Madison Powers and Ruth Faden

The Kennedy Institute of Ethics

Georgetown University

Recent health services research literature has called attention to the existence of a variety of disparities in the health services received by racial and ethnic minorities. As well, racial and ethnic disparities in health outcomes from various health services, including screening, diagnosis, and treatment for specific diseases or medical conditions have also been noted. Such findings provide the impetus for the consideration of two primary moral questions in this paper. First, when do ethnic and racial disparities in the receipt of health services matter morally? Second, when do racial and ethnic disparities in health outcomes among patient groups matter morally?

Our approach in answering these questions takes the form of two theses. Our first thesis, the neutrality thesis, is that disparities in health outcomes among patient groups with presumptively similar medical conditions should trigger moral scrutiny. Our second thesis, the anti-discrimination thesis, is that disparities in receipt of health care or adverse health outcomes among racial, ethnic or other disadvantaged patient groups should trigger heightened moral scrutiny. The theses are presented as lenses through which the morally salient features of health services can be viewed. Most theories of justice can accept some version of both the neutrality thesis and the anti-discrimination thesis. However, as we shall see, these theories differ in the nature and strength of their moral conclusions and in the reasoning they employ in reaching those conclusions.

The bulk of this paper will focus on the foundations of the theses, their relation to competing accounts of justice, and the considerations relevant to their moral analysis. In Section II, we articulate the moral foundations for the neutrality and anti-discrimination theses, and in Section III, we examine some potentially morally relevant considerations that inform the conclusions from the perspectives of alternative theoretical frameworks. Finally, in Section IV, we consider the moral implications of these findings for physicians and other health care providers.

The preliminary task, however, is to clarify several conceptual issues lurking in the formulation of the theses. Although the theses overlap in certain important respects, it is even more important to be clear about how they differ.

The first conceptual distinction has to do with who is covered under the thesis. The neutrality thesis covers disparities in health outcomes among any patient groups with presumptively similar medical conditions and prognoses. By contrast, the anti-discrimination thesis refers specifically to a subset of what falls under the neutrality thesis–the special case in which the outcome disparities involve racial, ethnic or other disadvantaged patient groups.

The second conceptual distinction has to do with what is covered. The neutrality thesis covers only disparities in health outcomes. But the anti-discrimination thesis, which specifies that the disparity must occur in a disadvantaged social group, means that disparities in the health care services people receive, and not just the outcomes they experience, also matter.

The neutrality thesis is thus intended to cover any instance in which it is established that there are differences in outcomes among patient groups that are in relevant respects otherwise medically similar. If it was determined, for example, that white men with colon cancer had poorer survival rates than African- American men with colon cancer, then the neutrality thesis should trigger the same moral scrutiny as if the situation was reversed. In addition, this claim would hold even if it was clear that there were no differences in the medical services the two groups received. However, what if it was determined that white men were less likely than African- American men to have screening colonoscopies after age 50? As long as this disparity did not result in different medical outcomes, there are no moral implications under the neutrality thesis.

In contrast, the anti-discrimination thesis assumes that disparities in both health services received and disparities in health outcomes are independent and distinct reasons for moral concern when the disparities disfavor racial and ethnic groups. These groups are “morally suspect categories,” understood here as analogous to legally suspect categories in equal protection law. Under the anti-discrimination thesis, either type of disparity-- alone or in combination is treated as morally problematic as long as the disparity disfavors a morally suspect group. This is markedly different from the neutrality thesis, in which disparities in utilization are only problematic if they have a disparate impact on health outcomes.

Underlying the neutrality thesis is the implicit assumption that the moral value of medical interventions is generally instrumental. In other words, whether it is good or bad to receive or fail to receive-- a medical intervention depends on the impact each option would have on individual health and well-being. . In the case of racial and ethnic minorities, however, a different moral value is at stake. The very fact that a minority population might receive fewer services believed to be beneficial suggests the potential for morally culpable discrimination. This is a significant moral concern in its own right, regardless of the medical consequences. Under the anti-discrimination thesis, disparities of either sort trigger an additional or heightened level of moral scrutiny beyond that warranted by health outcomes disparities generally. i

Thus far, we have merely articulated some of the implications of and analytic differences between the two theses and the implications of the differing forms of moral judgment that can flow from the use of either moral lens. In this section, we offer a philosophical defense of the two theses and link them to the more general theoretical foundations on which they rest.

A principle that has come to be known as the formal principle of equality is often the starting point for discussions as to when some sort of disparity or inequality in the way persons are treated (in a more general sense than meant in health care contexts) is morally problematic. It is a minimal conception of equality attributed to Aristotle, who argued that persons ought to be treated equally unless they differ in virtue of some morally relevant attributes. It is, of course, critical to determine in any particular context just which attributes are morally relevant and which are not. Often these determinations are matters of disagreement and controversy that can be traced to significant differences in rival theories of justice. The degree of agreement across theories of justice on the matters under discussion in this paper is, therefore, surprising.

Consider first a type of theory of justice many would think least likely to agree with either the neutrality thesis or the anti-discrimination thesis. The libertarian theorist rejects any pattern of distribution as the proper aim of justice, arguing instead that whatever pattern of distribution emerges from un-coerced contracts and agreements is morally justified (Nozick, 1974). Moreover, coercive attempts by the state to enforce a preferred pattern of distribution are themselves viewed as unjust. To the libertarian, inequalities are counted as merely unfortunate and not unjust, unless they are the product of some intentional harm or injury.

Initially, one might think that the libertarian position leaves little room for objecting to disparities in health outcomes among patient groups, whether defined along racial lines or otherwise, or to disparities in the receipt of health services among racial and ethnic groups. As long as patient preferences are not overridden and no harm to those patients was intended, no injustice or other moral failing would obtain. Indeed, it seems highly unlikely that the libertarian could accept the neutrality thesis, failing to see any basis for demanding moral scrutiny merely because some patient groups fare less well than other patient groups.

The libertarian conclusion may well be different, however, when, as contemplated by the anti-discrimination thesis, the patient groups involve morally suspect categories. Some conceptual room is left open for endorsement of the anti-discrimination thesis, and that room is a consequence of the limited domain of moral judgment for which the libertarian theory is meant to apply. The libertarian view is primarily a theory of societal obligation, or what society collectively owes its members, and not a comprehensive moral doctrine spelling out the full range of individual or other non-governmental moral obligations. Libertarians often assert that particular individuals have duties of mutual aid, even fairly stringent ones, even though state coercion to enforce them would be unjust (Engelhardt, 1996), as do certain non-governmental institutions and professional bodies that assume certain social functions as part of their self-defined moral missions. Thus, even in the libertarian view, the failure of individuals and institutions to offer health services to all racial groups on an equal basis can be a significant basis for moral condemnation.

A point of particular significance for this discussion is that nothing in the libertarian view necessarily excludes the existence of parallel moral obligations that are rolespecific, such as those ordinarily obtaining between physician and patient. Such special obligations are often referred to as agent-relative obligations. Some libertarians have argued that because of the existence of these agent-relative obligations, which in their view form the core of our moral requirements, coercive state action is morally condemnable. Such interference is said to be morally condemnable insofar as it may interfere with an individual's most basic agent-relative moral duties (Mack, 1991). The libertarian, therefore, may limit what government may do to enforce cer tain individual moral obligations, but it does not purport to be a comprehensive moral doctrine that effaces those individual obligations.

The upshot is that the libertarian view, even in its strictest form, need not reject a thesis asserting that disparities involving racial and ethnic minorities should trigger special moral scrutiny. However, libertarians will locate their judgment of moral failing in the failure of specific individuals or institutions to discharge their moral duties, not in the society at large. Nor would the libertarian necessarily see the moral problem as a failure of government to enforce neutrality in the receipt of care or achievement of the outcomes that specific individuals and institutions are properly committed to achieving.

In sum, even libertarianism, the theory of justice least compatible with the neutrality thesis, can substantially endorse the anti-discrimination thesis as applied to disparities in the receipt of services and in health outcomes. When using the lens of the anti-discrimination thesis, a libertarian might reach a more modest moral conclusion than the one we shall defend,and a libertarian does not endorse the more inclusive moral concern shown for disparities in health outcomes embodied in the neutrality thesis. However, in Section III, we explore some instances in which the libertarian view might agree with our conclusion that some patterns of racial and ethnic disparities should be counted as injustices, and not simply moral failings.

A family of justice theories known as egalitarian theories offers more solid support for both the neutrality thesis and the anti-discrimination thesis, even as those theories diverge substantially in their theoretical foundations. Egalitarians, unlike libertarians, are intrinsically concerned with the existence of inequalities. Egalitarians themselves differ as to how much inequality they find morally tolerable, the reasons they find inequalities to be morally problematic, and the kinds of inequalities they consider to be the central job of justice to combat.

One strand of egalitarianism prominent in the bioethics and health policy literature borrows heavily from the work of John Rawls (Rawls, 1971). The first principle of the Rawlsian theory is that everyone should be entitled first to an equal bundle of civil liberties (e.g., political and voting rights, freedom of religion, freedom of expression, etc), which shall not be abridged even for the sake of the greater welfare of society overall. Secondarily, everyone should be guaranteed a fair equality of opportunity. That principle of fair equality is given a robust, substantive interpretation such that permissible inequalities in such things as income and wealth work to the advantage of the least well-off segments of society. Fair equality of opportunity is thus a term of art, signaling more than a formal commitment to non-discrimination, but also an affirmative commitment to resources necessary to ensure that all citizens of comparable abilities can compete on equal terms. For Rawls, this commitment means a guarantee of educational resources sufficient for all persons to pursue opportunities such as jobs and positions of authority available to others within society.

Norman Daniels seizes on Rawls's core arguments (Daniels,1985). He accepts the core Rawlsian framework but offers a friendly amendment to the Rawlsian theory. Daniels claims that once we acknowledge that there are considerable differences in the health of individuals and that the consequence of those differences is that individuals differ substantially in their opportunities to pursue lifeplans, we must relax Rawls's own assumption about the rough equality of persons. Once this assumption is relaxed, the theory has implications for how we think about healthcare resources. If, as Daniels argues, health is especially strategic in the realization of fair equality of opportunity, and that healthcare services (broadly construed by Daniels) make a limited but important contribution to health, then we derive a right to healthcare sufficient to pursue reasonable life opportunities. The logic of Daniels' account clearly lends support to the neutrality thesis in as much as disparities in health outcomes are precisely the sort of consequences that the principle of fair equality of opportunity treats as unjust and therefore, as proper objects of remedial governmental action.

In addition, Daniels' version of the Rawlsian theory can be seen as lending support for the anti-discrimination thesis, although this is not an element of Daniels' theory that he himself highlights. For example, the theoretical support for treating inequalities in health outcomes among racial groups as unjust, as distinguished from a rationale that makes inequalities among persons generally unjust because of their adverse impact on equality of opportunity, lies in its endorsement of Rawls' core notion of a formal principle of equality. Rawls and Daniels both start their discussion of equality of opportunity with the formal principle that morally irrelevant distinctions should not be employed as a basis for determining the range of life opportunities open to persons. Matters of race, gender, and the like are counted as irrelevant, so if their claims are plausible, then even disparities in services received ( as well as disparities in health outcomes) based on racial and ethnic categories warrant some moral scrutiny.

Other members of the egalitarian family of justice theories offer more direct support for both theses. The “capabilities” approach argues that it is the job of justice to protect and facilitate a plurality of irreducibly valuable capabilities or functionings (Sen, 1992; Nussbaum, 2000). Capabilities theorists, led by Amartya Sen, generate slightly different lists of the core human capabilities central to the job of justice, but all converge on the idea that a variety of health functionings, including longevity and absence of morbidity, are among those centrally important human capabilities. Unlike the modified Rawlsian concept, which makes the importance of health and hence healthcare derivatively important because of health'se specially strategic role in preserving equality of opportunity, the capabilities approach reaches similar conclusions about the intrinsic importance of health, and more directly, the goods instrumental to its realization. Based on Sen's theory, inequalities among any of the core capabilities are matters of moral concern. Thus, as the neutrality thesis asserts, any finding of disparities in health outcomes should trigger moral scrutiny.

Among the core capabilities included on Sen's list are capacities for all to live their lives with the benefit of mutual respect and free from invidious discrimination.Thus, support for the anti-discrimination thesis also flows naturally from the capabilities approach inasmuch as the value of equal human dignity and respect is of fundamental moral importance, as is health. Disparities in services received, no less than disparities in health outcomes, therefore trigger a heightened moral scrutiny under a theory that renders inequalities of both sorts morally problematic.

Libertarian and egalitarian theories are two broad theoretical traditions that at face value seem to have the greatest divergence in their implications. However, they have been shown to result in greater convergence, at least on the anti-discrimination thesis, than might otherwise be suspected. Apart from the (perhaps) unexpected convergence of two quite different comprehensive moral theories on the interpretation of the formal principle of equality, there are additional philosophical arguments favoring the anti-discrimination thesis that do not require taking sides with any comprehensive moral views.

Recent work in political philosophy by John Rawls begins with the assumption of what he calls a reasonable pluralism of comprehensive moral views (Rawls, 1993). In a democratic nation, persons motivated to reach agreement on the basic social structure, understood as shared basis for social cooperation, will seek an overlapping consensus on some evaluative questions. That consensus will necessarily include a commitment to the view of each person as a free and equal citizen. While critics have questioned how much substantive moral content can be derived from this perspective, they generally agree that some underlying commitments are widely shared in any democracy (Gutmann and Thompson, 1996). Among them are the ideas that the interests of all should be given equal weight regardless of race, creed, color, gender or other attributes deemed morally irrelevant. Although such a notion does not settle the deeper moral question of which attributes are morally irrelevant, the crucial point is that such views form the bedrock of most Western democracies. Underlying this desire for equal respect and concern is the vague but powerful idea of human dignity and the importance we attach to equality of treatment for the least advantaged that the more powerful members of society have secured for themselves (Harris, 1988).

Thus, although there is a diversity of possible justifications for the importance of health and healthcare services, there is widespread basis for agreement that inequalities in health outcomes that track racial and ethnic lines, especially when racial and ethnic lines also track other indices of social disadvantage, are ethically problematic. This feature of democratic theory, reflected also in equal protection law, justifies at minimum the added moral scrutiny required by the anti-discrimination thesis.

So far we have established that egalitarian theories, and in particular capability theory, provide moral justification for the neutrality thesis. Thus, even with a libertarian view, the failure of individuals and institutions to offer health services to all racial groups on an equal basis can be a significant basis for moral condemnation. Even if the moral scrutiny demanded by the neutrality thesis and the added moral scrutiny demanded by the anti-discrimination thesis are warranted, this is not the final word. All that has been established thus far is that governments and health care institutions have a moral obligation to investigate identified disparities. The key questions are how governments and health care institutions should interpret the moral meaning of the results of such an investigation, whether disparities should be considered injustices, and under what conditions. On many moral accounts, an evaluation of the explanations for the disparities is needed to make a judgment about whether the disparities represent an injustice. In other words, whether disparities in health outcomes or in the services patients receive constitute an injustice depends for some on the causal story that stands behind the disparity. Thus, while there may be wide agreement about the moral imperative to investigate identified disparities, at least with respect to morally suspect groups, there is far less agreement about how to interpret the moral significance of the results of such an investigation.

The moral significance of causality is a difficult sticking point in moral philosophy. There is a natural inclination in theories of individual morality, as there is in law, to bind moral responsibility and causal responsibility together. We do not ordinarily think, for example, in law or morality, that an individual is morally culpable for adverse consequences arising from circumstances over which that individual had no control. Lack of causal efficacy is the end of the story for many assessments of moral and legal responsibility. Moreover, a judgment of causal responsibility is a threshold concern for many accounts of individual moral and legal responsibility, and the presence of some causal contribution to the harm of others opens the door to legal analysis. Theories of justice, however, are more varied and often more controversial than the individual model in their understandings of the relation between causal and moral responsibility.

Some theories of justice employ something similar to this individual moral responsibility model in their assessments of the justice of social institutions. Libertarians, for example, link a judgment of injustice to some intentional harm. That view holds that adverse consequences or disproportionate burdens borne by some individuals or groups as a consequence of the structure of social institutions do not warrant a judgment of injustice. The libertarian views these consequences for the most part as merely unfortunate, not unfair.

The libertarian view is an especially stringent rendering of the claim that moral responsibility for society and its political institutions is linked necessarily to a direct causal responsibility. It is a stringent standard as it demands that the causal connection be an intentional harm.

However, there is theoretical room for the libertarian to reach an even stronger conclusion that racial and ethnic disparities in health outcomes and the receipt of health services are morally condemnable failings of particular persons or institutions. In some cases, the libertarian can conclude that these disparities are injustices. There are at least three ways that the libertarian can reach such conclusions.

First, for the libertarian, patterns of inequality are not morally troubling in themselves. However, this assertion is qualified by the proviso that those patterns are morally unproblematic only as long as they are not the consequence of prior injustices in social exchanges or agreements. This nod to historical context is crucially important. If the social and institutional history that causally contributes to present patterns of inequality are in and of themselves unjust, perhaps the result of past intentional harms whose adverse consequences remain today, then present patterns of inequality may be judged as unjust, and not merely a matter of moral failing of individuals or non-governmental institutions. There is nothing intrinsic to the libertarian view that makes it hostile to such historical claims regarding the legacy of racism, the intentional harms based on racial or ethnic prejudice, or the moral taint on the advantages obtained from such practices.

Second, for one brand of libertarian theorist, the constraint on coercive state appropriation of private assets for the purposes of achieving certain patterns of distribution does not entirely restrict what states can do with respect to redistribution. While private assets are put beyond the reach of states, not all resources are private. According to some libertarians, redistribution for the purposes of combating inequalities in the health care context are acceptable when it involves public resources or the decision to devote resources to activities that benefit the public at large. Medical education and the construction and operation of health care facilities are clear examples of public resources being invested deliberately for the promotion of the common good.

Even if the libertarian can argue that there is no antecedent duty to support such activities for the common good, the claim of allegiance to the state itself is said by some libertarians to de pend upon strict neutrality between its citizens (Nozick, 1974). This requirement of neutrality clearly makes all disparities in services received, as well as disparities in health outcomes such as racial and ethnic health outcome disparities, unjust. If the neutrality requirement endorsed by some libertarians is a strict one, as it is in Nozick's libertarian theory, then the proper test of neutral state action is neutrality of effect on its citizens (Raz, 1986). Thus, one particular interpretation of libertarianism supports the neutrality thesis.

Moreover, the moral failing associated with its violation is an injustice. Of course, not all libertarian theorists endorse the political neutrality thesis and accordingly, those libertarians would be committed neither to the neutrality thesis we have defended nor to the finding of an injustice if neutrality of effect is not achieved.

A third possible exception to the libertarian's general reluctance to see an injustice in any disparities in receipt of services or health outcomes, even in the case of racial and ethnic minorities, lies in the libertarian's account of what constitutes intentional harm. The typical definition of an intentional harm is one that is generated from a fully conscious or present-to-mind motivational stance. Therefore, overt racist actions would surely count as intentional harms. For example, if services were not offered to racial and ethnic minorities because of a conscious intention to make their health outcomes worse, or as a deliberate assault on their dignity, these denials of services would count as intentional harms. In this narrow range of cases, the libertarian has no choice but to support the anti-discrimination thesis and conclude that the moral failings involved are injustices.

Less clear, however, is how the libertarian must account for more subtle, often unconscious, instances of racism. The resolution depends on the view of intention employed by the theory. In our judgment, nothing intrinsic to the libertarian theory rules out a more expansive account of what constitutes an intentional harm, even though the ideological thrust of most libertarian theories would be naturally resistant to any effort to look behind an agent's conscious state of mind. The libertarian would have to articulate a plausible rationale for adopting the narrow construal, and as long as the core intuition of what constitutes an injustice is tied to intentional harm, limits on the psychological transparency of an agent's own true intention would seem to need a persuasive argument for such a restriction.

Other justice theories, including two prominent versions of egalitarianism, make the locus of causal responsibility an important consideration. Consider first a rather permissive standard sometimes referred to as the brute luck conception of justice (Scanlon, 1989). Brute luck theories count as an injustice all those inequalities that are not due to the choices of individuals. All inequalities that are beyond a person's control are therefore judged as brute bad luck and deserving of remedy, or if the inequality cannot be eliminated, compensation. Such theories take an indirect account of the causal story leading to the inequality in as much as the only inequalities society does not have to eliminate are those said to be chosen. While responsibility for some inequalities is laid at the individual doorstep, the brute luck standard holds society morally responsible for all inequalities that the individual did not bring on by his or her own choices. For example, the brute luck view recognizes that inequalities that result from genetics, ill health not brought on by lifestyle choices, and being born into a poor, uneducated family are all illustrative of inequalities that should be remedied by society. The brute luck theory can be contrasted with an alternative claim that attempts to reign in the moral responsibility of society for unchosen ine qualities. The social structural concept argues that two conditions must be satisfied for society to incur an obligation to remedy inequalities: 1) the inequalities must not be the result of an individual's own choices and 2) those inequalities must not be attributable to natural fortune that the society had no hand in creating. Examples of natural bad fortune, for which no social remedy is due, include genetic differences and natural disasters. The focus is on the way social structures contribute to inequalities, and more specifically on the way that unjust social structures influence the creation of inequalities that reduce the life prospects of some people relative to others. Like the libertarian view, the social structural view demands proof that society had a causal hand in producing the inequality before it assigns society the moral responsibility for its elimination or reduction. The difference is that the social structural view does not require that the causal link between society and the inequality involve intentional harm. Instead, the social structural view adopts a less stringent requirement demanding only that the inequalities be an artefact or consequence of a particular social arrangement.

Let us next consider how the social structural and brute luck concepts might justify or limit the scope of application of a claim of injustice for disparities in health outcomes or health services. There are two important implications of the brute luck view. First, the brute luck standard provides robust justification for the injustice of inequalities that are covered by the neutrality thesis, but no special justification for the discrimination thesis. It would find all inequalities in health outcomes morally unjust, except for differences in health outcomes that are attributable to patient choice. . The brute luck view reaches this conclusion independent of whether the inequalities are concentrated within racial and ethnic minorities or the majority ethnic and racial population. The fact that inequalities cluster along racial and ethnic lines or along lines of social disadvantage adds nothing to the moral assessment insofar as no further factual information of any sort (including some sort of causal story) is needed to find an injustice.

Second, because the brute luck concept is indifferent to any casual inquiry beyond the role of individual choice, the brute luck view can provide no special justification for viewing inequalities in health services as injustices. For example, the brute luck view is indifferent to whether inequalities in health outcomes between patient groups are a result of disparities in access to health services or the impact of differential socioeconomic status and educational background. Both generate social duties to reduce or eliminate disparities in health outcomes. The fact of brute, unchosen inequality is enough.

The social structural concept takes a different view. Attaching a judgment of injustice to disparities in services or outcomes along lines of racial and ethnic minority status-- especially if burdened with other social disadvantages (the anti-discrimination thesis)-- is entirely consonant with the social structural view. The claim of the neutrality thesis, which is that disparities in health outcomes that do not necessarily involve disadvantaged groups also constitute an injustice, also can be accommodated by the social structural view, but only if a different set of morally relevant considerations can be brought to bear. Because the social structural view requires a causal story linking the social structure to health outcomes disparities, the case for injustice when disparities involve majority racial and ethnic patient groups would be more difficult to make than it would be for racial and ethnic groups who also experience broader social disadvantages. Even for these latter groups, a social structural view would necessitate the telling of a somewhat complex causal story to reach the conclusion that the inequalities are a matter of injustice and the responsibility of society to remedy. .

A key question faced by libertarian, social structural, and brute luck theories is just how much of the causal story needs to be sorted out before deciding whether a disparity constitutes an injustice. All of these theories exclude from the realm of social responsibility inequalities generated by the choices and actions of individuals. But is this blanket exclusion plausible? This is where many of our most influential theories of justice appear ham-handed when compared with the kinds of moral intuitions that influence much of social policy in the United States and other industrial nations. For example, health insurance and welfare laws generally eschew fine-grained apportionment of individual, social and natural causal contributions to ill health. In many respects, health insurance plays the role of a kind of social safety net, catching those who fall through, regardless of the cause.

There are at least two potential explanations for why the moral foundations of many aspects of social policy do not fit well with some leading theories of justice. First, the apportionment of individual, natural, and social responsibility is, in practice, extremely difficult to disentangle. Second, because apportioning causal responsibility is often so hard to do, it is fraught with the risk of error and is potentially unfair. There is no doubt that these difficulties both explain and justify why public policy relies on moral lenses that deliberately leave some elements of the causal story out of focus. We think that the right mix of moral lenses leaves such differences out of account when examining health outcomes , This is the insight captured in the claim of injustice attaching to the inequalities coming under the scrutiny of the neutrality thesis. It is also the moral basis of public health, which finds any disparity in health outcomes to be morally problematic, regardless of who is affected. However, we argue that a a special moral sensitivity to the constellation of race, ethnicity, and social disadvantage should be added back into the mix , especially when we have ample reason to believe that, although the precise causal story is complex, racial differences have made a dramatic contribution to the disproportionate burdens that are an artefact of the social structure. This is the insight captured by the claim of injustice attaching to the inequalities coming under scrutiny by the anti-discrimination thesis.

From this stereoscopic vantage point we turn to a few examples of how patient choices and behavior fit into the arguments thus far. Although neither the neutrality thesis nor the anti-discrimination thesis rejects the notion that patient choices and actions make a moral difference in assessing the injustice of disparities in health outcomes, we deny that patient choice and behavior necessarily vitiate a conclusion of injustice.

Consider, for example, how that argument for the moral decisiveness of a patient's own choice to refuse treatment offered and recommended might seem to settle the issue of injustice once and for all. One possible explanation for some disparities in health services is that racial and ethnic groups exhibit different preferences for some types of medical care. Some groups may have higher aversion rates, for example, to invasive coronary care procedures. In some instances, preference differences make all the moral difference and a conclusion of injustice associated with disparities in the receipt of care may be rebutted. However, even if disparities in utilization rates are explained primarily by differences in uptake, rather than differences in offering, that is not necessarily the end of the matter. For example, gaps in mammography use between white and African-American women have closed considerably over less than a decade. This has been a consequence of public health education and outreach campaigns mounted on the assumption that gaps in knowledge and awareness, not merely a matter of differences in individual preferences or cultural values, accounted for differences in mammography rates.

Others have argued that minority aversion to the utilization of beneficial treatments might be based on a reasonable distrust of medical institutions and personnel (Randall, 1996). Whether such distrust is widespread is an empirical matter, and determining whether such distrust is reasonable lies beyond our task here. However, to the extent that the formation of preferences among racial and ethnic minorities is a product of a legacy of intentional discrimination that results in disparities in utilization and health outcomes, the fact that patient preferences account for all or some portion of those disparities does not obviate their injustice. If the preferences themselves are the fruit of a morally tainted history of institutional relationships, those who occupy positions of authority within those institutions have continuing moral obligations to ensure that patient preferences that are detrimental to racial and ethnic minorities are not systematically disadvantaging. In short, our view argues for looking behind or beyond mere preference in some instances to make a moral assessment of racial and ethnic disparities in the uptake of health services and in the resulting disparities in health outcomes.

Libertarian theories of justice, as well as most forms of egalitarianism, are mute on whether preferences must be taken at their face value. Many brute luck theorists believe that some preferences are beyond voluntary control and are instances of brute bad luck for which there is a duty to remedy (Cohen,1993). The capability theorist also admits the possibility that some preferences are shaped by norms and institutions that involve unjust discrimination (for example, women's preferences for female circumcision . However, the idea of looking behind preferences is not the exclusive theoretical property of the brute luck theorist or any other particular theory. If the preferences themselves bear the moral taint of social structural injustices, then the social structural theorist cannot object. If the preferences bear the moral taint of intentional harms, then the libertarian cannot object. The difference is that each requires a different causal story to reach a conclusion of injustice when individual preference would ordinarily settle the moral matter in favor of there being no injustice.

Under all major accounts of justice, much of the work leading to a judgment of injustice involves getting the causal story straight, with some seeing overwhelming social determinants of such behaviors at work and others doubting the conclusiveness of the evidence and fearing the consequences of widespread belief in its truth. Although we lack the expertise to sort out these factual debates, our claim is a simpler one: there is too much at stake morally in ignoring the real possibility of some social structural causation. The demand for a precise apportionment of causal responsibility fails to take seriously the potential moral salience of the continuing effects of the legacy of racism and discrimination. Attaching a presumption of injustice to disparities in health outcomes that cluster along racial, ethnic, and socioeconomic lines is responsive to the need to fashion public policy with an awareness of the moral saliency of that legacy. Once again, we note that even the libertarian must attend to the importance of that history, for libertarianism is, in its own terms, a theory whose application is constrained by the assumption that patterns of inequalities are morally benign only when they emerge from a historical milieu in which injustices are not causally transmitted into the present context. In our view, few libertarians can claim that confidence when it comes to matters of race.

Moreover, at least for matters as central to human flourishing as health, we agree with the capabilities approach. The capabilities approach does not generally insist on the complete causal story to count disparities in health outcomes as instances of injustice. Moreover, the capabilities view demands additional moral scrutiny for racial and ethnic disparities in health care services and outcomes for moral reasons that have their foundation in capabilities other than health. These are capabilities that signal the importance of living a life as a free and equal moral person and enjoying the respect and dignity accorded to all citizens (Faden and Powers, 1999).

From the perspective of the health professional, the bottom line of this analysis can be summarized as follows. All the theories that we have reviewed have reasons to morally condemn disparities in health services and health outcomes involving racial and ethnic minorities. These theories have different reasons for reaching this conclusion, and they do not all agree that such disparities necessarily constitute an injustice. However, they all agree that race and ethnicity are morally irrelevant to the distribution of health care services and the outcomes with which these services are associated. Even from a libertarian viewpoint, the failure of individuals and institutions to offer health services to all racial groups on an equal basis can be a significant reason for moral condemnation.

In some respects, this is stating what is morally obvious. It is wrong for health professionals to discriminate on the basis of race or ethnicity. General moral duties of equal respect, as well as role-specific duties of the healing professions, obligate health professionals to accord equal consideration to each patient. The Hippocratic Oath requires physicians to apply treatments “for the benefit of the sick”and to “keep [patients] from harm and injustice” (Edelstein, 1967). The standard interpretation of the Hippocractic tradition concludes that such duties be applied impartially, and that no matter of personal preference or prejudice should compromise those duties with respect to any patient (Pellegrino and Thomasma, 1988). The Code of Ethics of the American Nurses' Association similarly argues that the foundation of their professional duties rests in duties of beneficence impartially applied to all patients (American Nurses' Association, 1985). Health care professionals are also obligated to address the moral context in which they work and to take responsibility for ensuring that equal respect and treatment is accorded by colleagues and by the health care organization where they work. To the extent that unconscious biases compromise their impartial duties toward their patients, there are derivative moral duties to identify and counteract those biases.

One aim of this paper is to defend the view that racial and ethnic disparities are not merely matters of individual moral failing on the part of health professionals, but are also social injustices. Insofar as health professionals and professional organizations subscribe to this view, they should take a leadership role in advocating for interventions to reduce these disparities. It is here that good empirical data, capable of teasing apart the various factors that contribute to racial disparities, are critical. Ethical arguments can justify the need for social action, but knowing precisely how to effectively intervene requires an integration of ethics with facts.

. We do not claim that the neutrality thesis and the anti-discrimination thesis offer an exhaustive account of the sources of value underpinning the broader range of moral concerns in health care policy. We have argued elsewhere that in addition to medical outcomes some arguments for universal health care may depend as much on their impact on aspects of human well being other than health (Faden and Powers, 1999).