge may have a diminished chance of further life because of comorbid conditions; a patient with a low chance of both survival to discharge and significant life extension beyond may, because of his or her youth, have a greater potential life expectancy than someone with a higher chance of actually living longer; and so on. We thus need a way to evaluate patients in terms of these determinants that would allow them to be compared with one another.
One way to do this is simply to leave it to clinical judgment, either of individual clinicians or a committee. Experienced clinicians, on the basis of that experience, will look and see, rate, judge, and decide. A second way is to articulate criteria in a point system such as the one developed by White, Katz, Luce, and Lo.Footnote 13 This system evaluates survival to discharge according to a SOFA (sequential organ failure assessment) scale (< 6, 6–9, 10–12, > 12), to expected life years in terms of comorbid conditions (none, minor, major, severe), and to age in terms of groups (12–40, 41–60, 61–74, 75+). Scores of 1–4 are assigned on each, the scores are added, and life-supporting treatment is given to those who do best. Random selection is used to break ties or impasses.
Point systems have inescapable difficulties. The scales used by White et al. have clear elements of arbitrariness to them, and it is impossible to replace them with other scales that do not. There is also the difficulty that, even on a common scale, different clinicians or committees may give the same patients different numerical scores in evaluating their survival to discharge or comorbid conditions. Nonetheless, point systems are significantly superior to prioritizing on the basis of simple clinical judgment, and it is hard to see what could be better. Uniform and transparent (and therefore defensible) prioritizing of patients requires an articulation of what the relevant criteria are, how they are weighted, and how they yield judgments in the end. Nothing other than a point system can provide this, and such systems have great utility. In many cases, once the criteria and weighting are made clear, clinicians will be able to agree without further calculations. When the appropriate prioritization is not clear after due consideration, numbers can sometimes make it so. We can also sometimes expect that interplay between clinical judgment and numbers may help clinicians arrive at a reflective equilibrium that will represent the best judgment that can be arrived at in the circumstances. Sometimes time or the availability of personnel does not allow for a point system to be used, and hence quicker decision procedures must be relied on (e.g., four-color triage or selection by one factor only, such as survival to discharge). But insofar as the criteria we have identified would be chosen by PDMs, it must be advantageous for patients to have them applied as rigorously as possible whenever possible.Footnote 14
Question 2: Numbers
It is common to suppose that numbers count, and that helping more people is better than helping fewer. The British Columbia Blood Contingency Plan frankly acknowledges this, writing that “when available resources are exceeded, the focus will shift from doing the best for the individual patient to the public health goal of doing the greatest good for the greatest number while balancing obligations to individuals and individual needs.”Footnote 15 It is, however, harder than one might think to explain why numbers should count.Footnote 16
We cannot say that they do if we think that human beings have infinite worth, for any multiple of infinity is not greater than infinity. Even when we set aside problems with transfinite numbers, there is still a difficulty in saying why numbers count. We may think that if three persons gain one unit of utility each, and one person gains one unit, we have three units versus one unit, and so numbers count because it is better to bring about three units than one. But we can arrive at the comparison of three versus one only if we sum the utilities of different people, and the problem is that no one experiences that sum.Footnote 17 Because no one experiences any greater utility than anyone else (so the argument goes), there is nothing to choose between helping more or fewer people. Nor is the reply that whereas no single person receives a greater benefit, more people receive the same benefit entirely satisfactory, for that does not amount to anything more than saying numbers count because numbers count and so does not explain why they do.
On these grounds, TaurekFootnote 18 has concluded that it is not (as he puts it) “objectively worse” for many people to suffer than a few. Thus decisions about giving blood to more versus fewer people should be made by a coin toss or some equivalent randomizing mechanism, the rationale being that because everyone has an equal claim on the blood, everyone should have an equal chance of receiving it. The Rawlsian theory of allocation, however, gives us a clear reason to take numbers into account. Faced with a choice between an equal benefit for more or fewer, PDMs will choose to benefit more on the ground that, because they could with equal probability be anyone, they will have a better chance of benefiting if more benefit.Footnote 19 This view seems better than Taurek’s. Not everyone will share the intuition that there is no objective difference between the sufferings of a few and of many, and for those people, distributing blood by random selection with indifference to numbers can hold no attraction. But we need not leave the issue at a clash of intuitions. Even if one accepts that there is no objective difference between the states in question, it is not clear why anyone should think that a lottery is a fairer mode of distribution than one in accordance with the preference of PDMs, which (we may remind ourselves) has a powerful claim to fairness. And even if that could be shown, it is not clear why fairness should trump utility so as to saddle us with a mode of distribution that does not give everyone the best chance of survival. It thus appears that numbers do count, just as almost everyone thinks, although not perhaps for any reason that one would naturally think of.
This immediately raises another question. If numbers count, would it be better to give optimal care to fewer people or suboptimal care to more? More specifically, in times of scarcity, would it be better to allocate blood so that everyone gets a lesser chance of receiving blood with a higher chance of survival if one does, or a higher chance of receiving blood with a lower chance of survival if one does? Healthcare professionals often bridle at the idea of doing anything less than the best for their patients. But if doing less for more will give patients the best chance of a good outcome, as it sometimes certainly will, PDMs must want that policy. And if so, there is no justification for healthcare providers to act on any conception of professionalism that dictates otherwise.
We cannot, however, stop at merely reducing the amount of blood given to suboptimal levels. The same logic implies that, in similar conditions of scarcity, it would be appropriate to also lower other standards. PDMs would thus endorse the recommendation of the British Columbia Blood Contingency Plan to use “expired” blood,Footnote 20 and that of the National Plan for the Management of Shortages of Labile Blood Components not to test blood for West Nile virus during the winter season.Footnote 21 And, as alternatives to doing nothing become bleaker, PDMs would presumably further lower standards to improve the odds of a good outcome, for example, to (as in some countries) accept transfusions from untested donors, and beyond that, to consider expedients not yet thought of. When being chased by lions, it is sometimes prudentially rational to jump into a river filled with crocodiles.
Question 3: Withholding and Withdrawing
I have argued that, in allocating blood, PDMs would give preference to those who have a greater life expectancy over those who have a lesser one, and to more people over fewer. The question now is whether these preferences remain when they can be satisfied by withdrawing and reallocating blood. This cannot be opposed by arguing that although physicians can sometimes withhold life-sustaining treatment, they can never withdraw it, for it is now widely acknowledged that they can. One may reply that physicians can withdraw life-sustaining treatment when patients request them to, or when the treatment is deemed not in the best interest of the patient or otherwise futile, but they can never do so for reasons of resource allocation. This, however, raises the question of why we should suppose they cannot, and two reasons are commonly given.Footnote 22 The first is that, by starting blood, physicians have made a commitment to the patient. The second is that the possibility of withdrawal will produce an unacceptable level of insecurity in patients and families. Let us now see how PDMs would evaluate these reasons.
It is not obvious that, by starting transfusion, healthcare providers have made a commitment to carry on until the treatment is refused or becomes not in the patient’s interest or futile. There is no explicit agreement to do this, and so the commitment must be implied. But not everyone understands that it has any such implication, and even if they did, that is something that can be revised. Blood could be given with the explicit understanding that it may be withdrawn under certain conditions, and so there need not be any misunderstanding about a commitment. Whether blood can be withdrawn and reallocated to secure a better outcome thus depends on the value PDMs place on the second element, security. No one can deny the importance of security. It is a terrible thing to withdraw transfusion services from a patient who is benefiting from them, and it is a terrible thing for such patients and their loved ones to live in fear that this might happen. The question is whether it is so terrible that it cannot be done in order to have a policy that gives all those needing blood the best chance of survival. (I here assume that estimates of life extension can be made that would make such predictions possible. Denying this would altogether eliminate the possibility of taking life extension into account in allocating resources.) We must also keep in mind that psychological factors count both ways. The fear stirred up by the possibility of withdrawal is accompanied by increased hope in those and the loved ones of those who were not initially chosen.
Given a choice between a policy that offered increased security to some and a better chance of survival for all, PDMs must choose the latter. In such a conflict, the value of a psychological state must give way to that of life. This choice, however, carries with it what Rawls aptly calls “strains of commitment.” What PDMs will choose is one thing; what they can actually live up to may be quite another. Perhaps the public (construed as PDMs faced with bearing the brunt of policies they have chosen) will accept the possible consequences of initially allocating blood with an eye to survival. Perhaps it will also accept the risks involved in withdrawal of care to improve individuals’ overall chances of a good outcome when the care is not life sustaining, as in aggressive discharge planning policies designed to optimize bed space.Footnote 23 But when the policy concerns the withdrawal and reallocation of life-sustaining blood, the strains of commitment may reach a breaking point. Patients and families may not willingly go along with the policy, and the reminder that they would have chosen it in the Original Position would then fall on deaf ears.
Anticipating this, PDMs could in Hobbesian fashion put in place enforcement mechanisms, as we do with so many other things in which there is a chance of backsliding, and Rawls acknowledges that this must be done.Footnote 24 It may, however, turn out that healthcare providers or the public will not have the stomach to carry out the enforcement. If so, we will have to forgo a first-best policy that would be favored by wholly prudentially rational beings in favor of a second-best one that will gain public acceptance. Because such a policy would be tailored for psychological factors or ungrounded moral intuitions, and because it is hard to predict what people will or will not accept on those bases, it is impossible to say how exactly that policy should read. Nonetheless, it is reasonable to suppose that it should approximate the first-best policy as closely as possible. Perhaps an acceptable policy would specify that such transfers can take place only when there is a very substantial difference between the two patients, or a very substantial number of people who will otherwise benefit. Or perhaps it would require in addition that the patient already receiving blood have relatively poor prospects of life extension, or be deteriorating. Other alternatives are possible, and point scales such as those discussed earlier could be used to give substance and precision to the proposals. What conditions the public will actually accept cannot be settled a priori by PDMs, but only by trying and seeing. It would, however, be surprising if no conditions could be found under which the public would accept such transfers, and the one that comes closest to what PDMs would choose will be best.
Question 4: Actual and Potential
Our final question is whether blood can ever be withheld from patients who need it now to put it to better use later. The answer is that sometimes it can be. PDMs must see themselves as possibly actual or potential patients, and thus the only question for them is where the maximum expected benefit would lie. Thus if it is certain that better or more patients would materialize, PDMs would say that the blood should be withheld. On the other hand, if future needs are less than certain, PDMs would decide on the basis of how probable the need and how great the expected goodwill. As these factors increase, the chance of PDMs benefiting from a policy of withholding the blood for future contingencies becomes greater, and as it does, the choice of that policy would become more attractive to them. Sometimes estimating expected good will be easy, as in disasters when the number and kind of incoming patients can be literally foreseen. Sometimes it will not be, as when centers have to rely on statistical estimates to gauge expected need. Given this, PDMs would favor a bird-in-the-hand approach, qualified by how good a grip there is on the bird and what they can expect to get from the bush.Footnote 25
Conclusion
I have argued that PDMs would hold that, in the red phase, the following conditions should apply:
1) Blood should be restricted to patients who are able to experience the benefit.
2) Absent overwhelming social utility, blood should be allocated to such patients solely on the basis of who has the best chance of maximum life extension, where this is calculated with the help of a point system by taking into account the probability of survival to discharge, comorbid conditions, and the age of the patient.
3) It is always better to give optimal care to more than to fewer and sometimes better to give suboptimal care to more than optimal care to fewer.
4) Blood allocated to patients should be withdrawn and reallocated to others in a way that will as much as possible maximize the life extension of as many patients as possible.
5) Blood should be allocated to actual or potential patients depending only on the probability and degree of benefit.
Some of these views will strike readers as intuitively right and others as not. But if the argument of the article is correct, both sorts flow from what PDMs would choose, and so we have a single source of justification for views we do not doubt but may have a hard time justifying, and reasons to hold views that we may initially doubt. And if I am further right that the guidelines that flow from this Rawlsian source coincide with those of utilitarianism and Scanlon, there is a powerful reason to accept those guidelines.
