For the Sake of Justice: Should We Prioritize Rare Diseases?
- 652 Downloads
This article is about the justifiability of accepting worse cost effectiveness for orphan drugs, that is, treatments for rare diseases, in a publicly financed health care system. Recently, three arguments have been presented that may be used in favour of exceptionally advantageous economic terms for orphan drugs. These arguments share the common feature of all referring to considerations of justice or fairness: the argument of the irrelevance of group size, the argument from the principle of need, and the argument of identifiability. It is argued that all of these arguments fail to support the conclusion that orphan drugs should be subsidized to a larger extent than treatments for common diseases. The argument of the irrelevance fails to distinguish between directly and indirectly relevant considerations of fairness or justice. The recent attempt to revive the moral relevance of identifiability has provided no novel reasons to think that identifiability is morally relevant in itself or due to considerations of fairness and justice. The argument from the principle of need does not fail due to any inherent flaw in the principle as such. Rather, this principle can be interpreted in different ways, and none of these interpretations support exceptionally advantageous terms economically for treating rare diseases specifically. It is concluded that we are awaiting justice based reasons for the preferential treatment of orphan drugs.
KeywordsCost effectiveness Identifiability Justice Orphan drugs Prioritization Rare diseases
Rare diseases are often singled out as an urgent area for medical research and drug development. Orphan drugs, i.e., biomedical treatments for rare diseases, are already today subject to economically advantageous regulations in most developed nations. The research on and development of these drugs are in various ways facilitated by economic incentives, e.g. tax cuts and extended monopoly for approved drugs . Moreover, once on the market, many both publicly and mainly privately funded health care systems in practice accept worse cost effectiveness ratios regarding orphan drugs than drugs for common but equally serious diseases . This article is about the justifiability of these kinds of economically favourable exceptions for rare diseases. The main question of focus is: should a publicly financed health care system accept a worse cost effectiveness, e.g. per saved life or QALY, regarding treatments for rare diseases (than for diseases that do not fall within this category)?1
The question of whether these economically favourable exceptions for rare diseases are defensible has been the subject of much debate.2 Hence, a thorough analysis of all the arguments for and against such exceptions is not feasible within a single article. Rather, I will focus on three recent ones that have not yet been scrutinized and which all refer to considerations of justice or fairness. Hence, the main claim of the article is not that we should not grant orphan drugs exceptionally advantageous terms economically, but the more modest one, saying that recent arguments of justice to the effect that we should all fail.
Two more simplifying limitations should be noted. First, I will confine myself to publicly funded and mainly publicly run health care systems. In these systems, allocations of resources are more explicit and often made overtly with consideration to cost efficiency (e.g. in Scandinavia and the UK) [16, 26]. I think that the basic problem is really the same within systems with more elements of private insurance, but questions of allocation are often more covert in these systems.3 I will therefore presume a health care system that openly recognizes that there is a limit to health care resources.
Second, I will focus on which treatments to offer in such a system, not how much to spend on research and development of new treatments. I do think that nothing in the arguments really hinges on this: they are all possible to use for and against any kind of economic advantages directed towards rare diseases, including the above mentioned subsidizing measures bestowed on developing orphan drugs. However, in order to keep things neat and simple, I will focus on economic advantages for treatments of rare diseases, advantages, then, relative to treatments of other non-rare diseases.
There are different definitions of rare diseases (or conditions). I will use the official one from the European Union and refer to rare disease as diseases afflicting no more than 5 out of 100,000 in the population European Commission 141/2000 .4 Moreover, I will confine myself to those rare diseases that are defined as ‘serious’ according to the European definition of orphan drugs, i.e., that are “life threatening or chronically debilitating” (European Commission 141/2000 , Article 3). The reason for this focus is that it is these conditions that has been the focus of debate, especially, as we will see, from the point of view of the considerations of justice discussed below.
Already at this point, it should also be noted that question is not if a publicly financed health care system should accept a worse cost effectiveness regarding treatments for serious diseases5 (even though I focus on unusual and serious diseases), although many rare diseases are serious indeed. However, many rare diseases are not serious and, more importantly, some common diseases and health problems are very serious too, like cardiac arrest, or prostate and breast cancer. So, in order to be more precise, the question is if rare diseases should be subject to economically favourable exceptions because they are rare (or because they have some other property extensionally equivalent, at least almost, to being rare). What I am after, then, is some sort of justification for the preferential treatment of rare diseases.
On the Need for a Renewed Look at the Issue
As already mentioned, the question of whether the economically favourable exceptions for rare diseases are defensible has been the subject of much debate. However, there are both practical and theoretical reasons to address the issue again. The practical reasons are all related to the increasing size and share of health expenditures on rare diseases. First, new rare diseases are constantly being identified.6 Second, and more importantly, the trend of personalised medicine is picking up pace. The rapid development of genetics has started a trend of re-classification of common diseases and diagnoses into more specific subcategories, e.g. depending on the fact that patients with different genomes are susceptible to different treatments . Some traditional common diseases, like for instance cancer, diabetes and arthritis, therefore are increasingly populated by new patient subgroups, each with a rare disease of their own. Third, treatments for rare diseases are being developed in an unprecedented rate, not least because of the economically advantageous regulations for them .
At the same time, there are no signs of orphan drugs becoming cheaper. Some parts of the costs are due to the existing regulations, like the extended monopoly for manufacturers of orphan drugs. But even if this were to be discounted, a part of the increased cost for orphan drugs are essentially related to them being drugs for rare diseases. The cost for developing any drug is very high and to a significant extent independent of the number of potential patients (about $1 billion on an average) .7 Being reimbursed for the cost of investments demands that the price for each individual will be higher the fewer they are. Hence, the price for each patient with a rare disease will typically be higher than the one for more common diseases.
These are a few reasons why costs for rare diseases constitute an increasing part of health expenditures; the opportunity cost of orphan drugs hence must be taken into account . The resources spend on developing and offering these treatments could be put to use in other areas of health care where it would result in more benefits, e.g. save more lives. Hence, how we spend the limited health care resources are is essentially an ethical issue .
Consider, for instance, Hunter syndrome, which have an incidence of approximately 1 in 130,000, i.e., a very rare disease . Hunter syndrome is an X-linked recessively inherited very serious disease caused by an enzyme deficiency, with symptoms including stiff joints, breathing difficulties, and progressive mental deterioration. There is an orphan drug ameliorating most of these symptoms, Elaprase, largely improving affected boys’ mobility and general health related well-being, helping them to move around and play like other children. In fact, Elaprase efficiently deals with all symptoms except the ultimately lethal one—neurological progressive deterioration (since it cannot break the blood–brain barrier).
Hunter is illustrative in several ways. First, like 80 % of rare diseases, it is monogenetic, i.e., caused by a single mutation . Second, the treatment for it is very expensive: the cost for Elaprase per patient is between $300,000 and $400,000 a year.8 Third, the treatment has limited effects. Although it no doubt has great impact on quality of life for those affected, it cannot influence life expectancy.
Many orphan drugs are even less efficient than this. In order to exemplify the above mentioned trend towards fine graining classification of common diseases, take Erbitux, which have been used to target a form of rare lung cancer. A treatment increases average survival with 1.2 months and costs about $80,000 . This kind of very expensive treatments for rare subtypes of cancer with measurable but very limited life preserving effects is by no means unique: for instance, Zelboraf restores certain subtypes of malignant melanoma almost completely, but only for about 8 weeks on an average to a cost of approximately $22,000, i.e., $143,000 per saved life year .9
However, there are also orphan drugs that are more efficient regarding all sorts of outcomes, both in comparison to Elaprase and Erbitux. For instance, Cysteamine both improves functioning and increases life expectancy by years by postponing the need for renal transplants in cystinosis patients . In sum, one cannot say anything general about the cost efficiency of all orphan drugs—some are better in this regard than others. Hence, any successful argumentation to the effect that we should not accept exceptionally low cost effectiveness for orphan drugs does not imply that we should not accept the cost for any orphan drugs—many orphan drugs will certainly be as cost efficient as the base line of cost efficiency a publically funded health care system settles for.
Still, only to put these examples into context, let us compare cost efficiency with a few treatments for common but serious diseases. A classic example10 is to treat coronary heart disease with cholesterol reducing statins, which, administered to the proper population, costs $15,000 per QALY (i.e., life years with full life quality in terms of the measurement in question) or 20 saved life-years per yearly Elaprase patient .11 Another example is certain cheap drug treatments of hypertension, which if left untreated may cause myocardial infarction or stroke, both common and potentially lethal; such treatments have a QALY of approximately $200 (!), once again if the target group is properly demarcated.12
These examples of treatments for common diseases share a few characteristics: they are directed towards conditions that are a far distance from being rare diseases, the conditions are serious indeed (lethal in many instances), the treatments are relatively cheap, directed towards a larger population, and very much more cost efficient than all of the mentioned examples of treatments for rare diseases. Moreover, it is impossible to say, both before and after treatment, whom in the population that actually has benefited from the treatment, i.e., who would not have survived were it not for the treatment. We only know for a fact (or almost with as high a level of evidence as is possible in medicine) that this goes for a significant part of the treated population. The characteristics of the examples are not accidental, as we will see: there will be reasons to return to them in the ensuing discussion.
The theoretical reason for taking another round with the question of economic advantages for rare diseases is that lately several novel arguments have been presented in favour of such advantages. These arguments have in common that they all refer to considerations of justice or fairness. The main part of this article is devoted to addressing these arguments. Some of these arguments have been presented directly in relation to the issue of rare diseases , while others have been presented in a more general philosophical debate on basic principles of rightness and justice . However, also the latter are clearly relevant for the present discussion, as we will see. Hence, we are entering an intriguing area—a practical problem that immediately reaches the heart of several discussions on basic principles in ethics.
The Argument of the Irrelevance of Group Size
…irrelevant group belonging or group properties should not affect a patient groups’ equal opportunities and equal outcomes regarding health. Whether or not a patient group is small or large is such an irrelevant group property and the groups’ size or the rarity of the condition should therefore not affect the groups possibilities to treatment… compared to other larger groups… 13 , 14
The authors explicitly state that this is a consideration of justice: “when worse cost efficiency foremost is due to… the size of the group to which the patient belongs… there is just cause to… accept a worse cost efficiency” , my italics. More specifically, they are thinking about the formal principle of justice , p. 422, saying that if there is no relevant difference between two cases, they should not be treated differently .
This line of reasoning, however, fails to distinguish between directly and indirectly operative factors. Cost directly affects patients access to treatments and, most agree (even though there are some disagreements about how), should do so in a publically financed health care system with limited resources.15 That is, most agree that cost is a relevant factor in determining health care prioritizations, besides being directly operative.16
However, a lot of factors, in turn, affect cost, i.e., they are indirectly operative, and most (if not all) of these factors are plausibly considered to be morally irrelevant in themselves. One candidate would be the number of patients having a certain disease, i.e., patient group size. However, it would take an unduly restrictive view on what factors may legitimately affect prioritizations not to allow indirect but in themselves irrelevant factors to affect direct and relevant factors.
To see why, consider the range of factors affecting cost. For instance, the cost of developing new treatments depend on the present level of scientific development in different areas of medicine, something which is due to contingent facts about, for instance, past scientific endeavours and funding. Likewise, the funding needed to develop new drugs is partly determined by human beings’ ability to metabolize them. None of these factors can plausibly be seen as themselves relevant factors for just prioritizations. However, they do affect cost.
The same goes for the efficiency of drugs—it is due to a lot of genetic and other factors that patients cannot reasonably be held responsible for or the allocation of which is a matter of justice. Still, no one denies that efficiency is an important factor in determining prioritizations and allocation of health care resources. So, while group size can be plausibly considered to be a factor that is, in itself, morally irrelevant, it may affect another factor that is morally relevant, e.g. cost or effectiveness. To discount for all factors that affect cost but are in themselves morally irrelevant would tantamount to discounting cost altogether.17
This is not to say that no factors affecting cost should be compensated for. For instance, if certain costs are due to manmade injustices, e.g. demanding the ‘genetically inferior’ to cover a larger portion of the costs for their medications themselves, this is could be just cause for adjusting cost (or preferably changing the institutional arrangements responsible for the injustice). However, none of the mentioned factors are of this kind, including rarity of disease.
Arguments Based on Principles of Need
Equity or equality is likely the most common kind of consideration of justice forwarded in favour of economic advantages for orphan drugs [27, 28]. In discussions on prioritizations in health care such considerations are often couched in terms of need: that needier patients should have precedence over less needy ones . This is also so regarding the discussion on prioritization of orphan drugs . The very thing that unites different principles of need regarding prioritizations in health care18 is the broadly egalitarian idea that we should prioritize the worse off (i.e., the needier), even if this leads to a net loss of goods totally , p. 35. This means that principles of need favour more egalitarian outcomes than would classic total utilitarianism, according to which we should maximize the total of net benefits.
However, there are four more specific interpretations of principles of needs, with different ideas on how to prioritize health care interventions: sufficientarianism, prioritarianism, outcome egalitarianism, and opportunity egalitarianism. In the following, I will argue that none of these principles can be used to justify economic advantages for orphan drugs, i.e., drugs for rare and serious diseases.
Probably, the most common way to interpret principles of need is as sufficientarian principles .19 Sufficientarianism is the view that there is a level of goods that we have special reasons to see to it that everyone reaches. However, beyond that level there are no (or much weaker) reasons to see to it that people have more goods.20 That is, everyone should get sufficient of the good. Regarding prioritizations within health care, the most famous sufficientarian is likely Norman Daniels, who has argued that we have special reasons to see to it that we reach the level of health identical to the level of normal species functioning .21
However, even if one were to disregard all the inherent shortcomings of sufficientarianism,22 this principle cannot be used to support economic advantages for orphan drugs. The main reason for this is that sufficientarianism itself does not say how we should prioritize health care or allocate health care resources below the threshold of a sufficiency. And in this context we are dealing with serious diseases, whether common or rare, i.e., conditions that, left untreated, results in a level far below any suggested level of minimal health. In those cases sufficientarianism is coupled with suggestions on how to prioritize below the level of sufficiency, these are usually egalitarian  or prioritarian  suggestions. I will deal with them separately below.
One other possible suggestion on how to distribute health care that seems to fit naturally with the sufficientarian spirit is that as many as possible have a sufficiency (when everyone cannot have it, as is the case with health).23 However, such a principle would most likely have us prioritizing common and serious diseases over rare and serious ones (if we have to make a choice), since this most likely will bring more people over the level of sufficiently good health (as the examples in the introduction suggest). Hence, if the sufficientarian interpretation of the principle of need gives any guidance at all on whether to subsidize rare diseases more, it likely speaks against it.
Principles of need may also be interpreted as expressing some version of prioritarianism [6, 25]. Prioritarianism says that benefiting people matters more the worse off they are.24 Improvements for those worse off should, then, be weighted to their advantage. Different weights on the value of benefiting the worse off result in different versions of prioritarianism. Nonetheless, all versions of prioritarianism allow the needs of the better off to override the needs of the worse off if their improvements are great enough or if the better off are numerous enough [6, 25].25 And likewise, the needs of sufficiently many override the needs of the few when these groups are equally bad off. In this way, prioritarianism is similar to utilitarianism. This feature of prioritarianism also explains why it cannot be used to prioritize rare diseases (because they are rare): if there is a choice between treating the few and the many and everything else is equal, prioritarianism favours treating the many . Of course, prioritarianism would give higher priority to the worse off, so seriousness of disease is (most likely) relevant: generally, we ought to give more weight to patients the more serious disease they have, even if this would mean a net loss of, e.g. QALYS. But, again, we are not in a context comparing the many with small needs with fewer with great, but many with great needs with fewer with great needs.
Another interpretation of the principle of need is that it says that we should spend health care resources in order to achieve as equal outcome regarding health as possible: “[W]e should aim at an equal as possible level of health for all citizens (which presuppose that we primarily focus on those who are farthest from such a level—i.e., those with greatest needs)” .
The first part of the quote is straightforward enough—it states that equality in terms of health is a desirable goal, i.e., outcome egalitarianism regarding health is stated. However, as the continuing quote within brackets suggest, it is tempting to conclude that outcome egalitarianism imply that we should treat those worst off first. However, this is not always so (as we will see shortly). Moreover, and more importantly in this context, if many seriously ill people are denied treatment so that we can afford it for the few, this may make things worse from an egalitarian perspective and not only from a benefit-maximizing one.
The reason for this is that (in)equality is a much more complex concept than is usually assumed : there are many aspects of equality i.e., ways in which outcomes may be more or less (in)equal (or better or worse in terms of (in)equality, as Temkin would have it). This is also demonstrated by the fact that there are different measurements of equality that ranks different outcomes differently regarding (in)equality. Let me elaborate.
It is no more likely that prioritizing rare diseases over common ones would increase the number of (more) equal relations than prioritizing common diseases would. Rather, we could get a larger number of equal relations on a lower level instead of a larger number on a higher level.26 On some aspects of equality, this could be desirable, for instance, since “egalitarians may be most offended if just a few are badly off while the vast majority are well off because the inequality then seems particularly gratuitous” , p. 18. However, a situation with more people (roughly) equally worse may be a greater deviation from perfect equality, as more individuals then have complaints for being worse off, which would, in one way then, be worse in terms of equality , pp. 19–20.
Moreover, as the examples of treatments for rare and common diseases above indicated, we are likely not comparing situations with an equal number of persons, but one with variations in population size, since prioritizing common diseases likely will improve number of life years saved. Plausible versions of egalitarianism would probably take population size into account, e.g. so called positive egalitarianism.27 Increasing the number of individuals who are (roughly) equally good off will improve the situation in terms of equality on positive egalitarianism.
Outcome egalitarianism may not favour orphan drugs, even if we were to assume, contrary to facts, that the few would be even worse off by not being helped than the many: the same amount of equality may be achieved by clustering as many as possible around a higher equal level than would be achieved in an alternative distribution with some even worse off but more relations of inequality between the rest. Imagine, for instance, measuring health on a scale ranging between 1 and 10 and two alternative distributions of health (each number representing one individual): then the first kind of situation could be illustrated by 2, 5, 5, 5, 7 and the second by 3, 4, 5, 6, 7. Comparing the sum of all health differences for all distinct pairs of individuals in these two populations (i.e., Total Pairwise Welfare Difference, see ) would result in the same number (20). For instance, using Average Pairwise Welfare Difference or the Gini coefficient would yield the same conclusion .
In sum, we would have to know much more in order to take a stand on what outcome egalitarianism would say regarding prioritizing orphan drugs, both on empirical and normative matters. First, we would have to know more about how it would affect the level of health of the individuals affected, compared to alternatives. And we would have to ponder more carefully what aspects of equality matters most when in conflict. Here, I have merely presented a few general remarks that indicate that such a careful analysis would not necessarily favour rare diseases but, rather, the opposite.
The principle of need has also been interpreted as saying that individuals should have equal opportunities to achieve (good) health: “The principle of need… partly expresses that all citizens should have equal opportunities to achieve good health…” .28 This idea has also been used to argue in favour of granting economic advantages to orphan drugs: “If we let the cost of development of drugs for rare conditions affect the opportunity to gain access to these drugs the citizens are denied equal opportunities to achieve good health” .
The principle of equal opportunities itself, however, is open to a range of possible interpretations. One possible interpretation is that “all patients should have access to all available treatments” , i.e., that we should never make prioritizations. Of course, this is a non-starter, since it is “impossible from the point of view of resources” .
Another non-starter would be to say that equal opportunity to health means equal probability of achieving health, since a great number of patients have the probability of zero29 of achieving anything close to health (like persons with Downs, Hunters, or many other chronically disabled patients)30: we would not want to level down everyone else to such a degree in order to achieve equal probability of health.31 Similarly, equal probability to get treatment would be best achieved by a giant health care lottery with equal chances of ‘win’ (treatment), regardless of cost, efficiency, prognosis, need etc. No one has and no one would, upon reflection, suggest such folly.
A more plausible interpretation of opportunity egalitarianism in the luck egalitarian spirit  would be to say that we should organize health care so that everyone would, as far as possible, have equal health were it not for personal choices for which they reasonably can be held responsible. However, such an interpretation would not favour granting economic advantages to orphan drugs, since it would be outcome egalitarianism coupled with a principle of responsibility (it should be noted, then, that luck egalitarianism is not a version of sufficientarianism, but a version of egalitarianism).32 And we have already seen why outcome egalitarianism will not do the job in any straightforward way.
Another interpretation is suggested in the following quote: “In order to realize the idea of equal opportunities it is required that different groups of patients have the same possibility as every other patient group to treatment for condition they suffer from” .33 As it reads, it says that different patient groups rather than different patients should have equal opportunities. This could be understood as saying that different diseases 34 should have equal opportunities to get treatment. Such a suggestion could support prioritizing rare diseases, but only if ‘equal opportunities’ is interpreted as equal probability of getting treatment for the condition: then rare diseases must be subsidized in order to equalize probability for patients to have access to them. However, this means (1) making diseases rather than patients the proper object of prioritizations, and this, again, (2) regardless of cost, efficiency, prognosis, etc. Especially the second implication is hard to accept, to say the least.
The fact that a treatment is less cost effective due to the small size of the patient group in this context seems irrelevant to take into consideration, since it would mean that small patient groups are denied equal opportunities to achieve good health .
However, this is only another way to formulate the argument of the irrelevance of group size, which, as we have seen, also fails in the end.
The Argument of Identifiability
The most promising line of defence, if successful, for granting economically advantageous terms for orphan drugs would likely be to refer to the moral relevance of identifiability of patients. This is so, since patients with rare diseases regularly belong to small, well-defined groups of conditions: we often know quite precisely which individuals would benefit from orphan drugs [20, 21], unlike some treatments for e.g. coronary heart disease or hypertension.35
In discussions of rare diseases, considerations of identifiability are typically discussed by reference to the rule of rescue [17, 34].36 This rule is formulated in various ways, often including definitional characteristics such as being an intuitive response to a situation, where an identifiable individual or group is facing a preventable threat to their existence . In this context, it is the component of identifiability that is discussed—other possible definitional characteristics of the rule of rescue are ignored (although we are also still in the context of prioritizing between treatments for conditions that are serious indeed, often lethal).
The question of the moral relevance of identifiability has itself been subject to much philosophical debate over the last decades.37 I think it would be fair to say that most authors have argued that identifiability does not carry any moral weight (unless, perhaps, for indirect reasons, such as being an indication of uncertainty regarding effects). However, recently Scott M. James attempted to revive the idea that identifiability is morally important; in an article in Utilitas, he argues that not taking identifiability (or, as he calls it, determinacy) into account can be unfair or unjust: “When potential beneficiaries are determinate, they have a special claim on us grounded in fairness” .
However, there are many senses in which an individual can be (un)identifiable. Sometimes, an individual is not identifiable due to our lack of knowledge: with further investigations we could (in principle) learn who the person in question is. Other times, as is the case with for instance (unrigged) lotteries, we cannot identify the winner before the result of the lottery is determined, but well after. However, sometimes we cannot even identify the beneficiary even after the realization of the benefit, what James calls metaphysical unidentifiability. From now on, I will call a metaphysically unidentifiable individual a statistical person and any individual who is not statistical in this sense I will call an identifiable person. Since being a statistical person is the strongest form of unidentifiability presented, presumably, the case for such a person having less moral weight is stronger (e.g. than if the person is merely unidentifiable due to present lack of knowledge that could be remedied). I will therefore, like James, stick to this kind of unidentifiability.
Let us get a firmer grip on what kind of unidentifiability we are dealing with then. Regarding the statistical persons who can be benefited by a certain course of action, we can say to what group of individuals they belong. However, we cannot say who in this group has benefited even after the benefits are realized (ex post), i.e., we cannot say what individuals would have been worse off (e.g. died) if we had not administered the treatment to the group in question. This is the case, for instance, with the treatments for coronary heart disease and hypertension mentioned in the outset: it is impossible to say, both before and after treatment, whom in the population that actually has benefited from the treatment, i.e., who would not have survived was it not for the treatment.
This kind of unidentifiability is easily conflated with lack of reliable evidence for the efficacy of treatment: you may think that the inability to identify those who benefit is due to uncertainty regarding to what extent there are beneficiaries. However, this is wrong: we know with at least as high a degree of certainty that a number of individuals who would otherwise have died prematurely will not do so as with treatments where these individuals can be identified both before and after treatments, such as Imiglucerase treatment for Gaucher’s disease38: the treatments of these common diseases are saving lives, we just do not know whose lives.39
The Moral Relevance of Identifiability
Now that we have clarified what sense of identifiability we are dealing with, let us turn to the crucial question: does the identifiability of one’s (potential) beneficiary make a moral difference? Yes, says James, since not benefiting the identifiable person may be to treat her unfair.40 In order to see this, he asks us to consider “a case in which one’s decision hinges exclusively on the putative significance of determinacy” , a thought experiment where a cruise-liner with 1,000 passengers are taken hostage by a group of terrorist determined to kill 499 of the passengers with the following options for you to choose: “On option A, the terrorists will choose at random … 499 passengers to kill… On option B, the terrorists will kill no. 1,000 along with 498 randomly chosen passengers…” . Passenger 1,000 was the only passenger strolling on the deck when the ship was taken hostage, while all other passengers were asleep in their rooms on sleeping deck. This means that “on Option A, no. 1,000 life will be spared; the 499 passengers to be chosen randomly will be chosen from the sleeping deck only” .
Obviously, in this example, passenger no. 1,000 plays the role of the identifiable person while the randomly picked persons play the roles of the statistical ones. So, the point is, for “no. 1,000, the difference between Option A and Option B is the difference between life and death” . Making the point even more precise, it is only for passenger no. 1,000 it is true that he will live or die for sure depending on what we choose—for the others, it is a matter of probabilities.
Now, James argues that “my reason to respond to determinate individuals is more stringent than my reason to respond to indeterminate individuals”  and that we, hence, should choose Option A rather than B. He claims that his argument for this does not hinge on our intuitive response to this case, since “most respondents (myself included) do not report any robust moral intuitions when confronted with the… case” .41 After all, “whether you choose Option A or Option B the outcome is identical: 501 lives saved and 499 lives lost” . If the argument does not hinge on our intuitions in cases where identifiability is the only difference, what is the argument?
Option B treats each passenger, including no. 1,000, as if her predicament was the same as that of any other, an approach that would be fair if the predicament of each was the same. But no. 1,000 predicament is not like that of any other… no. 1,000 can literally be saved at a negligible cost to others… 
That is, the “negligible cost” of having one’s probability of dying increased with 0.001, then.42
This argument is, I think, unconvincing. It rests on a mistake in moral mathematics or a failure to appreciate the severity of consequences for the additional statistical person dying if we choose option A. Regarding moral mathematics, it seems that the force of the argument rests on comparing (each) one of the other passengers with passenger no. 1,000. Of course, for each one of them, the increased probability to die if we choose Option A rather than B is 0.001. However, this increase accrues to every one of the passengers on sleeping deck, so that they add up to a probability of 100 % that one (more) of them will die. Naturally, ex hypotesi, we cannot know whom among the passengers on sleeping deck that would not otherwise have died if we choose Option A rather than Option B. But we do know someone did. And it seems to me that to say that we ought to choose Option A over Option B is to say that the lives of the passengers on sleeping deck matters less than the life of no. 1,000, since one more of them will then die for sure. I am not familiar with any account of fairness that would require us to make that choice.
Let me elaborate by using one of the analogies James describes in order to buttress his argument. He claims that our everyday behaviour reveals that we tacitly accept choosing Option A over Option B: “For example, I willingly—indeed cheerfully—increase my chances of dying prematurely by an even greater amount (i.e., greater than 0.001) every time I get into car or aeroplane, swim in the ocean, or eat ground beef” . True as this may be, this is not the proper analogy to the cruiser-line case. If there was an absolute probability 1 to a 1,000 that I would be killed every time I got into a car that would add up to a probability of 100 % were I to do it a thousand times, I would very much hesitate about getting into cars and would certainly keep a count how many times I did it. This is precisely how real the increased prospect of dying is for those on the sleeping deck, were we to choose Option A. Here, I suspect that James conflates probabilities understood as relative frequencies (as are the case with the everyday behaviour-examples) with objective probabilities, i.e., a limited number of equally likely cases where one will certainly occur (as is the case with the cruise-liner example or rolling a dice).
Moreover, I doubt very much whether identifiability is the only factor at work when judging this case—in fact, it seems less important than the number of individuals involved. If there was five persons on the cruiser-line and the terrorists were determined to kill two of them, on Option A the identifiable person on deck and one other randomly and on Option B two persons randomly, the randomly picked individuals would run a 25 % risk of being killed on the former Option and a 50 % risk on the latter. James concedes that the verdict most likely would, and indeed should, be different, in this kind of cases with drastically less people, and, hence, much higher probability for the statistical persons to die . However, ex hypotesi, we are still in a situation of metaphysical unidentifiability, i.e., we cannot say who would have died if we choose Option A that would have survived if we choose Option B.
The fact that we are much more reluctant to be anything but indifferent between Option A and Option B in the situation with fewer individuals goes to show that it is the “negligible cost”-factor rather than the metaphysical unidentifiability that does most (if not all) of the (alleged) moral difference. But to think that the size of the group to which the statistical persons belong matters from the point of view of fairness seems again to be a combination of a mistake in moral mathematics and a failure to appreciate the severity of consequences for the statistical persons. When we are dealing with small probabilities for a disastrous consequence distributed over a great number of individuals that add up to a great probability, we are tempted to think that the disastrous consequence is less likely to happen than if the same probability accrues to a single individual. But it is not—what is uncertain is to whom exactly it will accrue. It is also tempting to think that a person belonging to a large group is somehow less real than someone belonging to a small one. But they are not. And their lives are no less important.
There is also the question of the moral weight of identifiability. In order to recapitulate, James holds that identifiability sometimes matters, because it may be a relevant consideration of fairness. It is such a consideration, at least, when increased risk for the statistical individual is low enough (“negligible”). In such situations, arguably, the reason to benefit the identifiable rather than the statistical individual must be pro tanto, i.e., it has some genuine, although overridable, weight (the reason cannot just go in and out of existence) .
Furthermore, identifiability43 must be more than a tie-breaker, i.e., only making a difference when benefits are exactly equal for identifiable and statistical persons. Otherwise, the claim that identifiability matters is practically uninteresting. That is, in order for the idea of identifiability to have practical relevance, we must be prepared to pay some price in terms of overall benefits for the sake of identifiability and fairness.
But now, it seems, James’s idea faces a normative problem usually ascribed to his consequentialist adversaries, namely the problem of being too demanding.44 There may be a lot of individuals who should all be benefited before the statistical ones according to James. Since they all have a larger claim on benefits, the statistical individual should accept an increased risk, e.g. of dying, for the benefit of the identifiable person A, but also B, C, D, etc. This may add up to a very large probability, even though it is still true that in relation to each single identifiable individual foregoing the benefit among the statistical persons only means a “negligible” increase of risk. Statistical persons may end up with having to make huge sacrifices for the sake of the identifiable persons. I suppose that the potential huge sacrifice is exactly what James thought unfair with not opting for the identifiable person in the first place.
Identifiability and Rare Diseases
Now, we are equipped to return to the practical problem of prioritizing treatments for rare diseases. As explained in the outset, it is less and less true that treatments for rare diseases together constitute a negligible part of health expenditures. Due to the development of personalized medicine, more and more patients who previously had common diseases now belong to a subgroup differentiated as having a rare one. And there are more and more orphan drugs for these diseases. As mentioned in the outset, not least in oncology, the development has almost been revolutionary. For instance, in the US, there are approximately 50 approved personalized cancer treatments, which (almost all) cost $70,000 to $130,000 per treatment and (with some exceptions) result in gains in life expectancy measurable in weeks or months: “most of these drugs will have cost effectiveness numbers in the hundreds of thousands of dollars per… QALY… gained” . If society was to pay all dying cancer patients such treatments, it would increase $60 billion to health care expenditures. This is not to say that we should never pay for these treatments.45 But we have to carefully consider the opportunity cost and consider what cost in terms of many more statistical lives the treatment for these (often) identifiable individuals are worth.
This means that another of the analogies used by James to buttress his argument is not entirely apt for the context of rare diseases, namely the famous child in the pond-analogy (originally crafted by Peter Singer). James, like others, suggest that Singer is wrong to claim that the child drowning in the pond is analogous to the statistical persons in other parts of the world we could save, e.g. by spending more on famine relief: the identifiable child in the pond that we are passing do have a larger entitlement to being helped .
But the situation we are dealing with is a large and growing number of children and the rescuers are the ghost-like statistical persons, who in even greater numbers are dragged down by the desperate children. Consequently, we may, in the end, be facing something of a ‘bottomless pit’ objection regarding identifiability. This objection is usually directed towards the sickest first principle, i.e., the idea that we should give absolute priority to the worst off and not spend any health care resources on the second worse off until the needs of the worst off are met as far as possible. The objection says that the sickest first principle has the unacceptable implication that that we should spend all our resources on a few very needy who can only be helped to a very limited extent, instead of helping a great number of almost as badly off individuals very much . With an increasingly growing number of identifiable patients whom all can be helped to a very limited extent to very high costs, we may to an increasing extent face a bottomless pit of identifiable persons in health care. The expanding magnitude of the problem, I think, may make people less inclined to agree that the identifiable persons always should be prioritized at the expense of the statistical ones.
This article is about the justifiability of accepting worse cost effectiveness for orphan drugs in a publicly financed health care system. Recently, three arguments have been presented that may be used to answer this question positively. These three arguments share the common feature of all referring to considerations of justice or fairness: the argument of the irrelevance of group size, the argument from the principle of need, and the argument of identifiability. I have argued that all of these arguments fail to support the conclusion that orphan drugs should be subsidized to a larger extent than treatments for common diseases. The first and last arguments are invalid. The argument of the irrelevance fails to distinguish between directly and indirectly relevant considerations of fairness or justice. The recent attempt to revive the moral relevance of identifiability has provided no novel reasons to think that identifiability is morally relevant in itself or due to considerations of fairness and justice. The argument from the principle of need does not fail due to any inherent flaw in the principle as such. Rather, this principle can be interpreted in different ways, and none of these interpretations support economically advantageous conditions for treating rare diseases.
Once again, let us be clear about what I am not saying. I am not saying that orphan drugs should not be granted exceptionally advantageous terms economically than treatments for common diseases. To argue for such a drastic conclusion in such a difficult and controversial question is obviously beyond the scope of any single article. But I do claim that the best recent attempts to argue otherwise are unsuccessful. In the light of the general growing criticism against the preferential treatment of orphan drugs, it would be interesting to know exactly what the normative basis of this treatment is.
Hence, I will not consider the interesting and closely related question on whether or not to accept worse evidence for the effect of orphan drugs than treatments for common diseases .
In these systems, the practical question is rather: ‘what should private insurance cover?’ .
Although different areas of legislation have different definitions, this makes negligible difference for the discussions . For example, all the examples used would qualify as rare diseases according to legislations also in the US, Japan, or Australia, all of which have slightly different definitions.
I have suggested elsewhere that it should .
“[A]pproximately five new are described in the medical literature every week” .
The estimation of cost varies depending on what factors are included: it would be lower if one discounted, for instance, marketing costs. However, even if estimated differently, the expected cost for developing drugs for rare diseases need not be lower than for common diseases. However, bringing orphan drugs to the market is generally somewhat cheaper for two reasons: typically, regulation allows a lesser degree of evidence and therapies that have previously been in use off label for a particular rare condition can sometimes be relabelled as orphan treatments, and enjoy extended protection (I would like to thank an anonymous reviewer for bringing this to my attention).
Since Tony Hope’s seminal discussion of these topics .
Since 2007, the patent for statins has expired, likely making the cost effectiveness ratio even better. I would like to thank an anonymous reviewer for bringing this to my attention.
In this instance, non-smoking males 55 years or older with 170 mm Hg (SBU , p. 478). Although the level of evidence in both the statin-study and the report on hypertension must be deemed as high (SBU , p. 451), it must be remembered that the calculation of QALY rests on a number of assumptions, e.g. on what to measure and how to measure it, differences in which may affect results.
All translations in the text are mine.
It is unclear to what extent the authors themselves accept the validity of this argument , p. 854.
It is coherent to hold that cost has no relevance at all: it could even be held as a laudable expression of unwillingness to aggregate on the moral basis of separateness of persons (I would like to thank an anonymous reviewer for pointing this out to me). Although I find it implausible to hold that cost does not matter whatsoever, for instance since it would lead to an unacceptable degree of levelling down , pp. 900–902, there is no room for a comprehensive argument to this effect in this article. However, denying the relevance of cost does not imply prioritizing rare diseases.
The term prioritization is here used in a broad sense, not only referring to the ranking of individual patients, but also including the distribution of health care resources, e.g. by allowing a higher cost per QALY for rare diseases than non-rare diseases. However, as is the case with this sort of distribution, distribution has consequences for which patients that actually get treatments.
On a luck egalitarian conception of justice, we, perhaps, should consider all these indirect factors as relevant and, hence, try to even them out (I would like to thank an anonymous reviewer for pointing this out to me). However, luck egalitarianism would consider the difference between rare and common diseases as irrelevant in itself, since the rarity of disease does not determine whether it is the result of option luck or brute luck.
In order to avoid any misunderstandings, I am talking about principles of need in health care, i.e., substantial ideas on how health care ought to be prioritized and not the concept of health care need, i.e., what it means to say that someone(s) have a need for health care (although the former, of course, always presuppose or imply some view on the latter) [14, 25].
Moreover, this interpretation likely captures the core idea of principles of need most straightforwardly .
The second of the statements in this paragraph is sometimes called the positive thesis (of sufficientarianism) and the third one the negative thesis .
For instance, below that level we have special entitlements to publicly funded health care .
For instance, on how to set a non-arbitrary level of sufficiency .
What I have previously called the minimise insufficiency version of the principle of need .
At least if one excludes versions of prioritarianism that give absolute priority to the worst off. However, these versions are not very plausible .
This would be analogous to moving left on Temkin’s famous sequence .
For an elaborate formulation and defence of positive egalitarianism, see .
I will not enter the question about what could be meant by “good” health since it will not matter to the argument, but only presuppose that it means absence of disease.
As the state of medical art actually is, i.e., fancy philosophical thought experiments apart.
I presuppose that all these examples would constitute clear deviations from full health, since they are included in standard classifications of diseases like the International Classification of Diseases and Related Health Problems (World Health Organization .
Unlike outcome egalitarianism, where some levelling down would perhaps be acceptable to achieve more equal health, it is hard to see any upside with worsening someone’s probability of achieving health.
I suppose it could be claimed (although it never actually has) that those with rare diseases are less responsible than those with common ones to a degree that would make it justifiable to treat them differently in general, perhaps on the basis of the fact that a great many rare diseases are monogenetic and hardly any common diseases are. However, quite a few rare diseases, like certain kinds of melanoma, are not monogenetic and many common diseases, like diabetes, Alzheimer’s, and cardiovascular diseases, are at least partly due to genetic factors, so anyone trying to draw on such differences have a considerable amount of arguing to do.
The authors seem to suggest that this should be understood in accordance with the first interpretation above , p. 897, which is, again, a non-starter.
If patient groups should be individuated according to diseases. Another way to individuate them would be according to ‘rare diseases’ and ‘non-rare’, but then one would present the thesis rather than argue for it.
However, some patients with common diseases are identifiable in the sense defined below and, most probably, some patients with rare diseases are statistical. Nonetheless, there is sufficient overlapping between the groups, not least between patients with rare diseases and identifiable individuals, to make the argument interesting to scrutinize.
It has been argued that the rule of rescue is especially problematic within public policy areas where agent-neutral reasons presumably play a larger role , such as the one we are dealing with here. However, the rule has also been questioned as a sound principle in more concrete clinical situations .
Besides the already mentioned articles on the rule of rescue, there is a good overview of many of the contributions in .
The disease and its treatment have several similarities to the initial example of Hunter’s, but the treatment also has life prolonging effects .
In fact, we often have more reliable evidence for the beneficial effects of treatments for common diseases, like the ones mentioned, than for rare ones, simply since it is easier to get sufficient patient material regarding common disorders—lack of evidence is a greater problem for orphan drugs .
So James does not claim that identifiability matters in itself, but that it is sometimes wrong to benefit the statistical rather than the identifiable person because it would unfair (which matters morally in itself) and the only difference between these persons is the very fact that the latter, unlike the former, is identifiable .
I must say that my intuitive response to the case was that we should flip a coin, since the outcomes in terms of lost lives are identical on both options and flipping a coin would give passenger 1,000 a fair chance of being saved.
In the described (metaphysical) sense, for the described reasons (of fairness), in the described situations (of many statistical persons)—this will be presupposed in the following.
Actually, if we compare individuals pairwise, this problem arises also if identifiability is seen merely as a tie-breaker.
See  for an insightful discussion on what to consider more precisely when determining whether to pay for these kinds of treatments.
- 3.Brock, D. W. (2002). Priority to the worse off in health-care resource prioritization. In R. Rosamond, M. P. Battin, & M. Silvers (Eds.), Medicine and social justice. Essays on the distribution of health care (pp. 362–372). New York: Oxford University Press.Google Scholar
- 4.Carlsson, P., Hoffman, M., Levin, L.-Å., Sandman, L., & Wiss, J. (2012). Prioritization and financing of drugs for treatment of patients with rare diseases. (In Swedish: Prioritering och finansiering av läkemedel för behandling av patienter med sällsynta sjukdomar.) Appendix 4 in the Official Governmental Inquiry (SOU) 2012:75. Price, access, and service—continued developments in the drug and pharmaceutical markets. (In Swedish: Pris, tillgång och service—fortsatt utveckling av läkemedels- och apoteksmarknaden). http://www.regeringen.se/content/1/c6/20/25/65/36dbeb36.pdf. Accessed May 14, 2014.
- 6.Crisp, R. (2002). Treatment according to need: Justice and the British National Health Service. In R. Rosamond, M. P. Battin, & M. Silvers (Eds.), Medicine and social justice. Essays on the distribution of health care (pp. 134–143). New York: Oxford University Press.Google Scholar
- 7.Daniels, N. (2008). Just health: Meeting health needs fairly. New York: Cambridge University Press.Google Scholar
- 8.European Commission. Regulation No 141/2000 of the European Parliament and of the Council of 16 December 1999 on orphan medicinal products. http://ec.europa.eu/health/files/eudralex/vol-1/reg_2000_141/reg_2000_141_en.pdf. Accessed May 14, 2014.
- 15.Hayes, E. (2013) GSK prices Tafinlar at $7,600/month, undercutting Roche’s Zelboraf. Elsevier Business Intelligence. http://www.kantarhealth.com/docs/default-source/press-articles/gsk-prices-tafinlar-at-7600-month-undercutting-roches-zelboraf.pdf?sfvrsn=4. Accessed May 14, 2014.
- 20.Jarvis, L. M. (2013). Orphans find a home. Chemical & Engineer News, 91, 10–12.Google Scholar
- 21.Jarvis, L. M. (2013). Small audience, large payoff. Chemical & Engineer News, 91, 12–15.Google Scholar
- 24.Juth, N. (2012). Genetic information—Values and rights: The morality of presymptomatic genetic testing. Saarbrücken: Lambert Academic Publishing.Google Scholar
- 25.Juth, N. (2013). Challenges for principles of needs in health care. Health Care Analysis 2013 March 12. Published online ahead of print. doi: 10.1007/s10728-013-0242-7.
- 28.McCabe, C. (2010). Balancing economic, ethical and equity concerns in orphan drugs and rare diseases. European Journal of Hospital Pharmacy Practice, 16, 22–25.Google Scholar
- 32.SBU-rapport nr 170. (2004). Måttligt förhöjt blodtryck. (In Swedish: Moderately elevated blood pressure.) SBU: Stockholm.Google Scholar
- 33.Segall, S. (2010). Health, luck, and justice. Princeton: Princeton University Press.Google Scholar
- 35.Temkin, L. S. (1993). Inequality. New York: Oxford University Press.Google Scholar
- 37.Ward, S., Lloyd Jones, M., Pandor, A., Holmes, M., Ara, R., Ryan, A., et al. (2007). A systematic review and economic evaluation of statins for the prevention of coronary events. Health Technology Assessment, 11, 1–160., iii–iv.Google Scholar
- 38.World Health Organization (WHO). (2010). International Classification of Diseases and Related Health Problems (ICD 10). http://www.who.int/classifications/icd/en/. Accessed Aug 05, 2014.