In whose interest?

It would be best not to have to die at all, but failing that, many of us would like to have some control over the time and manner of our deaths, should we find ourselves in a condition so hopeless that there is no point in going on. At this date, in most of the world, including Britain and most of North America, the legally permissible forms of such control do not include voluntary euthanasia or assisted suicide. A physician may not, in most jurisdictions, administer or prescribe for self-administration a lethal drug for the purpose of ending life, even at the explicit request of a fully competent patient. Yet a physician is legally permitted (sometimes legally required) to take other steps that hasten death, if requested to do so by the patient or a proxy: he may end or withhold treatment (chemotherapy, antibiotics, surgery); withdraw or fail to initiate life support (respirators, feeding tubes); give high doses of opiates to relieve pain or induce coma, where these may also shorten life.

In his lucid and powerful book, the Canadian philosopher L.W. Sumner argues that this distinction is indefensible. His target is what he calls the Conventional View, namely that there is ‘an ethical “bright line” between assisted death, on the one hand, and all other end-of-life measures which may hasten death, on the other’. By assisted death Sumner means either providing (assisted suicide) or administering (euthanasia) a lethal medication in order to cause the patient’s death as a means of relieving his suffering. Sumner argues that the differences between assisted death and the legally permitted measures, such as turning off a respirator, are for the most part morally irrelevant. He favours a regime like the one that now exists in the Benelux countries, where both assisted suicide and euthanasia are allowed, under carefully specified conditions.*

Sumner holds that end-of-life treatment, like all medical treatment, should be governed by two values: respect for the patient’s autonomy and concern for the patient’s well-being. Usually they coincide, but where autonomy is not possible, as with an infant, well-being must govern. And if autonomy and well-being conflict, in the decision of a competent patient, Sumner believes autonomy should take priority.

With respect to the value of well-being, the most important and obvious point to be made is that one is justified in hastening death only in cases where death is better or at least no worse for the patient than continued life. Death may either be good, when it ends intolerable suffering, or neither good nor bad, when it ends a state of irreversible unconsciousness. Normally, death is a great evil for the person who dies, which explains the stringency of prohibitions against killing; but death is not always an evil, by comparison with the alternative. Sumner accepts the ‘deprivation account’, according to which death is bad because it deprives us of the goods that would be brought by continued life. If continued life would bring nothing but misery, in the form of intractable pain, nausea, delirium and helplessness, then a death that relieves us of that misery is not bad but good. That is what leads competent patients in sufficiently dire conditions to refuse further treatment, request removal of respirators and feeding tubes, or ask for terminal sedation. Evidently, the same end could often be achieved more quickly and effectively by a lethal injection, so the value of the patient’s well-being does not draw a ‘bright line’ between assisted death and other measures.

If we turn to the value of autonomy, we do find a difference. The currently recognised right of a patient to refuse treatment is very powerful, because it can be exercised for any reason consistent with the patient’s mental competence, and not just to end unbearable suffering. Adult Jehovah’s Witnesses are permitted to refuse blood transfusions on religious grounds, even if it will result in their deaths. This is the consequence of a general right to bodily integrity: no one, including a physician, may intrude on that integrity by surgery, injection of drugs, or other physical transgression contrary to the wishes of the patient. Without the patient’s explicit consent it would normally be a form of impermissible assault. (Exceptions for the treatment of an unconscious patient in an emergency are usually construed in terms of presumed consent: we believe the patient would agree if he were conscious and understood the situation.)

The right of bodily integrity introduces an asymmetry between withdrawing treatment and assisted death: a physician must end treatment if the patient demands it, even if the death that results is not in the patient’s interest; but no one, including Sumner, proposes that patients should have a right to request assisted death unless their medical condition makes life not worth continuing. The strong ‘negative’ obligations on others – obligations to refrain from interference – that are entailed by individual rights do not entail corresponding ‘positive’ obligations. There is a general negative obligation not to provide life-saving treatment against my wishes; but there is no general positive obligation to help me end my life just because I wish it. (Sumner gives the startling information that ‘in the Netherlands approximately two-thirds of requests for euthanasia or assisted suicide are refused by physicians,’ presumably because they are thought to be inadequately grounded.) Still, this asymmetry with respect to obligation doesn’t imply that there is a corresponding asymmetry between the permissibility of withholding treatment and the permissibility of assisted death, in the more usual case where both autonomy and well-being are served by either measure. Sumner’s position is that when the two values coincide, either method of hastening death is permissible.

He needs to defend this position, however, against the objection that there are two further differences between assisted death and the other methods that distinguish them morally. This is a crucial part of his argument, because it’s those differences that are at the heart of the Conventional View. Euthanasia and assisted suicide have two features, at least one of which is absent from each of the other measures. First, lethal medication causes death directly, rather than merely removing an obstacle to its occurrence from other causes. Second, the death of the patient is the intended aim of administering the medication, not merely an effect that is foreseen.

By contrast, when treatment is withheld or life support is terminated, death is caused by the disease or by one of its consequences, like the inability to breathe or eat. And when death is caused by extremely high doses of opiates, the intended aim is to relieve pain, and death is merely a foreseeable side effect. The claim that needs to be answered is that assisted death involves direct, intentional killing, and that this is wrong even if supported by values of autonomy and well-being that would justify allowing a terminal cancer patient to die of a treatable infection or shortening his life as a side effect of pain relief. This is clearly the assumption behind the Conventional View, the alleged moral ‘bright line’. The distinctions between doing and allowing, and between intending and foreseeing, are the subject of extensive and continuing discussion in moral philosophy, and there is disagreement about their significance. Sumner summarises the discussion, but his main point is that, whatever may be the moral importance of those distinctions in other contexts, they cannot mean the same in end-of-life choices, for a stunningly simple reason.

As applied to choices that lead to someone’s death, those distinctions have been developed to deal with cases that have the following three features: (1) death is a bad thing for the person who dies; (2) that person does not agree to it; (3) the choice is motivated by the aim of benefiting, or avoiding harm to, a different person. In those painful circumstances where someone or other will suffer harm no matter what we do, it is essential for a usable morality to determine which alternatives must be excluded, and which remain permissible. There is controversy over the correct standards to govern such cases, which include self-defence, collateral damage in warfare, and triage in medical care or rescue operations. But such standards have no bearing on end-of-life decisions, because all three of these crucial features are absent in typical cases of assisted death. When a competent patient requests euthanasia to end his suffering, (1) death is a good thing for him; (2) he agrees to it voluntarily; (3) its purpose is to benefit him and no one else. Even in the euthanasia of a noncompetent patient for whom there is no hope, such as an infant, the first and third conditions hold.

The whole point of identifying certain ‘indirect’ ways of causing death as morally permissible is to distinguish them from what is strictly forbidden, namely the intentional killing of an innocent person in order to benefit another. (Killing one person to harvest his organs and save five others is wrong; rescuing five persons and leaving a sixth to die, when one cannot rescue all six, is not wrong.) But euthanasia is not killing one person to benefit another; it is benefiting a single person by ending his life. The fact that the killing is intentional and causally direct is, to be sure, a feature that it shares with culpable murder, but it lacks the other features which make that morally significant. There is no moral reason, therefore, to contrast intentional euthanasia with the more indirect measures permitted by the Conventional View. The intentional/foreseen distinction has moral relevance only in cases of interpersonal trade-offs where death is an evil. Neither of these features is present in euthanasia and assisted suicide.

The same applies to the treatment of noncompetent patients who cannot exercise autonomy, such as infants born with a congenital defect that will end their lives after a year of suffering. The Conventional View permits hastening their death by not treating an infection, or not surgically correcting a bowel obstruction, for example. Sumner points out that these ‘failures to treat’ aim at the death of the infant just as much as active euthanasia; but the main point is that for both types of measure that fact is morally irrelevant, since the infant is not being harmed for the benefit of someone else: he is not being harmed at all, but spared a worse fate. (Sumner emphasises that all measures to hasten an infant’s death, ‘active’ or ‘passive’, are completely unjustified in cases of congenital disability that would allow the patient to lead a life worth living, such as Down’s syndrome.) The Conventional View depends on the confused transfer to this case of standards that apply to cases of a completely different kind.

The final major topic addressed by Sumner is that of respect for the autonomy of patients who are not now competent but who once were. Examples are cases of advanced dementia, and unconsciousness or semi-consciousness produced by stroke or brain damage. Advance directives in the form of a living will are designed for such cases, but they are not always available. Sometimes relatives offer evidence from things said in the past for what the patient ‘would have wanted’. Where an explicit or inferred past wish of the now incompetent patient clearly coincides with the patient’s present interest (relief of irreversible suffering), or does not conflict with it (terminating irreversible coma), there is no objection to any of the methods of hastening death. However, the priority Sumner gives to autonomy over well-being creates problems, which he recognises, for cases where an advance directive requests euthanasia or non-treatment under more ambiguous conditions.

Suppose, for example, that you sign a directive asking to be euthanised, or allowed to die of a treatable infection like pneumonia, if you should reach a condition of advanced dementia in which you don’t remember anything and can’t recognise anyone. It is possible to be in such a condition without intolerable suffering, and even to enjoy eating peanut butter and jelly sandwiches. Does autonomy really give your past self the authority to kill off this later self, even though its life retains the simple experiential value available to an infant? One response, proposed by Ronald Dworkin, is that we must distinguish between the experiential and the critical value of a life. Your directive expresses the view that in such circumstances, despite its primitive experiential value, your existence would not be worth continuing – it would be, in the critical sense, bad for you to go on living. So there is not really a conflict between autonomy and well-being in such a case.

Sumner is torn. He says it is hard to imagine administering a lethal injection to such a person. Yet he also says: ‘As a general thesis, it is hard to see how one might argue that current pleasures and enjoyments, under conditions of diminished or non-existent autonomy, always take precedence over previous, fully autonomous, expressions of one’s deepest and most enduring values.’ I’m doubtful of this grant of authority to the competent former self, whose values have not in fact endured, since they mean nothing to the demented present self. On balance there may be no right answer to the question whether it is good or bad for that person to go on living in that condition. But Sumner’s general point still applies: the same standard should be required to justify withholding life-sustaining treatment on the basis of such an advance directive as would be required to justify euthanasia: either both are permissible, or neither is.

Sumner’s detailed description of the legal landscape and its history is highly instructive. We are at the beginning of a great cultural shift in end-of-life treatment, made necessary by medical advances that can slow down the progress of fatal diseases while offering the patient continued life that is sometimes of doubtful value. Decisions to withhold further treatment and provide only palliative care are becoming common, and the provision of palliative care and adequate pain relief is now taken seriously by the medical profession. For most terminal patients those options are sufficient. Even where euthanasia and assisted suicide are available, in the Netherlands, they accounted for 1.7 per cent of all deaths in 2005, whereas withdrawing treatment or terminal sedation accounted for nearly one out of four deaths. Sumner argues that there should be much more regulation of the latter methods of shortening life, since they are so much more common, and equally in need of justification.

Sumner poses the question whether the wider availability of assisted death would reduce the incentive to provide adequate palliative care, since there would be a much cheaper and quicker alternative. He replies that this has not happened so far where assisted death has been legalised. Though the two types of measure have the same aim, to eliminate suffering, physicians apparently feel a strong motive to ensure that assisted death is chosen only as a last resort, when palliative measures fail. This optimistic view may hold for wealthy societies; I am not confident that it would hold up if assisted death became legal in societies with much more limited resources. Sumner’s book provides a superb example of the relevance of philosophy to public policy. The reason is that public policy governing treatment at the end of life is to a great extent shaped by philosophical confusions. It may not be too much to hope that a book such as this will help to rectify the situation.