Terminal sedation is a form of palliative treatment that has received much supportive and critical attention. The term first appeared in the literature in the 1990’s in Enck’s review of two articles concerning symptom control at the end of life.1 These papers suggested that up to 52% of patients near death had symptoms so resistant to typical palliative treatments that only sedation was sufficient to control them.2 3 By the end of the decade, debate was under way regarding whether terminal sedation was indeed ethical and whether it should be an endorsed practice. Billings and Block, for example, criticised what they saw as a practice that in some instances amounted to “slow euthanasia”.4

The literature that supports terminal sedation as ethical justifies it in two main ways. First, some authors hold that terminal sedation does not hasten death, or that it does not do so in a foreseeable way. Second, some authors hold that even if terminal sedation does rarely, foreseeably hasten death, notions of informed consent, proportionality and double effect justify its use. In addition to these justifications, there also runs through the literature an assumption that terminal sedation should be restricted to use in imminently dying patients. However, on review of the literature, no explicit justification is found for this assumption, and no justification follows from the principles justifying terminal sedation generally. Given that current palliative care includes the treatment of patients with chronic illness and patients early in the course of an incurable disease, the possibility of early, death-hastening terminal sedation has arisen, and therefore the deficit in the literature around such instances is worrying.

Regarding these considerations, this paper has two points:

• that terminal sedation may foreseeably hasten death in cases where fluid and nutritional support is withheld or withdrawn, and that therefore the first justification of terminal sedation is untenable;

• that the requirement that terminal sedation should be used only in imminently dying patients has been assumed rather than explicitly argued for, and no strong arguments for such a restriction are forthcoming from the typical justifications for terminal sedation.

DEFINITION

The definition of terminal sedation is uncertain and is contested in two areas. First, it is unclear about what type of sedation “terminal sedation” refers to.5^–14 However, the discussion has been clarified by the suggestion of Morita and colleagues that the purpose, depth and duration of the sedation used as a palliative treatment should be explicitly specified.14 15 In this way, sedation is distinguished as: (1) primary (where the intent is to sedate) or secondary (where the sedation arises as a side effect of other treatment); (2) mild to deep (based on objective criteria) and (3) intermittent or continuous. In this scheme, the phrase “palliative sedation” best refers to any use of sedation for palliation,5 7 15 while “terminal sedation” becomes the designate of a particular subset. Morita and colleagues suggest that in the literature, “terminal sedation” has been used to refer to instances of deep and continuous sedation (primary or secondary).15 The present paper also uses “terminal sedation” to refer only to deep and continuous sedation given as palliation.

Second, the circumstance in which it is ethical to use terminal sedation has been much debated. Many definitions have been put forward, but at the core they share the ideas that terminal sedation is (1) the use of a pharmacological agent or agents to induce unconsciousness (2) for treatment of truly distressing and refractory symptoms.10 15 Sedation thus described is justified by concepts of informed consent and by the principles of proportionality and double effect.16 In a review of the literature, Morita and colleagues also found that “terminal sedation” included the conditions (3) that the patient has been diagnosed as having a terminal illness and (4) that death is imminent.15 These conditions, however, are not argued for but are simply assumed in the literature. Furthermore, the concept of informed consent and the principles of proportionality and double effect cannot justify these conditions.

DOES TERMINAL SEDATION HASTEN DEATH?

Some authors suggest that terminal sedation simply does not hasten death. Sykes and Thorns reviewed four studies examining survival time after admission to a palliative care unit between those patients given sedation and those not given sedation. They found no evidence that sedation shortened life.17 Morita found that adding high dose opioid or sedative treatment during the last 48 hours of life had no effect on patient survival.18 In all these studies, however, the depth and intensity of sedation were not specified. Therefore, we can take them as evidence to show that palliative sedation on the whole does not hasten death, but no studies address clearly the specific concern with terminal sedation, taken as deep and continuous sedation.

In contrast, many authors presuppose that terminal sedation hastens death in some cases,4 6 7 13 19^–23 and, except for a few papers expressing concern over the reduction of consciousness itself,8 24 the hastening of death underlies the majority of criticisms of terminal sedation. Clinical beliefs seem to concur with this presumption. One survey of Dutch physicians found that over two-thirds of physicians who have used terminal sedation believed that, in their most recent case involving its use (explicitly distinguished from euthanasia), death was hastened by days to weeks.25 In a Japanese study, the number of physicians who felt that terminal sedation hastens death was just over one-third.26 But knowing that terminal sedation hastens death statistically is distinct from knowing that it will do so in the case at hand, and if the risk that terminal sedation will hasten death is only statistical, then the risks and benefits of it are best considered in the same way as are the side effects of other treatments—for in virtually all cases of administering medicine, there exists the possibility that the medicine may cause side effects of morbidity or mortality. What is held to be particularly problematic about terminal sedation is that it foreseeably hastens death in the case at hand.

The circumstance in which terminal sedation does foreseeably hasten death involves cases in which a patient has a prognosis of months and is still eating and drinking or receiving nutritional and fluid support. Providing deep and continuous sedation while withholding or withdrawing fluid and nutrition from such a patient will hasten the patient’s death. This paper refers to this practice as “early terminal sedation” (ETS). Many authors believe that in such cases of terminal sedation, death is foreseeably hastened by dehydration and starvation.4^–8 11 13 20 23 27^–33 Some may object that in many cases the underlying prognosis is so brief that a lack of fluids and nutrition will not shorten life, or suggest that most persons close to death no longer eat and drink significant amounts. The point fairly made in these arguments is that ETS is rare. For example, in a retrospective review, Sykes and Thorns found that at most 1% of cases of palliative sedation could have required justification for foreseeably causing death.34 But what matters is that such rare deaths due to ETS are legally and ethically sanctioned despite potentially shortening life by months. On legal grounds, the Supreme Court of the United States, the Supreme Court of Canada and decisions in British law all permit palliation even where it may foreseeably hasten death.32 33 35

THE IMMINENCE CONDITION

As experience in the Netherlands and in Oregon suggests, some people will prefer ETS to continued life. Yet the pervasive intuition against ETS, judging from the literature, is that terminal sedation should be used only in those patients who are imminently dying. What exactly constitutes “imminence” is somewhat vague in the literature but nonetheless roughly coherent. Where it is stipulated beyond such terms as “dying” or “last stages”, it is stipulated as hours or days or, at most, weeks.1 8 10

This restriction on who should be given terminal sedation is what this paper refers to as the “imminence condition”. A trenchant example of the importance of the imminence condition concerns the role of sedation in cases of extubation.36 Consider the patient receiving ventilator support with a prognosis of months or more, provided the patient remains ventilated. The debate concerns whether, if the patient decides that ventilator support should be withdrawn, the patient should be sedated before or after extubation. The imminence condition provides one way of framing this subtle point of debate. In one hypothetical case, the patient is sedated to unconsciousness and then extubated. In this case, by virtue of the longer prognosis at the time of sedation, the patient could be said to experience ETS. In a second case, the patient is first extubated, at which point death becomes imminent. At this point, sedating the patient to unconsciousness could be considered terminal sedation but could not be considered ETS because the patient’s death has already been made imminent by extubation. The arguments regarding this example are far more complex than is here indicated, but that they occur at all is in part due to the weight of the imminence condition. Given the point that this example illustrates, what is surprising is that the imminence condition has not received explicit justification in the literature on terminal sedation.

Early in the history of palliative care, to assume the imminence condition was an unproblematic move and went unnoticed, because palliative care dealt only with imminently dying patients. But over the past 20 years there has been a great shift in the patient populations treated by palliative care practitioners, to include treatment of patients with life-threatening or chronic or sometimes even potentially curable illness.37 The real possibility exists that terminal sedation could be used in these patients as well. Some have already suggested that deep, intermittent sedation, at least, is appropriate in these cases.28 38 39 Such issues are underscored by studies showing that unskilled or emotionally exhausted physicians may be more likely to consider using euthanasia for difficult symptoms in their patients.26 Furthermore, a growing literature has begun to ask whether terminal sedation should ethically extend to patients experiencing psychosocial–spiritual distress.27 28 Studies have shown that sedation, generally speaking, is already used in 0.4% to 5.3% of palliative patients for treatment of psychosocial–spiritual distress.40 Such distress can appear at many points during illness, and some have reasoned that this could lead certain patients to consider terminal sedation early in their illness.27 In all these cases, the potential appears that ETS could become a palliative option.

HAS EARLY TERMINAL SEDATION BEEN JUSTIFIED?

The justificatory reasons for using terminal sedation generally (as opposed to ETS) fall under clinical efficacy, notions of informed consent, the principle of proportionality and the principle of double effect.16 Terminal sedation will achieve the aim that the patient desires, namely cessation of suffering, and it therefore is clinically effective and cannot be restricted on the basis of futility. The principle of proportionality upholds consequentialist thinking in asserting that the right choice is the least harmful or most beneficial one. Patients may aim at death through choosing to have care withheld or withdrawn, and there is reasonable legal and ethical consensus that, provided the requirements of informed consent are met, the practitioner should comply with the patient’s choice regarding the omission of treatment. A distinction between causing, foreseeing and intending death and merely causing and foreseeing death informs the law in the United States, Canada and Great Britain, allowing physicians to foreseeably hasten death where the patient’s suffering warrants palliation with the consequence of death. The important distinction between a hastened and foreseen death and a hastened, foreseen and intended death rests on the well-known “principle of double effect”.41 In most modern versions, an act producing both a harm and a good is acceptable if (1) the good is intended while the harm is merely foreseen, (2) the act is not in itself prohibited, (3) the good does not require the harm as a means and is attainable in no other way and (4) the good is proportional to the harm. This paper accepts the principle of double effect without argument, although acknowledging that there has been much criticism of it, some of which has been visited upon its use to justify terminal sedation.42

The traditional justifications of terminal sedation are unhelpful to supporting the imminence condition. Morita and colleagues comment that in the case of primary, continuous, deep sedation alone, the principles of autonomy and of proportionality and of double effect have yet to be shown to be justificatory,16 and indeed it seems unlikely that they will prove justificatory in the case of ETS. The principle of autonomy does not support the imminence condition and may in some cases oppose it, for the imminence condition restricts the enactment of well-considered, free patient choice for a non-futile treatment. Even if the patient fulfils the requirements of informed consent, the imminence condition prohibits the enactment of ETS.

Similarly, serving the patient’s best interests may also oppose the imminence condition. It may seem that ETS permits a great harm in the premature death of the patient, and this is I think the strongest intuition in favour of the imminence condition. But the objection, found so often in the literature supporting physician-assisted suicide and euthanasia, is that the burden of suffering may outweigh the harm of an earlier death—that a longer prognosis only increases the burden.

The principle of double effect requires that the intent of the physician be only to palliate. But as Williams and others have indicated, no further palliation is added by withdrawing or withholding fluids and nutrition in patients who are already palliated through sedation.32 33 At best it may be said that the physician intends only to comply with the patient’s autonomous request or intends only to prevent further suffering rather than intending the patient’s death. Such arguments, if made robust, could indeed remove the physician from culpability, but they in no way support or lead to the imminence condition. Instead, the principle of double effect, along with the principles of autonomy and proportionality, may in some cases oppose the imminence condition.

In the realm of individual ethics, it may be that the imminence condition is acting as a surrogate estimate for intractability. As we have seen, the intractability of a symptom increases the expected harm that the symptom will entail. When a patient has only days or weeks to live, there is less chance that suffering judged to be profound and intractable will have been mistakenly judged to be so. This is especially relevant in cases where the suffering is more psychosocial–spiritual than physical. Several authors have expressed trepidation about the use of terminal sedation for psychosocial–spiritual suffering because of the worry that such suffering may be even more difficult to judge correctly as intractable than is so-called physical suffering.5 8 12 43 44 Similarly, the Dutch courts have excluded “existential distress” as a valid reason for performing euthanasia, because the condition falls beyond the scope of reliable medical diagnosis.45 These concerns have led some to suggest that psychosocial–spiritual suffering may be in principle always treatable by means other than terminal sedation.8 44 Others suggest that the intractability of such suffering must be thoroughly “tested” by investigation and by treatments other than terminal sedation12 43 before considering terminal sedation. And some agree that, provided that the psychosocial–spiritual suffering has been accurately assessed as intractable, it could ethically be treated with terminal sedation.12 36 38 43 Such a move would see psychosocial–spiritual suffering and physical suffering to be equally deserving of ETS. But if it is the intractability of suffering that matters, and if the imminence condition thus acts as a surrogate for intractability, then this surrogacy also reduces the imminence condition to intractability. If this is correct, the imminence condition should be rejected as an independent factor defining terminal sedation. Instead of maintaining that deep and continuous sedation without fluid and nutritional support should be used only when death is imminent, we should maintain that it should be used only in cases where the suffering is clearly intractable, whenever that suffering occurs.

There may be reasons to accept the imminence condition, however, and, as evidenced by the fact that it is repeatedly assumed in discussions of terminal sedation, it does have intuitive support. Perhaps the most intense positions supporting the imminence condition and rejecting ETS derive from concerns with euthanasia. These positions start with the premise that euthanasia is ethically wrong, add the premise that ETS is ethically tantamount to euthanasia and conclude that therefore ETS is ethically wrong. The difficulty is, however, that the priority of premises and progression toward conclusion can be reversed, as this paper intimates. That is, one may hold that, on legal and moral grounds, ETS is acceptable, that euthanasia is tantamount to ETS, and that therefore euthanasia should be permitted. This reveals the “work” that the imminence condition does. By restricting terminal sedation to cases of imminent death, the intention of the physician and patient becomes causally unimportant. That is, such sedation cannot be euthanasia, for it does not hasten death. This restriction by the imminence condition obscures the identical justifications for terminal sedation and ETS. It is beyond this paper to enter the debate regarding the equivalence of ETS and euthanasia, but two points are worth emphasising. First, arguments against ETS based on its supposed similarity to euthanasia can be turned on their heads, and, second, the imminence condition has played a strong but unjustified and relatively unchallenged role in these arguments against ETS.

Justifications to support the imminence condition may come from areas beyond individual ethics, such as law and public policy. Such arguments could conceivably support the imminence condition, but they must first be clearly made and must stand scrutiny. The point of this paper is to draw attention to the fact that no justifications for the imminence condition have been explicitly made in the literature. This should compel us to find and examine the basis of this support for the imminence condition far more than has been done thus far.

CONCLUSIONS

Terminal sedation is best considered as the subset of palliative sedation consisting of deep and continuous sedation. If terminal sedation is given along with the withdrawal or withholding of the nutrition and fluids—constituting “early terminal sedation” (ETS)— this will in some cases result in the foreseen hastened death of the patient. Accepted legal and ethical justifications exist independently for terminal sedation and for the withholding and withdrawing of any treatments, including nutrition and hydration. But combined, the two practices allow for intentional hastened death. The literature on terminal sedation has suggested prohibiting ETS by insisting that terminal sedation should be used only in patients facing imminent death. This restriction has here been designated the “imminence condition”. There is, however, a deficit in the literature on terminal sedation insofar as the imminence condition has been assumed rather than explicitly argued for. Further, no support for the imminence condition exists within the traditional justifications of terminal sedation, namely the principles of autonomy, proportionality and double effect. If no robust justification for the imminence condition is forthcoming, we should reject the imminence condition and accept that some cases of ETS will occur and will foreseeably and considerably hasten death. However, the continued assumption in favour of the imminence condition throughout the literature on terminal sedation suggests that medicine and law see it as intuitively important. This paper has attempted to draw attention to the fact that if we are to accept the imminence condition and reject ETS, we will need justification to do so, and that presently we lack such justification in the literature.