Inequalities in Health, Inequalities in Health Care:

Four Generations of Discussion about Justice and Cost-effectiveness Analysis

 

Since at least the 1980s, bioethics has addressed questions of justice in health policy. Much of the discussion has focused on health care, including the question of what role, if any, the techniques of cost-effectiveness analysis should play in the allocation of health care resources. In our view, this discussion has evolved through four generations of perspective and analysis, each asking different questions and seeking different solutions. In this paper, we use these four generations of commentary as a vehicle for exploring ethical questions about cost-effectiveness analysis, including some of the limitations of the standard arguments made against CEA based on its distributive implications. We then explore newer alternatives in CEA, including what is now called cost-value analysis, designed in response to these moral arguments. We also use the generational device to illustrate the importance of focusing on inequalities in health, and not merely health care, a view with early roots in "first generation" questions about a right to health.

I. First generation

In the first generation of commentary, the initial questions of interest were: Is there a moral right to health? Is there a moral right to health care? Despite the inclusion of language about a right to health in the UN Declaration of Human Rights, the bioethics literature did not much pursue the question of a moral right to health. Inequalities in health were viewed as unfair by at least some commentators, particularly when these inequalities were associated with poverty or environmental or occupational hazards. However, because health did not fit the model of a good or service that societies could in some strict sense distribute, a right to health was viewed as a conceptually confused and impractical objective. Instead, interest focused on whether there is a right to health care and, and if so, how to understand the nature of the entitlement grounded by the right. That is, to what medical goods and services are individuals entitled?

The answers to these questions followed predictable lines in moral theory. Libertarian, utilitarian and egalitarian views emerged and were never reconciled. In many respects, we have never fully left this original debate, as these divides continue to resonate in subsequent generations of discussion.

Libertarians rejected a moral right to health care. The job of justice, they argued, is to protect individual liberty and property rights, not to bring about some patterned conception of distribution. Inequalities in access to health care may be unfortunate, but not unfair, unless some intentional violation of anotherís liberty or property rights is responsible for the inequalities. At the same time, however, many libertarians made room within their theories for some role for the state (usually vouchers) in ensuring that the poor and unfortunate have access to a decent minimum of health care services (Engelhardt 1986; Lomasky 1981).

Commentators who argued from a family of egalitarian positions, by contrast, all defended an universal moral right to health care. None of them, however, provided accounts robust enough to adequately specify the content of that right.

Perhaps the strongest egalitarian view is one that understands justice as equality of welfare including health status. Robert Veatch, for example, argued that justice requires equality of objective net welfare over a lifetime for each individual, and that a "pragmatic derivative" of this principle favored a right to health care necessary "to provide an opportunity for a level of health equal as far as possible to the health of other people" (Veatch, 1981).

A more moderate, and more narrowly crafted, form of egalitarianism is that of Norman Daniels (Daniels 1985). He argued that all individuals are entitled to a decent minimum of health care services. His main argument rests on two claims. First, a robust commitment to equality of opportunity (construed in a very broad sense) recognizes that good health is specially strategic in realizing most species-typical life plans, and second, health care services are specially strategic in achieving improved health status.

A third, and still weaker version of egalitarianism simply holds that at least a part of what justice demands is some priority to the worst-off.

Some utilitarians also defended the view that there are good reasons, grounded in the interests of maximizing welfare generally, for societies to guarantee that all citizens have access to at least a decent minimum of health care. The problem for utilitarians was establishing how much societies should allocate to health care and how these resources should be spent. Some specifically looked outside of moral philosophy to cost-benefit and cost-effectiveness analysis as potential methods for answering these questions (Beauchamp 19 ).

Still another strategy offered by Allen Buchanan concluded that a legal entitlement to a decent minimum could be justified morally through a variety of arguments, including a principle of enforced beneficence (Buchanan 1984). His approach expressly rejected the need to identify a principled basis for specifying the content of the decent minimum and instead maintained that its identification was a matter of collective choice and fair procedure.

II. Second Generation

The first generation of commentary on justice and health policy concluded then with a near universal agreement that just societies should provide all its members with guaranteed access to at least a decent minimum of health care, regardless of whether this guaranteed access is accorded the status of a right. There was no agreement, however, as to how accounts of justice specify the scope of services to be included in the decent minimum, or even if they should. For those writers concerned to influence public debate and policy, there was a sense of frustration that this conclusion about a decent minimum, while welcome, was too anemic to be of much practical use. Accordingly, the litearture of the second generation focused on a central, and practical, question: how should priorities for health care be set within a budget constraint?

The dominant answer to this question came from health economics-- use cost-effectiveness analysis (CEA) to set health care priorities. The role of bioethics in this second generation of commentary was largely reactive and critical, pointing out moral objections, usually in justice and usually egalitarian in structure, to the use of CEA in health care.

That there would be egalitarian objections to CEA was inevitable. The preferred pattern of distribution in CEA is a function of two factors: benefits measured in terms of expected health gains and costs measured in terms of expenditure of economic resources. More "health-bang" for the buck is the sole aim. No independent distributive principles or side-constraints are inherent to the use of CEA approaches. Therefore, there are no restrictions on the morally permissible trade-offs that CEA can license.

It is beyond the scope of this paper to review all the moral criticisms that have been leveled against all the different variations of CEA that have been proposed. We focus here on only a few of the most prominent objections that have been raised against one variant of CEA, cost-utility analysis (CUA). In CUA, the utility or health benefit that is attributed to an intervention is quantified using an unified measure that attempts to incorporate impact on both length and quality of life. The Quality-adjusted Life Year (QALY) is standardly taken as the generic name for the class of these time-based measures that focus on judgments about the value of time spent in different health states. Because QALYs allow for the benefits of all health care interventions to be compared on a common metric, CUA can in theory be used to solve allocation issues for the total health budget of a country, a managed care organization, or other decision making entity charged with setting priorities for a wide variety of medical conditions, patient populations, and interventions.

We turn briefly now to a review of some of the central moral objections to the use of CUA and QALYs to set health care priorities. One is the charge of ageism (Harris, 1985). Because QALYs measure benefit in terms of longevity, QALYs disfavor the elderly who have fewer remaining years. However, not all who are disfavored are old. Young terminally ill persons will get low priority, and some older persons will fare well if their improvement in longevity or quality of life is considerable and their condition is relatively cheap to treat.

Another prominent objection to CUA and QALYs is that they discriminate against persons with disabilities. There is some truth in this claim but it must be carefully understood. QALYs do not always work to the systematic disadvantage of persons with disabilities. Persons with disabilities do well in CUA priority settings when they stand to derive significant improvement from interventions, especially interventions of modest cost. Thus it is more accurate to say that, at least in some cases, it is not disability itself that is disfavored by QALYS, but those with diminished capacity to benefit.

This is not the whole story, however. Imagine we have a treatment that can save lives but leaves some persons blind while no such adverse effect occurs for others. Do we give the people who will be blind a lower priority - because of a lower resultant quality of life - than we give those who will remain sighted? To do so would be to give lower priority to one set of patients just because of their disability. It is the disability itself that is the basis of lower priority, and for many this seems like a morally impermissible trade-off.

We can multiply examples of controversial trade-offs in CUA. Should severity of illness matter in priority setting? For example, should we permit unrestricted trade-offs such that more QALYs can be gained from capping teeth than from performing appendectomies? What about the so-called rule of rescue? Should we permit considerations of quality of life to trump life-saving interventions just because a great increase in quality of life for a non-life-threatening condition produces more QALYs than the small length of expected years saved by performing some life-saving interventions? If not, is there any increase in quality of life large enough to make the QALY bargain worthwhile?

Such puzzles have spawned a cottage industry of philosophical arguments for and against QALYs. Some urge us to bite the bullet and accept the maximizing principle at the heart of these tools of economic appraisal. Others attempt to justify QALYs on the basis of models of ex ante rational choice. These writers claim that, although the actual operation of CUA can have troubling distributive implications, it is rational ex ante, or behind some sort of veil of ignorance, to accept these unrestricted trade-offs.

While some have fought the good fight against using QALYs at all, others have tried to establish arguments to show what sorts of trade-offs are morally permissible and which are not. Francis Kamm, for example, has explored a series of novel cases to plumb the limits on trade-offs we would permit at the societal level based on what it would be reasonable to ask individuals to sacrifice (Kamm 1993). From those cases we get some guidance on the limits of acceptable trade-offs in, for example, situations in which rescue medicine is pitted against interventions that are highly efficient but of way less consequence to the individuals affected.

Although the commentary has been rich and insightful, at the end of this second generation of debate many see little progress in the traditional arguments of philosophers. The destructive project was successful-- the moral flaws of CUA were exposed-- but a constructive alternative to replace or augment CUA remained elusive.

III. Third Generation

The third generation of the debate, which is currently quite active, is characterized by deep doubt about resolving the question of how we ought to balance considerations of justice against aims of efficiency. The most modest proposal is to temper the efficiency-driven recommendations of CUA by relying on intuitive moral rankings, such as those generated by public discussion of the Oregon Medicaid program, to adjust for the methodís inegalitarian thrust. One suggestion offered by some pioneers of CUA argues for using CUA as only an aid to the deliberations of policy makers (Russell 1996). Policy makers are then expected to modify the results of CUA through a case by case consideration of the trade-offs that may offend. Thus, instead of basing decisions on a strict CE ratio ranking, what is often referred to as a "QALY league table", the suggestion is that policy makers adjust these rankings to reflect egalitarian and other moral concerns.

Other contributors, however, ask us to look well outside of CUA and the ethics commentary it has generated for guidance in the setting of health care priorities. Still others are working on strategies for empiricizing equity concerns and embedding them in mathematical CUA models. We briefly summarize three categories of approaches that have emerged in this area and offer some preliminary remarks about their prospects for success.

From Substantive Justice to Democratic Procedures: Some argue that we should leave decisions about the setting of health care priorities to the democratic process. The appeal of this suggestion is obvious. Those affected by the decisions ought to have their say, particularly when the limits of philosophical argument are reached. However, some who are attracted by a democratic procedural solution are reluctant to endorse whatever outcomes arise from simple majoritarian democratic processes. It is, after all, a commonplace observation that the majorityís raw preferences themselves may be unjust. Such preferences can run afoul of important distributive claims of minorities (ethnic, medical, or otherwise defined).

Some therefore favor an alternative account of democratic legitimacy, one in which the acceptability of decisions is conditioned on additional procedural and substantive constraints designed to ensure fair deliberative processes. Sometimes these constraints fall fairly comfortably within the procedural side of the procedural/substantive continuum. For example, in the context of priority setting within managed care, Daniels and Sabin argue for publicly accessible decisions and rationales, together with some mechanism for appeal and review (Daniels and Sabin 1997). The modesty of these constraints is evident. They do nothing to ensure the rightness of the rationales put forward, nor can they provide any assurance of substantively just outcomes.

Other, more ambitious suggestions are responsive to the worries about the inadequacy of purely procedural constraints. For example, one proposed constraint is that reasons be allowed to count only when other parties to the deliberation can accept them as relevant (Daniels 1998).

There are many problems with the more ambitious deliberative democracy approaches, but three are illustrative. The first problem is that there is a great risk of no solution emerging from deliberation because the depth of moral disagreement is so great. Some deliberative democrats acknowledge that possibility, but offer no assessment of how extensive the domain of indeterminacy is likely to be in a pluralist society. Even if all reasonable parties to the deliberation agree on the relevance of some reasons, there remain ample grounds for suspicion about how much agreement can be expected on their weight, and thus ample grounds for doubt whether the proposed constraints on reasons can solve the priority setting challenge.

The second problem is that the kind of reason that any individual party to the deliberation might reasonably accept is heavily theory-laden, or embedded in particular conceptions of justice. The objection is not merely the simple worry that deliberators will work from within competing single-principle conceptions of justice, such as libertarianism, utilitarianism, or some version of egalitarianism. This simple worry is worry enough. However, a further concern is that those who think that a variety of competing principles of justice may be relevant may disagree about which one is most applicable in a particular distributive context.

Finally, even if we suppose that democratic deliberation, suitably constrained, might confer some added measure of democratic legitimacy on decisions, which the armchair arguments of philosophers lack, that is not the end of the story. Will the public as a whole, or some representative sample, be designated to do the deliberation? If we rely on representative groups then we have problems in determining the criteria of representativeness. At best we merely mimic the decisions of majorities in much the same way some proponents of CEA and CUA claim that they are mimicking the outcomes of markets if markets functioned perfectly. By contrast, if such decisions could be left to actual democratic processes and not some simulacrum of democratic deliberation, we face the problem of extraordinarily high cognitive demands that would be placed on citizen-deliberators.

In all three cases, we are back at the same impasse that this democratic alternative to CEA was meant to circumvent. This is not to say that there is no independent moral value in deliberative democracy, or even to suggest (as some deliberative democrats argue) that progress in refining our conceptions of justice permanently eludes us. The problem is that when all is said and done, the requirement that deliberators offer public reasons will not give us a better handle on what counts as a good reason than we would get from moral philosophy. Our doubts are focused on what these new morally fortified conceptions of democracy can hope to deliver without coming back to the hard work to be done by substantive views of justice in determining which inequalities matter most in any given context.

Moralizing Preferences and Empiricizing Equity: Some adherents of the use of formal methods for setting health care priorities accept as telling the moral objections to CUA that were the focus of the second generation of commentary on justice and health policy. Bolstered by empirical data that suggest that many of these moral objections are consonant with the publicís preferences for how health care should be distributed, some go so far as to contend that the implications for public policy of CUA and QALYs are down right misleading (Nord 1999 ; Menzel 1999). Their solution to this problem is to explicitly introduce into CEA more fully moralized preferences, preferences that capture the societal values that appear to guide allocation decisions. In other words, their solution to the moral challenges to CUA raised in the second generation of commentary is to empiricize concerns about justice.

Nord advocates that CUA be entirely replaced by a new variant of CEA which he labels cost-value analysis. In cost-value analysis, the value or benefit part of the equation is no longer to be conceptualized in terms of units of health, as measured by QALYs. Instead, benefit or effectiveness is now to be conceptualized in units of social value that directly reflect the public=s preferences for interpersonal trade-offs.

The essential claim here is that CUA asks the wrong question. Like all proponents of formal methods, Nord and his colleagues assume that health care priorities should be set so as to satisfy public preferences. But public preferences for what? CUA elicits preferences of individuals for trade-offs between time and quality in their own lives, aggregates those preferences, and then uses the results as a basis for interpersonal trade-offs. While the techniques used in CUA such as the "standard gamble" and the "time trade off" give an idea of what individuals would want for themselves, they do not reveal what people would prefer as a principle for allocation. Thus, these techniques leave unanswered, and indeed unasked, what is seen by Nord, Menzel and others as the relevant question for establishing health care priorities-- the question of what the publicís preferences are for interpersonal trade-offs.

According to this view, it is because CUA has asked the wrong question that the results of some CUA analyses have been so morally counter-intuitive. Nord and his colleagues cite studies that suggest that, when respondents are asked directly about interpersonal comparisons using a technique called the "person trade-off", their preferences are markedly different from those elicited in traditional CUA surveys. Many respondents reject the pure health maximizing model embedded in CUA in favor of some weighted priority for the severely ill, even if - and this is crucial - the severely ill have reduced capacity to benefit from treatment and even if they cost more to treat.

It is in this sense that Nord claims that CUA is misleading. If decision makers are presented with quantitative empirical information about intra personal preferences, and if only intuitive justice considerations are factored in, equity concerns regarding the limits of permissible interpersonal trade-offs will be overwhelmed by the seemingly precise, quantitative results of CUA. If, however, CEA is based on distributive preferences rather than preferences for health states, these justice considerations will be quantified and embedded in the CEA results, and thereby protected in the public policy process.

The empirical techniques for assessing distributive preferences are still in their infancy, as is commentary on what role, if any, such preferences should play in the formulation of public policy. There are many unanswered questions. What would it mean to assess the validity of preferences so obtained, to be able to say that the numerical representations of societal preferences are in some sense accurate? By what criterion can we establish that the preferences as measured and summarized constitute a predominant or even a majority preference in the society generally? How valid are the preferences, even as measures of individual values? Person trade-off judgments, like all social psychological measures, are sensitive to how questions are framed. Framing effects are conventionally viewed as measurement problems that admit of methodological solutions. But this is only partly correct. If respondents are systematically more likely to express an "utilitarian" preference in one frame than in another, which preference is the accurate one? If respondents are confronted with this finding and asked to reconcile the apparent inconsistency in their judgments, is the resulting answer more accurate? If so, is this because their answers now represent more settled, more reflective judgments? But how can we know when a judgment is sufficiently reflective or sufficiently informed?

Even if these conceptual and methodological problems can be successfully addressed, some potentially powerful moral objections remain.

Aggregate statistics can mask deep moral and political divisions. Consider a study conducted by Nord in which Norwegian politicians were asked to give their preferences for resource allocation in five different contexts. In each case, the respondents had to chose between two different groups of patients that varied by either the magnitude of benefit each group would receive, how sick the two groups were, or the ages of the two groups of patients. In each context, choosing one group could be interpreted as the QALY maximizing or "utilitarian" preference. The utilitarian option was rejected by a sizable majority of politicians in every case, causing Nord to title the article in which these data are reported "Health politicians do not wish to maximize health benefits." (Nord 1999, p. 150). However, an interesting, but not altogether surprising, finding emerged when the responses were analyzed by whether the politicians were conservatives or social democrats. Conservatives were systematically more likely than social democrats to express an utilitarian preference. Indeed, in two of the contexts, as much as half of the conservatives made the utilitarian choice. Nordís response to this finding is to call for the assessment of preferences in samples large enough to avoid "political biases". For him, moral and political differences are seemingly methodological problems, to be ironed out as the person trade-off technique for measuring societal preferences matures. Arguably, however, what Nord dismisses as bias is in fact a moral and political disagreement that needs to be featured in public debate, rather than buried in large samples and summary statistics.

What we have in these data is, in effect, the same philosophical impasse that moral theorists, and perhaps deliberative democrats, have reached.

Yet another problem with attempts to empiricize equity concerns through the person trade-off method is that in the person trade-off context the respondent is working with hypothetical potential patients who have no particular history or life circumstances apart from having a specified medical condition. This sort of abstraction involves a significant loss of moral information. The information loss is a double edged sword, however.

The up-side is that it facilitates an impartial assessment. It protects against biases that work against the interests of marginalized groups. The down side is that it masks from public consideration features of a situation that may be morally relevant in making just allocation decisions. What moral difference would it make, for example, if the burden of disease is considerably greater for some ethnic or other historically disadvantaged groups? Or for some persons with biographies that differ in their occupational or environmental risk exposures?

Arguably, some differences among the burdens borne ought to translate into differences in the benefits due under a more contextualized account of justice than cost-value analysis, as presently conducted, permits. Otherwise, we would seem to be committed to a view that leaves out reasonable principles of rectificatory justice for historically disadvantaged groups and compensatory principles giving priority to those whose sacrifice is great, especially if unfairly imposed upon them. We think such principles should figure in a fuller account of justice in health care allocation. Thus, there is a risk that these person trade-off attempts to empiricize equity concerns may themselves mislead with respect to considerations of justice because of the attendant loss of moral information.

Perhaps such matters could be addressed by eliciting other, more contextualized preferences, ones designed to get at these added concerns about rectifying and compensating for injustices that contribute to health status inequalities. However, such concerns point to a family of cases in which those most vulnerable to bias may come out on the short end of the moral stick. If so, once again, we risk being at an impasse no less unresolvable than those that plague moral philosophers.

Christopher Murray and colleagues, the developers of a variant of QALYS called Disability Adjusted Life Years or DALYs, suggest that at least some of these issues could be finessed by moving away from methods that aggregate the distributive preferences of individuals towards methods that involve some deliberative group process for establishing societal preferences (Murray and Acharya 1997). Murray attempts to provide a moral rationale for the choices incorporated in DALYs through what he calls the principle of "filtered consensus". This principle privileges values that are widely shared by many people after deliberation and discussion. If the reasons behind these widely shared, robust values are not in conflict with "ideal-regarding principles" of morality, they can be incorporated into the construction of DALYS.

Elements of Murrayís principle of filtered consensus are evident in the protocol used to create disability weights. The protocol requires respondents to compare different health states using two variants of the person trade-off method. Respondents are confronted with any differences in comparisons that may result from these variations, as well as with the implications of their choices for social policy. Respondents also share their rankings with fellow participants, as well as their reasons for choosing as they did, in group discussion. Person trade-off assessments are thus continuously revised through a process of private refection and group deliberation that can last from eight to ten hours. The ultimate goal is to achieve a consensus.

The developers of DALYs have embarked on a project that aims in effect to integrate two of the three current responses to moral objections to CUA-- deliberative democracy theory and the empiricizing of equity concerns. Both Murray and Nord suggest that it is an empirical matter which method for eliciting distributive preferences-- the individual or the group-- provides more valid data. This response seems to us to miss the central point. Although we view the group deliberative process as having more methodological and moral appeal, it is unlikely that either method will satisfactorily address all the relevant moral concerns. At very least, the DALY project pushes us back into a more careful exploration of how much it is reasonable to expect from the constraints of deliberative democracy theory that surpass the work of good moral philosophical argument.

The Libertarian reemergence: According to the libertarian, the reason the priority-setting project has floundered is because it asks the wrong question. Because priority setting should be a matter of individual choice among health plans or insurance benefit packages in the market place, there is no need for social agreement on the elements of a decent minimum. Such decisions ought not be the province of government bureaucrats or dependent upon the collectively arrived at and collectively binding judgments reached through democratic processes.

The rationales for deference to individual choice are numerous. Respect for individual autonomy ranks high among the reasons for all libertarians. In addition, however, reliance upon individual market choice is claimed to promote greater efficiency, make individuals more accountable for lifestyle choices associated with premature mortality, and avoid the moral conundrums that inevitably draw centralized decision makers into debates about morally contested services such as abortion, infertility treatments, and end of life care.

Most of the arguments of an earlier generation of libertarians are reproduced in this latest wave of market-oriented (typically voucher-based) solutions to the priorities problem. However, several new arguments tap into concerns that egalitarians share. One claim is that health care policies should be sensitive to the way in which access to health care figures into an overall strategy designed to reduce health and other social inequalities. Since far more than health care affects health status and overall life prospects, the freedom to purchase "bare bones" benefits packages and use the savings for such goods as better nutrition or housing in safer neighborhoods may do more to improve the health and well-being of low income persons than a richer health care benefit package (Goodman and Musgrave, 1992).

Moreover, higher income persons make greater use of the guaranteed health care services and are more likely to persuade decision makers of the medical necessity of utilizing a broader range of costly diagnostic and therapeutic interventions. The unintended effect of minimum benefits requirements may what some call "middle class capture" (Shapiro, 1998). The risk is that the poor subsidize the preferences of the better off, while losing their own opportunity to use their limited resources in alternative ways that might do more to improve their own health status.

Although the libertarian challenge reveals the extent to which it is a contingent matter whether discrete moral aims, say, those prized by libertarians and those most valued by egalitarians, can be satisfied simultaneously through the careful crafting of social policy, the major limitation, as we see it, is that it swims against the current of economic theory with respect to health care. Many economists have long regarded health care and health insurance markets as unlikely to conform to the expectations of Neo-Classical economic theory for the reason that even very well informed consumers make choices under conditions of severe informational asymmetry. The need to make informed choices in advance of medical need, and in the face of substantial uncertainty about risks to health, undermines the standard market expectation that individuals acting in their own best interests can make the kind of prudentially rational choices traditional market theory supposes (Arrow 1963). The suggestion that a fortuitous confluence of expanded choice and reduction of inequality in health will result from individuals attempting to set the totality of their health-affecting priorities (not just their health care priorities) adds an additional layer of implausibility to an already overly romantic vision of human capacities.

IV. The Fourth Generation

The third generation debate about alternative strategies for addressing justice and efficiency in the allocation of health care resources is in many respects just getting underway. At the same time, however, a fourth generation of commentary also is beginning in which the spotlight is on health, as opposed to health care. While the first generation of commentary quickly abandoned any serious pursuit of a right to health in favor of a right to health care, this fourth generation promises to make inequalities in health, and not inequalities in health care, the central moral concern.

In many respects, the literature on justice and health policy is just waking up to what has been recognized elsewhere in health policy and public health for a very long time. There is substantial inequality in life expectancy and health-related quality of life between the rich and poor nations of the world, as well as between the rich and poor within nations. There also is considerable evidence for what is referred to as the "social gradient in health", suggesting that this differential in health status is not merely between the ends of the social continuom-- the rich and poor-- but exists in relative terms across the spectrum of social standing.

Of particular relevance to the right to health care debate is a related finding about the limited role of access to care in explaining this social gradient. An impressive body of empirical literature suggests that medical care has quite limited consequence for aggregate health status and thus for narrowing the gap in health inequalities in comparison to other policies that affect education, income, wealth and social status and provide for public health. Some studies suggest further that what is at issue with respect to the health of a society is the degree of socioeconomic inequality within that society, and not merely the proportion of the society that is poor (Kawachi, Kennedy and Wilkinson 1999).

While this last claim in particular remains controversial, there is little disagreement about other features of the empirical story about health inequalities. If, as we and others contend, the central moral concern in health policy is that inequalities in health, both globally and within the United States, are too high, than a number of issues arise.

The claim that health care is special in virtue of its special strategic role in reducing inequalities in health status, which in turn is specially strategic to equal opportunity in life prospects, is undermined. Insofar as the argument for a right to health care is based on equality of opportunity, health care no longer seems especially targeted as a candidate for being a right. For those for whom this egalitarian appeal was central, an alternative moral foundation for a right to health care must be found. Elsewhere we have argued that there are good moral reasons for ensuring universal access to health care, even if access to health care has little impact on aggregate health outcomes and thus is likely to have little effect on health inequalities (Faden and Powers 1999).

If, however, the main moral reasons for valuing health care and securing a universal entitlement to it have little to do with aggregate health outcomes, then the entire rationale for relying on CUA strategies to set health care priorities is undermined. If health care is important for reasons that have little to do with improving the health of populations, then why allocate health care resources according to a formula intended to maximize the amount of "health bang for the buck"?

But what if the nature of medical care changes such that the conventional wisdom about the limited significance of health care in equalizing health status also changes? Many are predicting big changes in what medicine can do, and if they are right, both the health maximizing argument and the argument that health care has special strategic importance as a vehicle of ensuring equal opportunity may find new philosophical life.

At least for now, however, we and others maintain that the importance of reducing health inequalities calls for a more comprehensive approach that, in some instances, can put social policies other than health care access in a position of higher moral priority. This also raises the stakes for what sort of moral theory resides behind our egalitarian concerns. Just what inequalities ought we seek to reduce or prevent? Arguably, some inequalities will matter more than others, especially those that tend to sustain, compound, and reinforce inequalities in other aspects of life.

A related issue is whether CUA approaches should be used to guide priority setting in other health and indirectly health-enhancing policies. What would be the cumulative effect of doing so?

This consideration introduces a nest of other questions that depend heavily on the pervasiveness of contingency that infects all maximizing approaches. As many commentators on the distributive effects of utilitarian moral theory have noted, it is an open question whether maximizing principles, such as the one at the heart of CUA (e.g. maximization of health utility), tend toward greater or lesser inequalities. Is it more likely that those who are worst off with respect to health (or other aspects of welfare) will get priority with CUA, and therefore that allocation policies based on CUA will reduce inequalities?

One answer says yes because of the principle of declining marginal utility. As some utilitarians argue more generally, dedicating resources to those whose well-being is at the lowest level tends to produce more utility than devoting those resources to persons whose utility, health or otherwise, is already high. If this turns out to be the case, we may have a confluence of equity and efficiency, a consequence that may in the long run be of more moral significance than the issues of permissible trade-offs that have occupied the attention of most commentators on CUA in the second and third generations of these debates.

However, it is premature to predict this happy coincidence. For the nature of the contingencies that determine the actual distributive implications of a health maximizing approach can also lead us to benefit more those who already have much. Investing resources in those better positioned to take advantage of life-extending and quality enhancing care might be the policy that maximizes aggregate health, but in the process, it might widen the gap between the healthy and the less healthy.

More likely perhaps, some mixture of both possibilities will occur. If so, egalitarians may want to be health maximizers when such policies reduce inequalities, but support restrictions on maximizing for policies where inequalities would be exacerbated.

Conclusion

In previous generations of commentary, we have expended considerable energy trying to find the magic bullet for deciding whether and when to maximize. One reason that project has been so unsuccessful is the failure to see the contingent relationship between efficiency and equality. Sometimes it is possible to have it both ways, but sometimes it is not. A similar contingent relation holds between expanded freedom of choice and reduction of inequality.

Whether we look to CUA, deliberative democracy, CVA, or the arguments of moral philosophers, we reach two overarching conclusions. First, an appeal to the substantive elements of justice, most visible in the first generation, have never left the scene, and it is a mistake to think that an easy exit from philosophical reflection can be achieved. Second, the main difference between the first generation and the present discussion is that the prospects for a more successful role for substantive accounts of justice depend upon abandoning the search for a single principle of justice appropriate for distribution of one good in isolation from a larger context in which other sources of inequality intersect.

Arrow, K., "Uncertainty and the Economics of Medical Care", American Economic Review 53 (1963), 941-73.

Beauchamp, T.,

Buchanan, A., "The Right to a Decent Minimum of Health Care," Philosophy and Public Affairs 13, 1984, 55-78.

Daniels, N., Just Health Care (New York: Oxford University Press, 1985).

Daniels, N., and Sabin, J. E., "Last Chance Therapies and Managed Care: Pluralism, Fair Procedures, and Legitimacy", Hastings Center Report (1998), 27-41.

Daniels, "Distributive Justice and the Use of Summary Measures of Population Health Services," in Institute of Medicine Committee on Summary Measures of Population Health, Summarizing Population Health: Directions for the Development and Application of Population Metrics, Washington, DC: National Academy Press, 1998, 58-71.

Engelhardt, H. T., The Foundations of Bioethics, NY: Oxford University Press, 1986.

Faden, R. and Powers, M., "Incrementalism: Ethical Implications of Policy Choices", Washington, DC: Henry J. Kaiser Family Foundation, 1999.

Goodman J., and Musgrave, L., Patient Power: Solving Americaís Health Care Crisis, Washington, DC: The Cato Institute, 1992.

Harris, J. R., The Value of Life, (London: Routledge and Kegan Paul, 1985).

Kamm, F. M., Morality, Mortality: Death and Whom to Save From It (New York: Oxford UP, 1993).

Kawachi, I., Kennedy, B., and Wilkinson, R., The Society and Population Raeder: Income Inequality and Health, NY: The New Press, 1999.

Lomasky, L., "Medical Progress and National Health Care," Philosophy and Public Affairs 10, 1981.

Nord, E., Cost-Value Analysis in Health Care, Cambridge University Press, 1999.

Menzel, P., "How Should What Economists call "Social Values" be Measured?", The Journal of Ethics 3, 1999, 249-273.

Murray, C., and Acharya, A, "Understanding DALYs", Journal of Health Economics 16 (1997), 703-730.

Russell, L. B., et. al.,"The Role of Cost-Effectiveness Analysis in Health and Medicine", Journal of the American Medical Association 276 (1996), 1172-7.

Shapiro, D., " Why Even Egalitarians Should Favor Market Health Insurance," Social Philosophy and Policy, 15, (1998), pp 84-132.

Veatch, R., A Theory of Medical Ethics, NY: Oxford University Press, 1981