FAIR EQUALITY OF OPPORTUNITY AND FAIR PROCESS

Norman Daniels’ theory of justice and health is based on John Rawls’ theory of justice as fairness.1 Rawls argues that a social contract among free and equal citizens would include three general principles of justice: a principle protecting equal basic liberties; a principle guaranteeing fair equality of opportunity; and a principle limiting inequalities to those that work to make the worst off as well as possible. Health and healthcare are not a topic for Rawls, as he assumes all members of society to be healthy.

In his new book entitled Just health: meeting health needs fairly, Daniels extends Rawls’ theory into the realm of health and healthcare by establishing two links between the three general principles of justice and health.2 First, the principle of fair equality of opportunity (FEO) grounds the special moral importance of health because health contributes to the range of opportunities open to us. If, as a matter of justice, we have social obligations to protect individual opportunity, promoting and restoring health is one component of fulfilling these obligations. A just society must provide the level of health services that is necessary to protect the normal opportunity range of its citizens, and thus FEO. Second, all three principles of justice allow distinguishing just from unjust health inequalities. Health is not only determined by access to health services, but also by socio-economic status. Important social determinants of health, such as education and income, are also classic objects of justice. This relation forms the basis for distinguishing between just and unjust health inequalities. If the distribution of the social determinants of health is consistent with the three principles of justice, any remaining health inequalities are just.

Although Rawls’ principles can guide our general thinking about justice and health, there is one problem. The principles, and in particular FEO which continues to be the primary focus of Just health, provide little practical guidance when resources for health services are especially scarce. Daniels reiterates that FEO is too general and indeterminate to solve real-world limit-setting problems in real time (pp24, 47, 103, 107).2 As a result, reasonable people disagree about how to meet health needs when not all can be met.

To illustrate this situation, consider a slightly embellished version of one of Daniels’ own limit-setting problems in an industrialised country (p135).2 A health provider contemplates covering a new high dose chemotherapy with stem cell support for advanced cancer. Preliminary data show that the intervention produces a two month gain in life expectancy in roughly 60% of patients with no other treatment options. However, the intervention is costly and consumes resources that otherwise could be used for other treatments. There is considerable disagreement about whether resources should be conserved for more effective interventions, or whether a last chance should be given to this group of desperate patients.

Daniels acknowledges that FEO is too indeterminate to resolve this “best outcomes/fair chances” problem as well as other limit-setting problems—notably the priorities and the aggregation problem—and that, in consequence, reasonable people disagree about how to set limits fairly. Daniels claims that a fair process still allows making a legitimate and fair limit-setting decision in situations of reasonable disagreement. He specifies the requirements of a fair process in the “accountability for reasonableness” framework (AFR). In this paper, I assess the claim that AFR results in fair limit-setting decisions (nothing will be said about AFR and legitimacy). I will show that Daniels is remarkably unclear about the link between AFR and fairness, and that once this link is clarified, the need for further specification and amendments.

ACCOUNTABILITY FOR REASONABLENESS

The central idea behind AFR—which is based on Daniels’ extensive previous work with James Sabin3^–6—is public accountability for limit-setting decisions that must operate within the range of reasonableness. AFR sets out four requirements for a fair process (pp118–19)2:

1. Publicity condition: both direct and indirect limit-setting decisions and their rationales must be publicly accessible.

2. Relevance condition: the rationales for limit-setting decisions must provide a reasonable explanation of how a health provider seeks to meet the varied health needs of a defined population under reasonable resource constraints.

3. Appeals and revisions condition: there must be mechanisms for challenge and dispute resolution regarding limit-setting decisions.

4. Regulative (enforcement) condition: there is either voluntary or public regulation of the process to ensure that the above conditions are met.

AFR is presented as a normative framework. When there is no agreement on sufficiently fine-grained distributive principles to resolve conflicting claims on resources, limit-setting decisions should be accepted as fair if they have been made via processes that satisfy all four conditions of AFR.

HOW DOES ACCOUNTABILITY FOR REASONABLENESS RESULT IN FAIR LIMIT-SETTING DECISIONS?

Daniels introduces AFR as a “classic appeal to procedural justice”(p109).2 However, there are several different notions of procedural justice. Each of them reflects a specific relation between procedure and just outcome—or fair outcome, respectively, when justice implies justice as fairness (Rawls 1971, pp83–90).1 So which of these notions of procedural justice applies to AFR? I argue that AFR is not a paradigm case of any of the classic notions of procedural justice and propose instead that what might be called “constrained pure procedural justice”i best captures how AFR results in fair limit-setting decisions.

Pure procedural justice

Daniels himself seems somewhat undecided about AFR’s appeal to procedural justice. His primary claim is that AFR invokes pure procedural justice. However, in an explanatory footnote he immediately curtails this claim by stating that AFR will be “… imperfect at best for anyone who insists that some particular principle tells us what the just outcome should be” (p109).2 Despite this ambivalence on Daniels’ part, I think it is fair to start by regarding AFR as an instance of pure procedural justice. Pure procedural justice implies two things. First, there is no criterion for the just outcome that can be justified independent of procedure; and second, we can agree on a procedure to the effect that any outcome resulting from this procedure is considered just. In the case of gambling, for example, the distribution of money is just when it has been determined by the roulette’s spin (provided the roulette has not been manipulated).

Does AFR appeal to pure procedural justice as Daniels initially claims? I think not. If the conditions of AFR were like the rules of roulette, they would have to be purely procedural (this is not the case). Moreover, limit-setting decisions would be fair as long as these procedural conditions were adhered to in the decision-making process. In the introductory limit-setting case, any coverage decision for the new cancer intervention would be fair provided it resulted from a process fulfilling the conditions of AFR—even if that decision violated the requirements of FEO. Clearly, this cannot be an outcome that Daniels would endorse. Here is why.

First, AFR contains a substantive requirement, the “relevance condition”. This condition indicates AFR places some substantive constraints on making limit-setting decisions.

Second, Daniels himself emphasises that his appeal to procedural justice supplements FEO, but does not abandon it (pp6, 27, 110, 251).2 Although this statement conflicts with his prior claim about AFR appealing to pure procedural justice, his adherence to FEO is not surprising and seems more true to his beliefs. Daniels makes a great effort to show that his opportunity based account of justice and health is compatible with alternative theories, such as the capabilities approach or a theory of egalitarian opportunity for welfare (pp64–77).2 And although he does not seek to systematically defend Rawls, Daniels claims his extension of justice as fairness greatly increases the power of Rawls’ theory (p47).2 Simply abandoning FEO later on would make these previous efforts of grounding health as an object of justice worthless.

Third, FEO is the normative foundation for Daniels’ theory of justice and health. His theory therefore provides a criterion of fairness that is independent of process and, as such, allows judging the outcomes of concrete processes. I will argue in more detail below that this does not require FEO to determine one and only one just outcome. However, all limit-setting decisions made in a just society must be, on Daniels’ account, at a minimum consistent with FEO.

Fourth, the idea of a fair process is introduced as a purely practical necessity. Daniels is not a sceptic. He never excludes that philosophical analysis can produce widely accepted and sufficiently fine-grained principles specifying what it means to treat people fairly (p25).2 FEO is too general and indeterminate and therefore a source of reasonable disagreement, but there may be more specific distributive principles in the future. We need a fair process primarily because philosophers will not be able to meet the real time need of decision makers (p108).2 Nothing in this analysis suggests that FEO is abandoned in favour of pure procedure.

Perfect and imperfect procedural justice

If limit-setting is constrained by FEO, we are left with the two other classic notions of procedural justice, namely perfect and imperfect procedural justice. Both of them make two assumptions. First, there is an independent criterion for the just outcome, such as FEO in the case of Daniels’ theory. Second, there is a feasible procedure that gives the just outcome. However, a procedure can do better or worse. In cases of perfect procedural justice, the feasible procedure always results in the just outcome. For example, when dividing a cake, we usually think this should be done equally. Announcing the smallest piece will go to the person cutting it ensures an equal division (at least if we assume this person wants as large a piece as she can get, is capable of cutting the cake equally, and so on). This outcome satisfies our independent criterion for what is considered just. In cases of imperfect procedural justice, the feasible procedure will give the just outcome in some, if not all, cases. For example, we strongly believe that only a guilty person should be convicted in a criminal trial. However, although we have this independent criterion for the just outcome, there is no infallible procedure giving it in each and every case. Sadly, adjudication sometimes leads to the conviction of an innocent person.

Now AFR could be regarded as an appeal to perfect or imperfect procedural justice, as Daniels himself concedes (p109).2 On this view, AFR prescribes the conditions of a process that results, more or less accurately, in fair limit-setting decisions. In the initial example, when considering the coverage of an expensive last chance chemotherapy, AFR would help to determine or at least approximate what fairness requires.

In this line of thinking, is AFR more like cutting a cake or more like investigating a crime? Whichever notion of procedural justice Daniels might decide on, perfect or imperfect, he requires a clearly defined independent criterion of fairness that allows determining the fair outcome of AFR in the given case. I argue in the following section that AFR’s substantive requirement, the relevance condition, is too indeterminate to fulfil this task.

What does the relevance condition mean?

The relevance condition constrains the kinds of reasons and principles that can justify limit-setting decisions in a fair process (pp123–31).2 It prescribes that limit-setting decisions should only appeal to reasons and principles that are accepted as relevant by “fair-minded” people. Therefore, the requirement that limits to health-related interventions should be based on relevant reasons is the independent criterion of fairness. But what exactly is a relevant reason?

Daniels provides three slightly different definitions of relevant reasons. The first definition is obviously circular: “… the reasons or rationales for important limit-setting decisions… must be ones that “fair-minded” people can agree are relevant to pursuing appropriate patient care under necessary resource constraints”, whereas “fair-minded people will seek reasons… they can accept as relevant to meeting health needs fairly under resource constraints” (pp117–18).2

The second definition is based on a different explication of the term “fair-minded people”. It states that relevant reasons are accepted by fair-minded people who are “trying to find ways of cooperating with each other on mutually acceptable terms” (p124).2 However, this explication proves unhelpful for specifying relevant reasons specifically in the situations of reasonable disagreement that trigger AFR.

Reasonable disagreement about how to allocate resources fairly is the starting point of AFR (p103).2 We need a fair process because we are facing a situation of reasonable disagreement, not a situation of mere disagreement. Daniels himself does not make this distinction, but it helps to show that the second definition of relevant reasons does not further specify the situations AFR aims to address.

Reasonableness is an important concept in Rawls’ philosophy, the basis of Daniels’ own work. According to Rawls, a basic aspect of reasonableness is the willingness to propose fair terms of cooperation and to abide by them provided that others do.7 However, if we need AFR because reasonable people disagree about how to make necessary limit-setting decisions, the second definition of relevant reasons is unhelpful. It merely tells us that, when we seek mutually justifiable terms of cooperation but disagree, we should accept as relevant reasons that provide the basis for mutually justifiable terms of cooperation.

The third definition of relevant reasons states that relevant reasons must pertain to meeting health needs fairly under resource constraints (p118).2 Meeting health needs “fairly” can of course have as many different meanings as there are conceptions of fairness. Without further specification, the third definition of relevant reasons is more or less open-ended. In the context of Daniels’ work, however, it seems plausible to assume that fairness implies consistency with the requirements of FEO, a reading of the relevance condition that has been advanced and criticised.8 Whether or not this reading is persuasive, consistency with FEO cannot offer the independent criterion of fairness we are looking for. If FEO is too indeterminate to solve limit-setting problems, any definition with reference to FEO will be equally indeterminate. The third definition of relevant reasons is either open-ended, or it is just as indeterminate as FEO.

A possible fourth definition of relevant reasons is not stated explicitly in the text, but it is nevertheless tangible throughout Just health. Daniels provides two negative examples of relevant reasons: religious reasons and mere disadvantage (pp113, 126).2 The exclusion of religious reasons, along with the emphasis on reasonable disagreement throughout the book, has a strong resemblance to Rawls’ idea of public reason.9 This suggests that relevant reasons might be similar to Rawls’ free-standing arguments, a reading that has been advanced and criticised as well.10 Whether or not this reading is plausible can be left open here. However, it is important for the present argument that construing relevant reasons as free-standing arguments will not offer the needed determinate criterion of fairness.

Imagine again the initial limit-setting problem regarding coverage for a new chemotherapy. Would we be able to identify the fair outcome if we knew the coverage decision should not be based on reasons that refer to comprehensive reasonable doctrines? It is likely that various reasons for and against coverage, such as giving desperate patients a last chance and considering the marginal cost-effectiveness of the new intervention, will be regarded as “free-standing” arguments. Each of them could tip the balance for or against coverage, so that more than one outcome, and even conflicting outcomes, would satisfy the relevance condition. This can be, of course, perfectly acceptable. But the point is that appealing to public reason makes the relevance condition, at best, as indeterminate as FEO. Defining relevant reasons as “free-standing” arguments equally fails to provide a determinate criterion of fairness.

What does this imply?

If AFR appealed to either perfect or imperfect procedural justice, it would require a clearly defined, independent criterion of fairness that allows determining the fair outcome of AFR in any given case. When dividing a cake or investigating a crime we know there is one and only one just outcome that a procedure should give: we should divide the cake equally and either convict the culprit or discharge the innocent. These two cases differ only in how accurately the procedure determines what we know to be just in the given situation. In contrast, AFR’s substantive requirement is not sufficiently specified to similarly define the fair outcome of limit-setting processes, and AFR does not contain other substantive requirements. The previous paragraphs have shown that the relevance condition is, at best, as indeterminate as FEO. This implies that AFR lacks the one, and only one, determinate and independent criterion of fairness that both perfect and imperfect procedural justice require.

CONSTRAINED PURE PROCEDURAL JUSTICE

FEO is not indeterminate in the sense that it does not provide any guidance whatsoever. The principle excludes several options in the limit-setting process, such as unjustified discrimination on the basis of gender or race. However, FEO, along with Rawls’ other two principles of justice, is consistent with many different “equally just” possibilities. Fulfilling Rawls' principles of justice is a “threshold test of justice”.11 The principles constrain the range of options, but they do not require one and only one course of action; justice as fairness will always allow for more than one just outcome. As Rawls points out, choosing among equally just possibilities, then, is not a matter of justice, but a matter of fair procedure (Rawls 1971, p201).1

Because Daniels repeatedly emphasises his adherence to FEO (pp6, 27, 110, 251),2 consistency with FEO appears to be the most plausible reading of the relevance condition. The relevance condition then amounts to a “threshold test of justice”, determining whether a proposed limit-setting decision is consistent with FEO. This makes AFR fundamentally different from spinning the roulette, cutting a cake or investigating a crime. The outcomes of AFR are not by definition fair, nor does AFR determine or approximate the one independently fair or just outcome. AFR thus is not a paradigm case of procedural justice—it calls for its own category of procedural justice.

I propose that AFR is an instance of what might be called “constrained pure procedural justice”. Its central idea is an appeal to fair process to choose from an independently constrained range of options. Constrained pure procedural justice, I suggest, has two essential features. First, there is an independent, but indeterminate, criterion of justice that is consistent with more than one possible outcome. Second, a pure procedure allows choosing fairly between these equally just outcomes. The justice, or fairness, of constrained pure procedural justice thus hinges on both the independent criterion of justice and the fairness of the process.

Applied to AFR, this yields the following analysis: the relevance condition defines the independent but indeterminate criterion of justice—that is, on my reading consistency with FEO. The publicity, appeals and revisions and the enforcement condition specify the conditions of the adopted pure procedure.

Obviously, here is not the place to scrutinise Daniels’ position on the link between FEO and health and healthcare. I do want to close, however, with some critical remarks about the procedural conditions of AFR. I argue in the following section that these conditions must be further specified and amended to better achieve formal fairness, inclusiveness and representation in limit-setting decision making.

IMPROVING THE PROCESS

Formal fairness

A process can promote consistent treatment of similar cases, or formal fairness. By appealing to a predetermined procedure whenever a limit-setting problem arises, it becomes more likely that similar cases will be treated similarly. Daniels himself points out that AFR helps meeting the formal requirements of fairness through the publicity condition (p122).2 This effect is probably amplified through the appeals and revisions condition: if limit-setting decisions must be made public and these decisions can be challenged, decision-makers presumably have an interest in treating similar cases similarly and justifying differential treatment—or at least inconsistencies are more likely to be caught. Furthermore, it is essential for formal fairness that the requirements of AFR cannot be changed or abandoned merely at the convenience of decision-makers. This is ensured by the regulative condition.

However, it is not clear that the requirements of AFR are sufficient to achieve formal fairness.ii Giving different weights to competing considerations can be consistent with FEO even if this results in conflicting limit-setting decisions—for example, granting or declining coverage for the chemotherapy. Daniels is right that different health providers might consider limit-setting differently and that their conflicting coverage decisions do not violate formal fairness, even when the cases are similar, if the limit-setting process was fair (pp135–7).2 But what if the same health provider weighed limit-setting considerations differently in similar cases? Imagine that a health provider would give priority to fair chances and cover the chemotherapy in the first half of the year, but in the second half of the year, the organisation would prioritise best outcomes and refrain from coverage. Both decisions seem perfectly consistent with FEO, but they violate formal fairness. It is doubtful that AFR’s publicity and appeals and revisions condition are sufficient to avoid similar scenarios, in particular when publicity does not require active communication. A consistency condition should require consistent, or at least relatively consistent, decision-making from individual health providers.

Inclusiveness and representation

If deciding among equally just options is a matter of fair process, not a matter of justice, inclusiveness and public representation in the limit-setting process seem to be crucial. However, the conditions of AFR must be further specified and amended to achieve fair consideration of everyone’s claims in the decision-making process.

First, the publicity condition only requires making limit-setting decisions and their rationales publicly accessible. This assumes that the public, or members of private health insurance schemes, are already aware of limit-setting. However, public awareness may not always exist, and if it does, people may be too passive to engage in information seeking behaviour. Active communication of limit-setting decisions, for example through newsletters, internet-based discussion platforms or public events, is needed to achieve effective publicity. (Effective publicity is also a precondition of the appeals and revisions condition, since people have to be cognisant of limit-setting decisions in order to appeal them).

Second, AFR is not inclusive and representative enough. The framework envisages that health providers make limit-setting decisions and provide reasons, while those subjected to such decisions can merely appeal for revision. Daniels insists that public involvement only serves to enrich the deliberative process and to disseminate limit-setting decisions; it has merely instrumental value (p129–31).2 The idea of public involvement as valuable in itself appears meaningless to Daniels because he believes democratic representational structures or proxy selection processes are “completely absent”.

However, this position is not compelling. Although mechanisms to achieve public representation are certainly limited, they are not completely absent in democratic societies. Moreover, the absence of a perfect procedure does not preclude using a feasible, imperfect one. For example, we do not abolish the enforcement of criminal law because this practice sometimes sadly involves convicting an innocent person. If the mechanisms of public representation are limited, this is a reason to be more creative about stakeholder involvement. Participation in debates about funding priorities, communication with political representatives, formation of political associations to lobby and advocate, and so on are ways to influence the decision-making process and to promote fair consideration of everyone’s claims.12 A public involvement condition should require establishing mechanisms to this effect.

Third, AFR’s voting procedure needs to be clearly specified. The relevance condition allows for voting when reasonable disagreement is intractable (p113).2 However, it does not specify the voting procedure and vaguely suggests that a “majority (or perhaps a super-majority) vote” should be decisive. This indeterminacy may ignore important distinctions between aspects of decision-making situations, such as the number of available options, that might call for different voting schemes. It may also give rise to arbitrary practices. More concrete guidance on the voting procedure is needed.

Fourth, those formulating and implementing limit-setting decisions should not be driven by their own interests to the extent that they become unable to balance competing claims.12 Although the relevance condition seems to exclude violations of FEO, it is still conceivable that major conflicts of interests cause biases when choosing among equally just options. An impartiality condition is needed to preclude this from happening.

CONCLUSION

Does AFR result in fair limit-setting decisions? The answer to this question is not straightforward, since Daniels is remarkably unclear about the criterion of fairness that AFR satisfies. I have argued, first, that AFR does not correspond to any of the classic notions of procedural justice; second, that “constrained pure procedural justice” best reflects how accountability for reasonableness results in fair limit-setting decisions; and third, that the procedural conditions of accountability for reasonableness must be further specified and amended to better achieve a fair process. A fundamental question remains whether FEO is the appropriate constraint for limit-setting in healthcare—I have said nothing about this. However, if Daniels wants to defend an integrated theory of justice and population health that is primarily based on FEO and also includes the idea of a fair process, some revisions of AFR seem necessary.