Toward establishing a universal basic health norm.

Acharya, Arnab K.

Vast improvements in human health have been made during the past century. Indeed, gains in increased life expectancy and reduced physical impediments for much of the population were greater than in any previous century. Yet the gains were not uniform across the world or even within individual countries. The variations in health status among people cannot for the most part be explained through genetic differences. Instead, in most instances the variations in the last century and at the turn of the current century correspond to the variations in the distribution of control over material resources.

A cursory overview of the worldwide distribution of health status confirms that there are many who experience a low level of health. Restricting the discussion to mortality as an indicator of health, (1) we find that the distribution of life expectancy can be described in the following way: It is higher than seventy-two years for nearly a to 2 billion people living mostly in wealthy industrial countries, newly rich countries of East Asia, and the upper income class of middle-income, emerging-market, and the rapidly growing developing countries. Nearly 3 to 4 billion people, consisting of the vast majority of people living in middle-income countries, including those of emerging-market economies and the middle class of rapidly growing developing countries, can expect to live sixty to sixty-five years. Life expectancy is below sixty years for nearly 1 billion to 1.5 billion people living mostly in South Asia and Sub-Saharan Africa. Large groups of people can live beyond the age of seventy-two, but most people currently can improve their health significantly, especially those who expect to live for less than sixty years and those who can expect to live to between sixty and sixty-five years.

Given these facts, what norms should guide us in examining potential changes in policies and institutional arrangements that affect health status? One approach that figures prominently in recent policy discussions is what might be called an equality-demanding norm. (2) This norm stipulates that we

* This article has been significantly improved due to comments by Christian Barry, Peter Davis, Stephen Devereux, Madelyn Hicks, Peter Houtzager, Paul Howe, Connie Rosati, the anonymous referees for this journal, and the participants of the workshop "Public Health and International Justice," Carnegie Council on Ethics and International Affairs, New York, April 2002. I would also like to thank Ezinda Franklin and Arthur Smith for significant editorial assistance. None but the author is responsible for any errors in this paper. should strive to make health outcomes or access to health care more equal. A World Health Organization (WHO) official emphasizing the intercountry differences cited above writes: "It is particularly important to assess whether inequalities in health outcomes are increasing or decreasing over time and to make comparisons between countries." (3) Similarly, an influential publication on health equity that stresses inter- and intra-country inequalities states that it was "motivated by a common concern about unacceptable differentials in health." (4) In a recent commentary in the Lancet, Davidson Gwatkin emphasizes the importance of differences in maternal mortality rates within, as well as between, countries and regions. (5) Focusing on Indonesia, Gwatkin argues that the goal of health policy should be to bring the mortality rate of all lower-income quintiles to the present level of the highest-income quintile. (6)

I argue that the locus of debate in public health and international justice should move away from discussions that stress the importance of achieving some form of equality in health status. (7) Given the available resources, I deny that there is a moral imperative to pursue equality of health status or access to care. Under current resource constraints, a just international public health policy is not best served through demanding equality in health status or horizontal equity in access to health care across the world, and particularly within developing countries. (8) Resource constraints--the domestic budget together with foreign aid--are always severe in developing countries. If institutional mechanisms are arranged under these constraints in order to ensure commonly accepted egalitarian goals, such as horizontal equity, then the likely outcome would be to decrease the health status of many who do not currently enjoy particularly high levels of health. I believe this to be enough of a reason, for now, to abandon an equality-demanding norm regarding health status at the global level. An alternative to an equality-demanding norm is the prioritarian norm, which demands that the least well-off be served first. Although this view may avoid in practice some of the negative implications of the egalitarian view, I argue that it too is untenable given the resource constraints we currently face. We should instead develop a threshold norm that characterizes minimally adequate health status. People enjoy minimally adequate health status when they are capable of fully participating in carrying out their own life plans within a fairly lengthy time horizon. It is this norm, and not equality of health status or even granting priority to the least healthy, that should guide international health resource allocation. (9) I shall claim that an institutional order is just with respect to health to the extent that participants in this order do not (avoidably) fail to reach this threshold.

Two policy conclusions follow when one tries to meet the threshold norm while faced with current resource constraints: First, the scope for redistribution is limited within developing countries, since it will adversely affect the health of those who would be considered to be enjoying health just above the threshold level. Second, the potential for relaxing the resource constraints lies in our abilities to redirect resources from developed countries, perhaps even through reduction in domestic health expenditure in these countries.

THE GLOBAL DISTRIBUTION OF HEALTH STATUS

It might be suggested that there is a low-cost solution that would bring everyone to a high level of health and that only commitment is lacking. If such a solution existed, nearly everyone would agree to bring about a scenario where equality at a high level of health prevails. Since the cost is low, developing countries themselves would undertake this policy easily without facing a significant fiscal burden, hence without having to experience the slowdown of progress in literacy rates, buildup of infrastructure, and growth of job opportunities. Unfortunately, no such solution has yet been identified.

It is generally recognized that there are low-cost health interventions that can improve life expectancy considerably, as well as evidence that low-cost dietary supplements would remedy stunting and other physical incapacities. The landmark World Development Report 1993 on health policy prepared by the World Bank reported that significant health gains can be made by making available $21 annually (in 1992 U.S. dollars) per capita public expenditure on preventing and treating infectious disease. For many of the lower-income countries, $21 was approximately 7 percent of the GDP per capita in 1992, just about the average percentage of income spent in Europe for health. (10) Very few developing countries have undertaken this level of public expenditure; but even if they were to spend this much, the World Bank report indicated that the level of care would not extend beyond the treatment and prevention of infectious diseases.

Although the middle classes in many developing countries have made considerable health gains, they also face chronic noncommunicable illnesses at the onset of middle age, such as heart diseases, various forms of cancer, and diabetes, that require better medical infrastructure and greater access to well-trained physicians. None of this was budgeted within the 7 percent of GDP health budget recommendation, which cautioned against investment in tertiary care that usually delivers care for noncommunicable illnesses. The simultaneous high prevalence of chronic noncommunicable and communicable illnesses--a situation known as epidemiological transition--is common in many Asian and Latin American countries. (11) Thus, where significant improvement has been made for some people, further overall improvement is only possible through attending to high-price care while continuing to offer low-cost care. Given the fiscal constraints most developing countries face, the public expenditure of 7 percent recommended by international agencies on easy-to-treat illnesses would entail a reduction in free tertiary care services supplemented by some private expenditure on which the middle classes critically depend.

Much has been achieved in Kerala, India, and Cuba at low cost. (12) However, in recent years it has been noted that many of the poor and the near poor do not receive adequate care for noncommunicable illnesses in Kerala, as much of the care is only privately available. The government has initiated many well-placed programs for communicable diseases with higher public expenditure than the all-India level, but such illnesses as cardiovascular problems and diabetes among the working-age population often go untreated and remain a major cause of impoverishment in Kerala. (13) Bringing the public expenditure at the all-India level to the Kerala per capita level would still leave many Indians without adequate care for most noncommunicable illnesses. It may even in the process deny some free tertiary care to the poor and near poor.

There is, thus, a practical restriction on increasing health expenditure and on what can be achieved at currently allocated levels of expenditure through this increase. Even if the budget is increased to a higher amount without imposing too much of a fiscal burden, the outcome will not be much different. If developing countries were able to spend l0 percent of their GDP, the average expenditure for lower-income countries would not exceed $25-$65 per capita. Even in places (including the rich industrial countries) where much of health care is streamlined by emphasizing cost-effectiveness, the cost of medical care is still a constant concern for both the public and private sector. The average health expenditure per capita in the rich countries between 1997 and 2000 was roughly $2,000 annually, nearly the same level of per capita income expenditure as most middle-income countries. (14) There is not much scope for increasing this expenditure. Thus the world health budget would roughly remain around 7 percent to 10 percent of the world GDP.

Distribution of command over health care and other types of resources corresponds to health distribution. In low-income countries where progress has been made, additional increases beyond treatment and prevention of infectious diseases have only reached the middle class in a very limited way. If developing countries were to spend as much on health care as developed countries as a percentage of GDP, they might be able to bring many who suffer from communicable illness to the health level of the middle class, but reducing access to the care that the middle class receives for noncommunicable illnesses, though often inadequate by the standards of rich countries, will diminish their well-being.

OBJECTIONS TO THE EQUALITY-DEMANDING NORM

Equality-demanding norms have been proposed as a solution to remedy the currently unacceptable health distribution. Under current domestic budgetary constraints, however, equality will not be achieved at a high level. Instead, it can be achieved only by reducing the health of a large number of people in developing countries. Within the countries in epidemiological transition, health inequality can arise as a result of simultaneously providing care for chronic illnesses suffered mostly by the middle class, who are often associated with greater longevity and cheap, effective care for communicable illnesses, and for the great number of less-well-off people. (15)

Pursuing equality in health status is ethically impermissible when equality can be achieved only by reducing the well-being of a significant number of people in an important way who are not among the worse-off but nonetheless do not achieve very high levels of health status. In fact, the practical implications of equality-demanding norms lead to counterintuitive results when applied to the global allocation of health resources.

Why might equality-demanding norms be appealing? Thomas Scanlon notes that one of the reasons we are concerned about equality in general is that we think that those who are truly badly off can be helped through a transfer of resources from those who are better off. (16) Scanlon argues that we think such a transfer is desirable because it can be accomplished without other bad effects and will alleviate suffering without creating new hardships of comparable severity. Scanlon's view carries significant weight when we consider redistribution that affects health. Income losses may not affect well-being, while losses in health status will always decrease well-being; this is because health is constitutive of well-being in a way that income is not.

Transfers to achieve equality can often involve imposition of significant costs to those from whom resources are redistributed. Thomas Nagel and Derek Parfit have presented such cases. (17) Nagel asks us to imagine parents making a decision that would affect their two children. The first child is a happy and healthy child, likely to prosper in the family's present suburban community, which has affordable good schooling. The second child has a painful handicap that can only be treated in a more expensive setting, a city, where good schooling is unaffordable for the first child. The parents weigh the decision as to whether to move to the city. Nagel comments:

 If one chose to move to the city, it would be an
 egalitarian decision. It is more urgent to benefit
 the second child, even though the benefit we
 can give him is less than the benefit we can give
 to the first child [in the suburbs]. This urgency
 is not necessarily decisive. It may be outweighed
 by other considerations, for equality
 is not the only value. But it is a factor, and it
 depends on the worse off position of the second
 child. (18)

A modified argument from Parfit illustrates that our intuition does not always favor egalitarian outcomes. The argument depends at first on examining a situation where our intuition flavors equal outcomes, but perhaps for reasons other than the acceptance of equality-demanding norms. Consider two cohorts with equal living standards in different societies of equal size without any contact with each other, and that an international agency can redistribute resources engendering two distinct life expectancy outcomes from which we are to choose one over another:

Case 1

(S1') Society A: 35 Society B: 85 (S2) Society A: 55 Society B: 60

Very few of us would prefer the state of the world to be $1 over $2, although the combined years for both societies in $1 are higher. However, it is hard to argue that we favor S1 over $2 because we find the outcome more equal in $2. We prefer $2 to $1 because we feel that people in society A should live a long life, although the twenty years more for them is less than the loss of twenty-five years per person in society B under $2. A gain of twenty years when starting from a low level of health outweighs the loss of twenty-five years for those enjoying good health. We feel people in society A are cheated in S1, while in society B people live a longer life. To see that equality is not what we prefer, consider Case 2:

Case 2

(S1') Society A: 50 Society B: 85 (S2) Society A: 55 Society B: 60

Most will prefer S1' over $2. Society A loses five years while society B gains twenty-five years when options S1' and $2 are compared. Parfit and Nagel point out that we do not count as equally important or urgent just any increase to the worse-off, but instead opt for a sufficiently high increase when individuals are relatively more worse off. We do not consider the gain of five years for the more-worse-off group to be worth the loss of twenty-five years for the better-off, at least when the better-off would fare pretty badly after this loss.

Accepting that health inequality should not always be rectified when it requires leveling down casts doubt on whether we can accept equality-demanding norms as fundamental, even when we may in practical situations favor more equal distributions. We may accept leveling down when the gains to the worse-off are large in comparison to the losses to the better-off. (19)

OBJECTIONS TO THE PRIORITY VIEW

In light of these concerns, some philosophers have suggested that we should abandon equality-demanding norms and adopt instead what Parfit has called "the priority view.'' (20) According to the priority view, losses to the well-off are justified in relation to how badly off the worse-off are; that is, the worse-off should be helped because they are badly off, not because they stand in some particular way to others who are better off. This view differs from the egalitarian view in that resource reallocation is justified not because there are differences in well-being but because such transfers can raise the position of the least advantaged. (21) Thus, in considering Case 1 we note that people in society A are badly off if $1 prevails, while they would be much better off if $2 prevailed, independent of what happens to society B. In considering Case 2, the worse-off group is not nearly as badly off as in Case 1 and, therefore, the leveling down of others' well-being matters more.

The priority view has strong intuitive appeal, but we should nevertheless reject it. In developing countries there are many who can expect to live less than to the age of fifty. This is particularly true in light of the HIV/AIDS epidemic. The priority view would demand that some of these people be brought up to greater life expectancy, perhaps at the cost of some leveling down. Taking the HIV/AIDS example, one recalls that just a few years ago when antiretroviral treatment (ART) was significantly more expensive, the epidemic had the potential, even in countries with moderate prevalence, to absorb expenditures larger than the health budgets of most developing countries if ART was offered. Thus, treating HW/AIDS had the potential to bring down the health of a great many people who had already lived beyond the age of fifty. The possible five to seven years of life gains to some younger people would have wiped out quite substantial gains for many other people.

The priority view leads to counterintuitive results because addressing ill health can sometimes be very costly. Treating the very sick at an exorbitant cost would drive resources away from those who can be treated successfully at a relatively lower cost. If the loss is significant to those who can be treated more cheaply and are only relatively better off, we may not want to treat those requiring treatment at a much higher cost even though they are worse off, indeed badly off. This is the case especially if the number affected among the better-off is significantly high. Such cases are exemplified by the offering of costly treatment of rare cancers (or for HIV/AIDS a few years ago in countries with moderate prevalence) that are fatal while many others may be suffering from probabilistically less-fatal malaria, which can often be treated extremely cheaply. The principle behind this intuition is similar to that espoused by Francis Kamm: that it is permissible that the aggregation of significant lesser losses to many people outweigh even greater losses to a few, even though no individual in the larger group would suffer as much as each individual in the smaller group. (22) Thus one should most likely prevent lifelong disabilities of many children through mass polio vaccinations rather than treat childhood leukemia in resource-poor settings of developing countries.

Further, it is permissible to use resources to achieve large gains to a better-off individual instead of using the resources to achieve small gains to a less-well-off person, leaving the less well-off with no compensation. Imagine someone very ill who would require a large amount of resources to gain only a few months of life, while a better-off person can use this amount to live twenty years more from middle age but would otherwise live only a few more years than would the currently less-well-off person. Note that the denial of the resources to the worse-off does not involve significant losses to the worse-off from the present state. The action would be forbidden by equality-demanding or prioritarian norms. Of course, allocations that secure greater gains for the better-off will not always seem plausible. Imagine ten years being added to the better-off instead of eight being added to the worse-off, who will live fifteen years less than the better-off in the status quo. We are likely to think that the worse-off should receive the resources. This is true, I believe, because we think that people should receive resources to allow them to achieve an adequate level of health, and that they should therefore have access to ever-larger shares of resources so long as this does not lead to others falling short of this level.

THE THRESHOLD VIEW

An alternative view to egalitarianism and prioritarianism is what could be called the threshold view. This view centers on noting that we should have strong reservations against reducing people's welfare if it, although higher than some people's, is not at a very high level. Thomas Nagel recognizes this situation:

 If the choice is between preventing severe
 hardship for some who are very poor and
 deprived, and preventing less severe but still
 substantial hardship for those who are better
 off but still struggling for subsistence, then it is
 very difficult for me to believe that ... the priority
 of urgency goes to the worse off, however
 many more there are of the better off. (23)

In Nagel's view there are some people, not among the worse-off, whose well-being should be of concern whenever we try to improve the well-being of the worse-off. Thus, after specifying what counts as "struggling for subsistence," one would be forbidden to lower the well-being of people who are at that level even when one tries to improve the well-being of the worse-off. The stronger version of Nagel's view can be thought of as the "threshold norm." The threshold norm can be stated as follows: We should specify a threshold level of health and do our best to keep as many people as possible above this level. An institutional order is just with respect to health to the extent that participants in this order do not (avoidably) fail to reach this threshold.

This norm will give priority to all who are below the threshold level. However, it may impose a cost to the very sick, as it would advise against treating the few whose illnesses, although very serious, will require resource usage that deprives many others of health care, leading the less ill to fall below the specified level of health.

An acceptable threshold level of health must be viewed in light of what should count as adequate health given the current available medical technology. An account of the threshold level offered here answers two questions: First, what should be the level at which we would think someone is deprived of adequate health--that is, what counts as the threshold level? Second, when forming social policy, how should a particular distribution of health be evaluated when many fall below that level?

What Counts as the Threshold Level of Health?

In order to assess what the threshold level of health should be we need a workable way to talk about and measure health. It could be that whenever one says that he or she has good health we should be satisfied with that answer. In that case, our task would be to work simply toward obtaining high measures of people feeling and saying that they are in good health, perhaps by asking people to rank their health from 1 to 10. Unfortunately, this simple solution will not do, since it is now well established that for policy purposes ranking one's own health would lead to perverse results. One sees many instances where people report significantly less illness than that detected by clinical diagnosis. Conversely, one sees many examples of those with fewer clinically diagnosed problems reporting greater health problems. Eliciting assessments of individuals' health conditions is similar to eliciting assessments of individuals' conditions in life, such as one's position in society. Such valuations can be the result of adapting to one's condition in life and may merely, as Martha Nussbaum has recently put it, "[give] sanctity" to one's "quiet acceptance of deprivation." (24)

Len Doyal and Ian Gough assert that "health and autonomy (a large content of which can be thought of as adequate cognitive capacity) are the basic needs which humans must satisfy in order to avoid serious harm or fundamentally impair participation in their form of life." (25) According to this view, health must allow us to complete a range of practical tasks in daily life requiring manual, mental, and emotional abilities with which poor physical health usually interferes. Daniel Brock similarly argues that health must be seen with a view toward the range of opportunities we have within which we construct life plans. (26) It would be suitable within the context of this essay to understand health as being the absence of direct physiological (or psychological) hindrances in one's ability to work within a given economy, to carry out normal tasks, and to achieve higher goals in life that we set for ourselves. (27)

To accomplish a plan of life, one would need a sufficient time in which one is alive and well enough to be physically active. Thus, both a short life and a life full of pain and disability would be undesirable. A corresponding definition would be similar to Norman Daniels's notion of typical species functioning, even though he derives it from a different context. (28) Typical species functioning is defined as those biological aspects of any sentient being that engender the normal functioning of that being to be considered as a member of a natural species. Daniels writes, "impairments of these functionings reduce the range of opportunity we have within which to construct life plans and conceptions of the good we have a reasonable expectation of finding satisfying or happiness producing." (29) Daniels's view is that equality in health, in whatever form, may itself need not be a pressing goal in most circumstances, whereas equality in some of the things health can accomplish can be a central goal in social policy. For example, health would allow one to earn a living, be a part of the community, and pursue higher goals in life. A certain amount of health is essential in equalizing opportunity for people to lead a full life in society. Indeed, it may be that equal ability to do and be certain things due to good health may itself generate equality in many other spheres, such as equality in certain types of health functionings.

As material conditions differ, health requirements will also differ. Yet this does not mean that it would be meaningless to compare health achievements across countries. It would not be meaningless to assert that some have too many health resources and some have too few. Many international health programs are based on setting standards and combating the incidence of subpar health conditions. For example, the Food and Agriculture Organization of the United Nations (FAO), WHO, and UNICEF have set standards for nutritional intakes for adults and children. (30) Nearly 15.9 percent of total lost disability-adjusted life years (DALYs) and 11.7 percent of deaths in 1990 can be attributed to malnutrition, thus making it the leading cause of illness in the world. (31) This association with death makes malnourished people natural candidates for the worse-off in terms of health. Another reason why nutrition is important is that it affects adequate performance in such processes as growth, pregnancy, lactation, physical work, and resisting and recovering from disease. (32) Many health conditions are similar to malnutrition in that corrections of these conditions are required in order for a person to attend effectively to daily personal chores, to acquire employment relative to their human capital, to be an active member of a family or community, and to meet other goals in life. What nutrition is supposed to achieve is clear--yet the nutritional requirements differ by climate and different occupational opportunities available in a region.

This suggests that one can make a judgement such as the following: x is relatively more impaired due to ill health in society A, where x lives, than y is due to ill health in society B, where y lives. Assessments of impairments due to ill health need not be limited to relative comparisons, even when impairments are not understood beyond the physical terms alone but instead in terms of how impairments impede well-being or the course of actions in life. Applying an argument made by Amartya Sen: An objective assessment of health can be made because the impact of illnesses can be seen as intersubjecrive, observable, material facts about a person--facts concerning his or her physical and mental functional capabilities and environment, together with a socially shared evaluation of these facts in respect to how these factors affect the person's quality of life. (33)

Empirically we are not likely to observe dramatic differences in the assessment of health in terms of what can be achieved with particular physical conditions across cultures. This fact does not imply that from the outset we should conclude that health conditions would be viewed the same way across all cultures. The basic requirement is that health conditions be evaluated with respect to how health realistically facilitates people's plans of life and methods of livelihood. It could be that if we took all available corrective factors, including infrastructure accommodations, and assumed that all necessary activities required the same manual and intellectual dexterity, then the perspectives on health conditions would be the same across cultures. Simply, some medical conditions will affect everyone in the same way no matter where they live.

Health requirements are likely to vary nontrivially across countries under current circumstances. It will be most likely agreed that children should be afforded the same chances for all possible future jobs, political offices, and opportunities. Nearly all children should be given an equal chance of survival conditional on their congenital status. A child should be considered to be especially disadvantaged if he or she, as an adult, will not be capable of qualifying for most types of employment in a given region when another child in a different region with a similar condition could obtain employment. For example, when evaluating the need for biomedical interventions in the loss of usage of limbs, we would take account of the infrastructure that handicapped people face. In most developing countries, particularly in rural settings, this illness would have devastating consequences, whereas handicapped people in the rich countries have rights to a great deal of special resources.

Thus, the threshold level of health is determined by people being able to do and be what they can reasonably expect to within their probable environments. The threshold level should not be understood to be a relative level. It is the level of health that allows members of any society to be able to be functional within the societal context they live in. As societies change, the requirements for functionality may change.

Implications of a Threshold Standard

What would follow from the view that health care delivery should first prioritize the well-being of the less well-off, but not at the expense of causing the deterioration of the health of others at or just below a threshold level? The notion of a threshold level of health is similar to the concept of a poverty line. (34) Poverty lines are typically income levels that are determined by pricing a bundle of commodities; any entitlement below the line makes life intolerable.

An essential element of measurement of poverty rates is the Pigou-Dalton condition. The condition asserts that any transfer of income from a richer person to a poorer person must be considered a better outcome, even if this kind of transfer occurs between two people who are below the poverty level. Is this necessarily a better outcome? It passes the standard utilitarian test of a better outcome if marginal utility from improved health is comparable, positive, decreasing, and continuous. Yet, if this transfer were the only alternative in an economy, we would consider the situation tragic. (35)

Restricting ourselves to the poverty line, suppose that we are to choose between two situations: in both situations, A1 is worse off in comparison to A2, with neither party doing very well; A1 does better and A2 does worse in situation one (the status quo) than they would in situation two. The priority view would commit us to helping A1 solely because the person is badly off in absolute terms and not because he is worse off in relation to someone else. It may also commit us to the position of helping the badly-off regardless of what happens to other people, providing that others do not, in consequence, become the least well-off. The priority view may adversely affect the well-being of many not-so-well-off people considerably because its focus is on how badly off the worse-off person is and not the trade-off among people who are badly off. As Sen notes regarding the poverty line, bringing someone below a poverty line should have "some absolute level of significance and to cross it is a change of some importance." (36)

Should the Pigou-Dalton condition apply to people below the poverty line? If the answer is no, then inequality measures below the poverty line are important only because they measure how far people fall below an acceptable standard of living and also gauge the amount of resources required to lift everyone who is below the poverty line above it. Thus, the focus is not on inequality but on the degree of deprivation. Any policy that would increase the number of people below this line should be a cause for concern, and further policies that help people below this line at the expense of other people below or at this line cannot be condoned. This is the case because we would be rectifying inequality at the expense of harming someone who is already badly off. This type of policy, although it could be sanctioned under the priority view, violates the spirit of Scanlon's observation regarding when we would promote equality: those who are truly badly off in terms of any measure of well-being can be helped through a transfer of resources from those who are better off without causing the latter group any substantive harm.

Suppose that a health standard can be determined and would function like a poverty line. Further suppose that one of its features would be that people live to be at least sixty years of age. We would then not help the least well-off when the resources required would be so overwhelming that we would thereby lower the health status of a great many people below this standard. Would there be any scope for the redistribution of resources within low-income countries among people who do not live beyond age sixty and are often not healthy? (37) Such redistribution would be similar to distributional changes among people below the poverty line, which, as I have argued, is unjustifiable and untenable.

Accepting the threshold view implies that we are permitted to withhold health resources from those who require a large amount of resources just to achieve adequate health status so much so that this usage reduces the health of people who enjoy only a very modest level of health.

This is not an endorsement of prioritization through cost-effectiveness criteria. To show this, suppose that a considerable amount of resources is required to keep someone at the threshold level while the alternative would require that a smaller amount be given to someone above the threshold level for the same or larger amount of gains. In this case the threshold view demands the priority be given to the person who is in danger of falling below the threshold level.

Scope for Redistribution

In light of the fact that many in the world can live to the age of eighty with good health, it does not seem unreasonable that a threshold level would be that people live at least sixty-five years of good health (this level is, of course, open to debate). Further, this is near the life expectancy of better-performing developing countries that are less affected by HIV/AIDS. It is for these countries where we find general acceptance that the middle class has escaped poverty. It is reasonable to think that this could be the threshold level for many countries.

It would be unreasonable to expect a redistribution of health resources within developing countries, where the vast majority of the middle class receives reasonable but not state-of-the-art medical care. (38) Take an example: in the course of care for the highly prevalent problem of adult diabetes in low-income developing countries, much of the middle class is usually restricted to dietary control instead of insulin supplement; similarly, for many cardiovascular problems medication is scarce. Even this low level of care received by the middle class and the diagnoses of these types of problems are not usually available to the poor. (39) Under these conditions it is unreasonable to expect that health resources should be diverted away from people who have access to just enough health resources to enable them to live, say, only until the age of fifty-five with a middling level of health, toward those who are even less fortunate. Thus, health-sector reforms in developing countries, which are often implemented with the aim of making health resources more accessible to the poor, should have as a side constraint the preservation of the health status of these pretty badly off people. Unfortunately, health reform legislation in developing countries that have not yet completed epidemiological transition often urges the reduction of resources devoted to making tertiary care available, thereby reducing the already low level of care that is available to people who are badly off even if nonpoor.

Thus, if we adopt the threshold view that I have outlined, there will be very little scope for resource redistribution within societies where the large part of the population hovers around the threshold level or falls below it. Therefore, so long as any redistribution can take place, it must (with the exception of relatively wealthy developing countries with glaring within-country inequality) come from developed to developing countries. This is because many people in developed countries enjoy good health for many years, often at great cost, while many in poor coun tries face an extremely high probability of death throughout their lives that can be lowered at a low (though not insignificant) cost. (40) The highest-cost care is most often received late in life. One could therefore start by limiting such high-end care to make the resources saved here available to bring those in poorer countries up to (or at least nearer to) the threshold level. One measure for achieving such redistribution would be to allow changes in patent laws that may make drugs in developing countries more readily available, even while having some negative impact for the richer nations by lowering the incentives for new production of drugs and reducing the incomes of those involved in pharmaceutical industries.

The resources required at present to achieve equality in health status at the high Western level would probably be a bottomless pit, given that nearly $2,000 annually per capita is already spent on health in industrialized countries. However, the amount of resources required to achieve a reasonable level of health for nearly all people is likely to be finite and is well within 10 percent to 12 percent of world GDP, although it may require limiting the resources devoted to some high-end care in industrialized countries.

The resource transfers from developed to developing countries demanded by the threshold view may nevertheless be somewhat limited. This is so at least if one adopts a threshold view that is sensitive to differences in social context, since there may be greater health resource requirements in developed countries even aside from those that arise due to price differences. Nevertheless, it is possible to identify many types of care that have high levels of resource utilization without yielding much benefit. The resources expended on the provision of such care could instead be redistributed to those who might thereby be brought up to the threshold level.

(1) World mortality figures by countries, region, and the associated national income closely correspond to measures of disabilities. See Christopher J.L. Murray and Alan D. Lopez, eds., The Global Burden of Disease: A Comprehensive Assessment of Mortality and Disability from Diseases, Injuries, and Risk Factors in 1990 and Projected to 2020 (Cambridge: Harvard University Press, 1996).

(2) This term is borrowed from Thomas Pogge, "Can the Capability Approach Be Justified?" in Martha Nussbaum and Chad Flanders, eds., "Global Inequalities," Philosophical Topics 30, no. 2 (2002), pp. 167-228.

(3) Derek Yach, "Health for All in the Twenty-First Century: A Global Perspective," National Medical Journal of India 10 (1997), pp. 82-89; available at www.who.int/archives/hfa/techsem/971001.html.

(4) Timothy Evans et al., eds., Challenging Inequities in Health: From Ethics to Action (New York: Oxford University Press, 2001), p. xiii. Most methods for measuring inequality emphasize giving higher weight to the least well-off; some ignore the closeness of the difference altogether and focus only on the relative position of well being. Technically, inequality measures from two different regions are comparable in welfare terms only if the average well-being is the same; in practice this is often ignored. See Sudhir Anand and S. Nanthikesan, "A Compilation of Length of-Life Distribution Measures for Complete Life Tables," Harvard Center for Population and Development Studies Working Papers 10, no. 7 (2000); and Adam Wagstaff, "Inequality Aversion, Health Inequalities, and Health Achievement," Journal of Health Economics 21, no. 4 (July 2002), pp. 627-41.

(5) Davidson R. Gwatkin, "Assessing Inequalities in Maternal Mortality," Lancet 363, no. 9402 (January 3, 2004), pp. 23-27.

(6) Why the rate for the highest level of income should be considered the ideal, and not some higher level of maternal health for all, is left unexplained.

(7) For a focus on health inequality, see WHO, World Health Report2000 (Geneva: WHO, 2000); and Timothy Evans et al., eds., Challenging Inequalities in Health. For measurements, see, e.g., Emmanuela Gakdiou, Christopher J. L. Murray, and Julio Frenk, "Defining and Measuring Health Inequality: An Approach Based on the Distribution of Health Expectancy," Bulletin of the World Health Organization 78, no. 1 (2000), pp. 42-54; and Adam Wagstaff, P. Paci, and E. van Doorslaer, "On the Measurement of Inequalities in Health," Social Science and Medicine 33, no. 5 (1991), pp. 545-77.

(8) Horizontal equity refers to treatment of like conditions in a like manner. It is not clear what relevant factors make up the like conditions.

(9) It has been argued that agents are responsible for rectifying serious deprivations when and to the extent that they have contributed to bringing about these deprivations. I will not argue here against this "contribution" principle for allocating responsibility. For a discussion of this principle and its practical implications, see Christian Barry, "Applying the Contribution Principle" Metaphilosophy36, no. 1 (2005), forthcoming. It must be pointed out, however, that in a poor region it could be that agents who suffer from deprivation may have brought about, even if unwittingly, each other's deprivations. Assigning responsibility according to this principle brings about no change unless, as in many cases, such culpabilities also lie with well-off people, some of whom may live outside the region. If we nevertheless find these deprivations to be morally unacceptable, a view that assigns obligations of distributive justice to rectify outcomes irrespective of how these outcomes came about would be better able to assign responsibilities that can bring about significant reductions in incidences of these deprivations. Such responsibilities would extend beyond borders. I cannot defend the claim that such responsibilities exist here, but I shall provide a plausible account of responsibilities regarding international health where health budgets are limited and where any reduction in the health of those who are well-off induces significant reduction in well-being.

(10) World Bank, World Development Report 1993 (New York: Oxford University Press, 1993).

(11) Juan Luis Londono and Julio Frenk, "Structured Pluralism: Towards an Innovative Model for Health System Reform in Latin America," in Peter Lloyd-Sherlock, ed., Healthcare Reform and Poverty in Latin America (London: Institute of Latin American Studies, 2000).

(12) Jean Dreze and Amartya K. Sen, eds., Indian Development: Selected Regional Perspectives (Delhi: Oxford University Press, 1997).

(13) K. R. Thankappan and M. S. Valiathan, "Health at Low Cost--The Kerala Model," Lancet 351, no. 2 (1998), pp. 1274-75.

(14) World Bank, World Development Report 2004 (Washington, D.C.: World Bank, 2004), Indicator Table 3, CDROM.

(15) I am sidestepping the debate regarding the correlation between income and health. I believe there is no convincing argument to dispute that in many parts of the world the poor are the most ill and that the gradients in improved health status and income move in the same direction.

(16) Thomas M. Scanlon, "The Diversity of Objections to Inequality," Lindley Lecture, Department of Philosophy, University of Kansas, 1996.

(17) Thomas Nagel, Mortal Questions (Cambridge: Cam bridge University Press, 1979); and Derek Parfit, "Equality or Priority?" Lindley Lecture, Department of Philosophy, University of Kansas, 1991.

(18) Nagel, Mortal Questions, p. 123.

(19) The magnitudes of losses are understood here not in terms of years but in terms of their intuitive moral importance. A decrease in life expectancy of fifteen years for society A may be less of a loss than a decrease of life expectancy of ten years for society B if society A is a great deal better off than society B is at present.

(20) Parfit, "Equality or Priority?"

(21) It also differs from the difference principle as elaborated by John Rawls by noting that not just any least well-off person qualifies for assistance; the least well-off has to be in bad condition in order to be helped.

(22) Francis Kamm, "Deciding Whom to Help: Resource Prioritization, Population Health Measures, and Disability" in Fabienne Peter, Sudhir Anand, and Amartya Sen, eds., Public Health, Ethics, and Equity (Oxford: Oxford University Press, 2004).

(23) Nagel, Mortal Questions, p. 125.

(24) Martha C. Nussbaum, Women and Human Development: The Capabilities Approach (New York: Cambridge University Press, 2000), p. 139. See also Jon Elster, "Sour Grapes: Utilitarianism and the Genesis of Wants," in Amartya Sen and Bernard Williams, eds., Utilitarianism and Beyond (Cambridge: Cambridge University Press, 1982); and Amartya Sen, "Positional Objectivity," Philosophy & Public Affairs 22, no. 2 (1993), pp. 126-45.

(25) Len Doyal and Ian Gough, A Theory of Human Need (Basingstoke: Macmillan Press, 1991),p. 73.

(26) Daniel Brock, "Quality of Life Measures in Health Care and Medical Ethics," in Martha C. Nussbaum and Amartya Sen, eds., The Quality of Life (Oxford: Clarendon Press, 1993). This view is also expressed in Amartya Sen, The Standard of Living (Cambridge: Cambridge University Press, 1987).

(27) Goals in life cannot require a large amount of resources relative to societal constraints on resources; similarly, good health cannot be defined as being fit to climb Mount Everest. See John Rawls, "Social Unity and Primary Goods," in Sen and Williams, eds., Utilitarianism and Beyond, pp. 159-86.

(28) Norman Daniels, "Health-care Needs and Distributive Justice," Philosophy & Public Affairs 10, no. 2 (1981), pp. 146-79.

(29) Ibid., p. 185.

(30) Food and Agriculture Organization of the United Nations, "Mapping Undernutrition" (Rome: FAO, 2000).

(31) See Murray and Lopez, eds., The Global Burden of Disease, table 6.2 (p. 311), table 6.3 (p. 312).

(32) Helen Young, "Nutrition and Intervention Strategies," in Stephen Devereux and Simon Maxwell, eds., Food Security in Sub-Saharan Africa (London: ITDG Publishing, 2001).

(33) Sen, "Positional Objectivity."

(34) Some poverty lines are set in terms of cutoff points in a distribution; we suggest a poverty line to mean being able to afford a bundle of goods, as elaborated by Molly Orshansky; see Linda Barrington, "Estimating Earnings Poverty in 1939: A Comparison of Orshansky-Method and Price-Indexed Definitions of Poverty," Review of Economics and Statistics 79, no. 3 (August 1997), pp. 406-14.

(35) Martha Nussbaum, in "The Costs of Tragedy: Some Moral Limits of Cost-Benefit Analysis," Journal of Legal Studies 29, no. 2 (June 2000), pp. 1005-36, points out that a superior outcome in terms of cost-benefit analysis may still impose a terrible outcome, and this should be considered a tragedy.

(36) Amartya Sen, "Poor Relatively Speaking" in Resources, Values and Development (Cambridge: Harvard University Press, 1984), p. 342.

(37) To take another example, in 1986 the child mortality rate in Bangladesh stood at 173 per 1,000 among the poorest group, while for the richest group it stood at 98. Abbas Bhuyia, ICDDRB, Bangladesh, quoted in Adam Wagstaff, "Inequalities in Health in Developing Countries: Swimming against the Tide?" World Bank, Working Paper no. 2795, March 5, 2002; available at econ.worldbank.org/files/12290_wps2795.pdf. Compare the last number for the similar figure in the industrial countries: an estimate of 14 or 9. While it is possible to argue that statistics for Bangladesh are informative for policy purposes, when compared to the industrial countries these inequality figures induce no moral imperative.

(38) Hospitalization among the poor is rare and most of the middle class reported receiving significantly costly and sometimes insufficient care. Reported in Ajay Mahal, Janmejaya Singh, Farzana Afridi, Vikram Lamba, Anil Gumber, and V. Selvaraju, "Who Benefits from Public Health Spending in India?" NCAER Special Report (New Delhi: NCAER, 2000).

(39) The high prevalence of theses chronic illnesses is well documented for developing countries; see Murray and Lopez, eds., The Global Burden of Disease. The account of availability of care can be found in Deepa Narayan, Voices of the Poor (New York: Oxford University Press, 2000); see also Thankappan and Valiathan, "Health at Low Cost," and personal accounts given to the author in India in his fieldwork and in Tanzania accounts given by government and NGO officials.

(40) See WHO, World Health Report 2000. I have argued the scope for redistribution within developing countries is limited, although there are many dramatic cases of insensitive behavior of the extreme rich in such countries. I note that the scope for achieving a desirable distribution of well-being exists through greater emphasis on intercountry redistribution rather than through intracountry redistribution.