Terminal Sedation, Euthanasia, and Causal Roles

Introduction

When terminal sedation is made the object of bioethical discussion, it is mostly in a way that a literary critic might call synecdochal: A part is taken for the whole. What is made the object of discussion is a whole of which terminal sedation is only a part. Not unlike discussions of prenatal diagnosis or preimplantation diagnosis, it is only the combination with other medical behaviors that creates ethical problems. In the case of terminal sedation, it is the combination with the termination of some nonpalliative medical treatment. Only in this combination does terminal sedation become a form of euthanasia.

If one takes for granted the standard definition of terminal sedation, this medical procedure cannot by itself be considered to constitute an act of euthanasia. By this definition, terminal sedation is a treatment administered when other palliative treatments are not sufficiently effective, and which aims at keeping a severely suffering patient unconscious in the proximity of death. Relevant indications are agitation, fear of suffocating, vomiting, sleeplessness, and states of severe pain. According to this definition, the adjective “terminal” refers to nothing but the temporal incidence of sedation and leaves open the causal relation between sedation and subsequent death. It does not prejudge the issue of whether sedation stands in any causal relation to the patient's death. As a rule, it is assumed to have no causal role in this death; it makes it easier for the patient to live his last period of life, but does not causally contribute to death. It constitutes help in dying and not help to die.

What is more controversial is the classification and evaluation of combinations of terminal sedation with the termination of other medical treatments, either preceding or following sedation.[1] This combination has been increasingly applied in medical practice, especially in The Netherlands, the country about whose practice we are best informed. In The Netherlands, terminal sedation has been increasingly used as an alternative to active euthanasia. The most recent Dutch survey, which for the first time includes information about terminal sedation, shows that in 2001 terminal sedation was given in 5.6% of all deaths. It also shows that in almost two thirds of these cases artificial nutrition and hydration were simultaneously withdrawn.[2] This combination of terminal sedation and termination of treatment, understood as the termination of some or all life-sustaining nonpalliative measures, will be the focus of discussion.

Three Scenarios

There are 2 ways of combining terminal sedation with the termination of life-sustaining treatment:

1. Combination of terminal sedation with withdrawing or withholding life-sustaining medical treatment; or

2. Combination of terminal sedation with the termination of life-sustaining nutrition and hydration in addition to withdrawing or withholding life-sustaining medical treatment.

Both variants pose a number of questions, primarily conceptual ones. If these combinations constitute acts of euthanasia, how should they be classified – as active, indirect, or passive euthanasia? Obviously, the answer depends on how the causal role between sedation and death is interpreted. The possibility that sedation is causally contributory to death by itself is only of secondary practical relevance. Nevertheless, it cannot be left aside, because it may at least partly explain why some authors tend to classify the combination of terminal sedation and the termination of life-sustaining treatment as an intermediate form not adequately covered by the traditional classification.

Let's consider the different scenarios more systematically:

1. The patient's death is caused (besides the patient's disease) exclusively by terminal sedation.

2. Aside from the patient's disease, the patient's death is caused exclusively by termination of treatment.

3. Aside from the patient's disease, the patient' s death is caused by both terminal sedation and termination of treatment.

The first case is a clear case of euthanasia because both standard conditions of euthanasia are satisfied, namely (1) that it aims at relieving suffering, and (2) that it hastens the patient's death. This fact, however, is completely independent of its being combined with terminating treatment. If the risk of hastening death is accepted from the start, it is a case of “indirect euthanasia.” If hastening death is intended, it is a case of “active euthanasia,” though if death is only probable, in a conditional form.

The second case, in which only termination of treatment is causally contributory, is practically more relevant. In this case, terminal sedation has only the function to save the patient the suffering resulting from termination of treatment. How should this case be described? Is it an intermediate form not covered by the traditional classification, eg, a form of euthanasia between active and passive euthanasia?

I do not think that the traditional classification must be extended to make room for this kind of case. Both variants of terminating treatment, if they occur in the situation of terminal illness, are forms of passive euthanasia, independently of whether they are accompanied by terminal sedation. Whether termination of treatment means withdrawing or withholding medical life-sustaining treatment or whether it means withdrawing or withholding hydration and nutrition, in both cases death occurs as a consequence of termination of treatment, no matter whether terminal sedation is applied.

There are 2 widely shared intuitions that militate against this classification. The first intuition suggests that both variants are cases of active euthanasia; the second intuition offers that at least the second variant should be classified as a case of active euthanasia.

The first intuition draws on an analogy between the given kind of case and another, closely connected kind of case: A actively puts B in a position in which B is dependent for his survival on an act of A, which A then refuses to carry out. Think, for example, of A as an army starving a town B with the aim of making it surrender. If the population of B is not prepared to surrender and prefers to starve, the description that A “lets” B “die” does not only seem insufficient but is also misleading. This description suggests that B comes to death from causes independent of A's behavior. The description would certainly be adequate to a situation in which an epidemic breaks out in B whose consequences for B might have been prevented by A's intervention. In the case of a siege, it is grossly inadequate. It is surely more adequate to describe this case as a case of “killing,” or perhaps, as a case of “killing by letting die.”[3] Obviously, here there is an analogy with the case of termination of treatment under terminal sedation. In this case, too, the sedated patient has no chance to ingest food, water, or medicine, or to ask for any of these. Would it not be more adequate, therefore, to describe this situation as “killing” or “killing by letting die” instead of “letting die” – in other words, as a case of active rather than passive euthanasia?

Again, this description would be misleading and inadequate. It would not sufficiently account for important disanalogies. A first disanalogy is the temporal sequence of events. In the case of the siege, A's depriving B of vital resources precedes A's act of refusing help. The situation of needing help temporally follows the situation in which the population is cut off from resources. In contrast to this, in the combination of terminal sedation and termination of treatment the temporal sequence does not make a difference. It does not make a difference whether sedation is given before or after termination of treatment. A second disanalogy is that the terminal patient stands in need of help independently of sedation. Differently from the besieged town, the patient is in a dependent situation from the start. The exact analogy to the case of the siege would be a case in which C puts a healthy person D in a position in which D becomes dependent for its survival on C, and in which C (with, without, or against D's will) deprives D of vital resources. In this case, it would be as misleading as in the case of the siege to talk of C “letting D die.” This description, again, would suggest that D's death is caused by factors independent of D's behavior and whose consequences might only have been prevented by D's intervention. That would give a wholly inadequate picture. Since, however, this possibility is not relevant in the case of terminal sedation, there is no need for a corresponding linguistic differentiation. There is no need to introduce an intermediate category between active and passive, killing and letting die.

The second intuition takes a different direction. It suggests that the second variant, the termination of nutrition and hydration, has a special quality that is not adequately accounted for by describing the combination of terminal sedation and termination of treatment as “passive.” This intuition often is accompanied by 2 other intuitions, which individually support the “active” character of the termination of life-sustaining hydration and nutrition.

The first supporting intuition is that only withholding or withdrawing nutrition and hydration, but not withholding or withdrawing medical treatment, is a factor which acts on the patient “from outside.” If nutrition and hydration are withdrawn, the patient dies from an external factor. If the patient dies from the withdrawal of medical treatment, he dies from his illness. On this view, the ascription of causality essentially depends on a certain standard of normality. Because the withdrawal of life-sustaining nutrition and hydration is something “nonnormal,” it is this factor to which the causality for the patient's death is ascribed, whereas termination of a life-sustaining medical treatment is not seen as a causal factor in the same way. This view is reflected in the common differentiation between “life-sustaining” and “life-prolonging” treatment. Withdrawing “life-sustaining” treatments is more likely to be thought of as causally contributory to the patient's death than withdrawing “life-prolonging” treatments. In the first case, the external factor is identified as the crucial causal factor; in the second case, it is the internal one.

I do not believe that the “normality standard” implicit in this differentiation can be a legitimate basis of assigning causal roles. This standard lacks an objective basis. There are no objective differences corresponding to the differences in description. Even in cases in which the patient's death occurs only as a consequence of withdrawing medical treatment, withdrawing treatment is a (negative) causal factor of the patient's death in the same way that withdrawing hydration and nutrition is a (negative) causal factor in the other case. The fact that medical treatment is seen as something “artificial” (and, correspondingly, “life-prolonging”), whereas hydration and nutrition are seen as something “natural” (and, correspondingly, “life-sustaining”) does not correspond to any difference in the objective causal structure. The only decisive consideration for assigning causal roles on an objective basis is the counterfactual consideration that death would not have occurred if treatment, whether “artificial” or “natural,” had not been withdrawn.

The second supporting intuition is the quasi-instinctive rejection of this kind of euthanasia by a large number of caregivers. Intuitively, the description “passive euthanasia” has been called too weak to adequately reflect this deeply felt rejection.

This consideration again is problematic, for the reason that it is generally problematic to base descriptive classifications on normative evaluations: It is one thing to describe a certain state of affairs and another to evaluate it. Inevitably it leads to confusion to keep these insufficiently apart. The problem is that if you base a certain descriptive classification on normative consideration (as is often done with the distinction between “active” and other kinds of euthanasia), you deprive yourself of the chance to base the normative distinction on descriptive characteristics. In order to justify an evaluation by reference to descriptive features, you need a descriptive classification that is independent of evaluation. From a purely descriptive perspective, however, the withdrawal of nutrition and hydration under terminal sedation can hardly be described as a case of active euthanasia. One of the necessary conditions of active euthanasia is that it is a positive act contributing to the patient's death. This condition is not fulfilled in the case of withdrawing or withholding nutrition and hydration, which do not generally involve positive action.

If the conditions of the second scenario are fulfilled and sedation is not by itself causally contributory to the patient's death, it follows that a combination of terminal sedation and termination of treatment cannot be classified as “active euthanasia.” What about the alternative of describing the combination as “indirect euthanasia”? This description seems no less inadequate. As an act of “indirect euthanasia” the patient's death would have to occur as a side effect of terminating treatment in cases in which this is aimed exclusively at reducing suffering. This is the case only if the patient's suffering is caused primarily by the treatment and not by his illness. In this case, the aim of relieving suffering might be attained primarily by termination of treatment. This, however, will be a rare case. As a rule, the patient's suffering is primarily caused by his illness, and termination of treatment aims at the relief of suffering by letting the patient die from it. The patient's death is the means leading to the ultimate aim of relieving suffering. As far as the patient's death is expected as a consequence of termination of treatment, the patient's death cannot be conceived as an unintended side effect. Even though the intention of the physician responsible for termination of treatment may be primarily directed to relieving suffering and not to hastening death, hastening death is, in this case, the means for realizing that aim. Necessarily, however, an intermediate end held to be necessary to attain a further end is intended. This is the rational core of the doctrine of double effect: Whoever uses a means M to attain an end E does not necessarily intend the side effect S of which he knows that he realizes it by realizing E. But whoever uses a means M to attain an end E and who knows that he realizes an intermediate end I by realizing M and, furthermore, thinks that it is only by realizing I that E can be attained, cannot plead that he only intends E and not I. If I is the only way to attain E by means of M, he necessarily intends I together with M and E.

In scenario 3, in which both factors are necessary and not individually sufficient to cause the patient's death, the conceptual situation is again different. This case, again, may give the impression of not being adequately covered by the traditional classification. This impression is deceptive, however. The case can be adequately accounted for by looking at it as a combination of indirect and passive euthanasia, at least as far as the intention with which sedation is administered is not the intention to hasten death. As far as terminal sedation is causally contributory to the patient's death as a side effect, the case is one of indirect euthanasia. As far as the termination of treatment is causally contributory to the patient's death intended as a means, the case is one of passive euthanasia. The scenario simply combines both variants.

An Ethical Assessment

In the ethical assessment of the combination of terminal sedation and termination of treatment, the latter is the crucial element both in scenario 2 and in scenario 3. In both cases terminal sedation only serves to mitigate the consequences of termination of treatment for the patient. As a consequence, the same ethical criteria must apply to these cases as to other cases of termination of treatment in situations of terminal suffering. That means that a combination of terminal sedation and termination of treatment must be judged to be morally acceptable or morally unacceptable under the same conditions as termination of treatment without accompanying sedation. What are these conditions?

It is widely agreed that termination of treatment is morally acceptable whenever further treatment would be futile. Futility means that further treatment would be ineffective in the sense of neither improving the well-being of the patient nor allowing him a further period of life with a quality of life acceptable to him.[4]

Termination of treatment under conditions of futility is generally held to be permissible but not obligatory. In individual cases, there can be good reasons to continue treatment (eg, the interest of relatives to see the patient alive), though these reasons are rightly controversial. The question is, however, whether in situations in which there is an indication for terminal sedation do stricter norms have to be observed, and whether there is not also an obligation to terminate treatment. However this question is answered, the physician is certainly obligated to relieve the symptoms following from termination of treatment.

It is agreed, furthermore, that termination of treatment is obligatory in 2 kinds of situations: (1) if further treatment is clearly detrimental to the patient, or the risks for his life and well-being strongly outbalance the chances to prolong life and to improve well-being; and (2) if a competent patient refuses to be treated, or has a made an advance directive to the same purpose. Both conditions need to be spelled out in more detail, which is beyond the scope of this article.[5] On the other hand, there are conditions under which termination of treatment under terminal sedation is clearly unjustified. This is the case if the decision to terminate treatment is not primarily oriented at the well-being or the will of the patient but at the well-being or the will of relatives or caregivers. The same holds for cases in which there are better alternatives available about which the patient has not been informed or which are withheld. Finally, termination of treatment under terminal sedation is at least problematic if the patient objects to being “put to sleep” with the prospect of going on living for a certain stretch of time without regaining consciousness. Possibly he prefers a more active and less protracted death, eg, a death by suicide.

The most important moral imperative for termination of treatment under terminal sedation is, however, that if the patient is competent and not sedated without his explicit consent. It is difficult to tell how far this ethically crucial imperative is satisfied in concrete practice. Even the most recent Dutch survey does not give any information on this point.

An Ethical Comparison With the Alternative of Assisted Suicide

What are the ethical pros and cons of terminal sedation in combination with termination of treatment in comparison with alternatives such as assisted suicide? Are there any ethical grounds for preferring the first procedure (which is widely accepted within the medical community) to the other (which, though legal in a number of countries, is widely rejected within the medical community)?

To be adequate, a comparison would have to take into account the totality of consequences of these alternative procedures – for the dying, for physicians, caregivers, relatives, and society at large. I propose to restrict the discussion to 2 key dimensions and perspectives: on the one hand the perspective of the patient who finds himself in a situation in which he feels the need to ask for euthanasia, and on the other the perspective of society. This is not to say that the perspectives of physicians, caregivers, and relatives are negligible, but it seems that both of the perspectives chosen are crucial.

From the perspective of a patient who finds himself in a state of severe suffering which proves resistant to palliative measures, it is difficult to formulate a clear ethical preference for one of these procedures. Which of the procedures is preferable depends to a high degree on individual attitudes. A clear advantage of the option of terminal sedation is that it does not in all cases mean a point of no return, but that it can be administered in a way that makes it reversible so that the patient has a chance to decide anew about termination of treatment after having regained consciousness. The fact that this possibility exists in terminal sedation is important, especially if sedation is given in a situation in which the patient is incompetent and in which sedation makes him regain competence so that he is only then in a position to make use of his right to self-determination and to decide about the option of a final termination of treatment. If, as Richard Momeyer[6] has argued, respecting the self-determination of the patient does not only mean to respect his will, but also to give him an opportunity to express and to carry out his will, the chance of regaining competence is a clear ethical advantage.

This advantage is, however, absent if the patient has decided that he wants his life ended. In this case he may have 2 reasons to prefer assisted suicide. First, with an act of suicide he openly expresses for others and for himself the autonomy of his decision. Second, he saves his relatives the potential burdens associated with a prolonged process of dying. The first aspect will be important for patients who want to determine themselves the exact time of their deaths. Though terminal sedation will, in general, be administered on the patient's explicit consent, it lacks the symbolic confirmation of the patient's will to take the end of his life in his own hands, characteristic of suicide. For many patients it will also be important not to confront their relatives with a prolonged dying process. Even if treatment is terminated immediately after sedation, the process can take a number of days.

What are the interests of society in this matter? Because society coincides with the class of potential patients, society has 2 chief interests: first, an interest in having euthanasia provided as an ultima ratio in hypothetical situations of terminal suffering; second, an interest in preventing abuse and misuse. From this perspective, something can be said for preferring terminal sedation with termination of treatment to assisted suicide, at least on the background of the attitudes dominating the majority of “bioethical cultures” at present. First, the prospect of being able to resort to termination of treatment accompanied by terminal sedation in a possible state of severe terminal suffering is much better than the prospect of being able to resort to suicide. Terminal sedation, different from assisted suicide, is not dependent on the capacity of the patient to move his limbs or to swallow, and it is not dependent on the availability of appropriate substances or the assistance of others in providing them. The fact that it is not easy to find a physician who is prepared to assist a patient in committing suicide seems a consequence not so much of loyalties to professional norms than of perceived causality.[7] In assisted suicide, the causality of the patient's death is perceived to lie in the person of the physician to a much higher degree than in termination of treatment. Even if the doctor is only the penultimate and indirect cause of the patient's death and not the ultimate and direct one, the role of the physician is perceived to be more “active” in this case than in the case of termination of treatment. Even in cases in which termination of treatment is carried out by a positive act (such as turning a switch) the causality leading to the patient's death is more readily attributed to the patient's illness than to the physician's way of acting. This difference in perceived causality is further supported by differences in timing. The more immediately that death follows upon the physician's intervention, be it by action or by nonaction, the more probable it is that ascriptions of causality focus on intervention. The fact that termination of treatment precedes the patient's death by a certain interval of time makes it easier for the physician to contribute to the patient's death than in assisted suicide. Though it is questionable as to whether these perceptions correspond to any objective fact (the causal structure is identical in both cases), these differences in causal attribution may explain why there are markedly different psychological thresholds in relation to relatively “active” and relatively “passive” forms of euthanasia.

The second reason to give priority to terminal sedation from a societal perspective is based on the assessment that the risks of abuse and misuse are easier to avoid with terminal sedation than with assisted suicide. Though at first glance it might seem that the value of patient self-determination is much better safeguarded by a practice of assisted suicide than by a practice of terminal sedation, there is much to be said against this picture. Critics of terminal sedation, like David Orentlicher,[8] may well be right in pointing to the fact that the risks of terminal sedation in combination with termination of treatment are more to be feared than the risks of assisted suicide, since in terminal sedation the crucial causal factors do not lie with the patient but with the physician. But this is true only in theory. In practice, it is hardly imaginable that an incompetent patient, or a patient who does not want to die, might be forced to commit suicide, whereas it is imaginable that a patient in an advanced state of dementia is sedated and let die against his will. It seems quite improbable that a physician sedates a completely healthy person in order to give him an opportunity to die by lack of hydration. It is much less improbable that a physician provides a healthy person with a substance to allow him to die by suicide.

Final Remarks

The result of the comparison of the arguments pro and con regarding terminal sedation in contexts of euthanasia was that this option is clearly preferable to assisted suicide. This result does not, however, amount to saying that this option is the only morally acceptable one. It is compatible with saying that both procedures are morally acceptable, though the one is ethically advantageous. Short of a more thorough investigation of both alternatives, there is, however, one thing that can be said even at this stage of the argument: While there are good reasons not to make it obligatory for a physician to assist a severely suffering patient in committing suicide, there are good reasons to make it obligatory for a physician to give a patient terminal sedation under the same circumstances, provided that this is the only means to save him the symptoms consequent on termination of treatment. I would even go further and say that there are reasons to give the patient the right to demand terminal sedation under these conditions. One reason for this is the strong interest of the patient in having his suffering relieved in the last moments of his life. Another reason is the assurance, already in one's healthy days, resulting from the expectation of a peaceful death.