Opioid and sedative use are common ‘active’ practices in the provision of mainstream palliative care services, and are typically distinguished from euthanasia on the basis that they do not shorten survival time. Even supposing that they did, it is often argued that they are justified and distinguished from euthanasia via appeal to Aquinas’ Doctrine of Double Effect. In this essay, I will appraise the empirical evidence regarding opioid/sedative use and survival time, and argue for a position of agnosticism. I will then argue that the Doctrine of Double Effect is a useful ethical tool but is ultimately not a sound ethical principle, and even if it were, it is unclear whether palliative opioid/sedative use satisfy its four criteria. Although this essay does not establish any definitive proofs, it aims to provide reasons to doubt—and therefore weaken—the often-claimed ethical distinction between euthanasia and palliative opioid/sedative use.
- palliative care
- elderly and terminally ill
Statistics from Altmetric.com
Suppose that a fully conscious patient with a terminal illness, such as end-stage metastatic lung cancer, voluntarily requests to be killed by his/her doctor as a means to relieve his/her suffering. After other options are explored and discussed, and after it is ensured that the patient has provided informed consent, his/her doctor obliges and injects him/her with a lethal combination of drugs, with the full intention of those drugs killing him/her, as a means of respecting his/her autonomy and relieving his/her suffering. This constitutes the paradigm case of ‘euthanasia’ as I will use the term in this essay.i
One fruitful way to explore the ethics of euthanasia is via comparison with already widely accepted practices in the provision of mainstream palliative care, in order to draw similarities and differences between them. Palliative care comprises a broad range of pharmacological and non-pharmacological interventions, which can also be divided into ‘active’ practices (eg, the administration of drugs) and ‘passive’ practices (eg, withholding drugs, instituting ‘do not resuscitate’ orders, turning off mechanical ventilators, removing nasogastric feeding tubes). In this essay, I will focus on two specific active practices: the administration of opioids and sedatives.
Opioids are often administered to dying patients for pain relief. In large enough doses, strong opioids such as morphine cause respiratory depression and may directly result in death. Deliberate large boluses of opioids can be used for euthanasia. Alternatively, opioids can be used with the sole intention of relieving pain by beginning with a low dose, and then incrementally and cautiously increasing that dose to achieve satisfactory pain relief without overshooting the required dose. If the pain increases or decreases, or if evidence of opioid toxicity develops (eg, increased drowsiness, muscle twitching/jerking), then the dose can be titrated accordingly. Let us call this practice the ‘appropriately titrated administration of opioids’ (ATAO).
Furthermore, sometimes pain or other physical symptoms such as terminal agitation are intractable, that is, they cannot be satisfactorily alleviated by opioids or other drugs targeting those symptoms directly. In this case, the patient’s level of consciousness may be deliberately reduced to reduce his/her experience of that suffering. This ‘palliative sedation’ is usually achieved with a benzodiazepine or barbiturate drug. It may be intermittent or continuous, and may be mild, moderate or deep. ‘Continuous deep palliative sedation’ (CDPS), also known as ‘terminal sedation’ or ‘definitive sedation’, is the absolute last resort—used only where all other more conservative attempts have failed—and involves permanently sedating the patient to unconsciousness. The intention of CDPS, and all other levels of palliative sedation, is to circumvent the patient’s awareness of his/her symptoms, but not to hasten or bring about death. The patient is then allowed to die owing to their underlying disease process, such as advanced cancer. Like with opioids, large boluses of benzodiazepines and barbiturates can be used for euthanasia, but in palliative sedation this is not the case; the dose is carefully titrated and not overshot.
Given the nature of the drugs involved, it is at least biologically plausible that—despite their careful titration—the ATAOs and palliative sedation (particularly CDPS) do in fact hasten or bring about death, or in other words, shorten survival time. If this is true then they bare a similarity with euthanasia, which also involves an active intervention (usually a drug) which hastens or brings about death. Opponents of euthanasia may respond in two ways, either (1) rejecting the empirical claim that the ATAOs and palliative sedation shorten survival time (for which there is an extensive literature), or (2) accepting that they do shorten survival time, but arguing that they are ethical practices (unlike euthanasia) on the basis of some other differentiating feature(s). This is often argued through appeal to Aquinas’ Doctrine of Double Effect (DDE), and therefore depends on two further claims that the DDE is a sound ethical principle, and the ATAOs and palliative sedation meet its four criteria.
In this essay, in the first section, I will analyse the existing empirical literature and argue against the widespread conclusion that the ATAOs and palliative sedation do not shorten survival time. I will instead advocate for a position of agnosticism which I believe should be more widely adopted. Given this uncertainty and the resulting possibility that the ATAOs and palliative sedation do in fact shorten survival time, an evaluation of the DDE is essential. In the first part of the second section (the claim that DDE is a sound ethical principle), I will deconstruct the ‘mid-level theory’ of the DDE to its component ‘high theory’ deontological and utilitarian roots, exposing some of its hidden shortcomings and circularities. I will then argue through thought experimentation that the DDE is specific but not sensitive in biostatistical terms; if it deems an action ethical then it probably is, although if it deems an action unethical then there is still a reasonable chance that it is ethical, of which euthanasia may be an example. Finally, in the last part of the second section (the claim that ATAOs and palliative sedation meet its four criteria), I will consider whether or not the ATAOs and palliative sedation satisfy the DDE’s four criteria and are therefore ethically justified. By considering the differences between tout court (‘without qualification’) and pro tanto (‘to an extent’) bad outcomes, between direct and indirect means of relieving suffering, and between physical and psychological forms of suffering, I will arrive at serious reasons to doubt that the ATAOs and palliative sedation meet the DDE’s four criteria. I will also argue that satisfying the DDE’s fourth proportionality criterion necessitates abandoning the Doctrine of the Sanctity of Human Life, which is often held as a core reason for arguing against euthanasia in the first place. Although my arguments do not establish any definitive proofs, they ultimately culminate to weaken the often-claimed ethical distinction between the ATAOs/palliative sedation and euthanasia.
The empirical claim that the ATAOs, and palliative sedation, do not shorten survival time
This is an empirical question which requires an empirical, quantitative answer. In order to look for the highest quality evidence (meta-analyses and systematic reviews), a literature review of Ovid’s Evidence Based Medicine Reviews database was conducted, which comprises the whole Cochrane Library (Cochrane Database of Systematic Reviews, Cochrane Register of Controlled Trials, Cochrane Methodology Register, Database of Abstracts of Reviews of Effects, Health Technology Assessment, and National Health Service Economic Evaluation Database), as well as comprising the American College of Physicians Journal Club database. The following search strategy was used:
Death* OR dying OR surviv*, AND
Palliat* OR terminal* OR hospice* OR end of life OR end-of-life OR last week*, AND
Opioid* OR morphine OR sedat* OR benzodiazepine* OR midazolam OR barbiturate* OR thiopental OR thiopentone OR propofol.
Keywords were searched for in articles’ titles, abstracts and lists of keywords.
The search was conducted on 18 March 2018 and generated 276 results. Of those, three articles were identified pertaining to our research question,1–3 all of which were in regard to palliative sedation only (not the ATAOs). However, a separate review article4 that did evaluate the ATAOs was identified by searching through the authors from the bibliographies in the aforementioned three articles.
The most comprehensive of the articles evaluating palliative sedation was a Cochrane review by Beller et al,2 which searched the Cochrane Register of Controlled Trials, MEDLINE and EMBASE, limiting results to quantitative study designs (both experimental and observational) provided that they had a control group. Fourteen studies were identified, 13 of which measured survival time, although unfortunately none were randomised controlled trials (RCT) or quasi-RCTs utilising experimental study designs. All were observational studies looking at a consecutive series of patients, and only three used prospective data collection; the others were all retrospective. A total of 4167 adults were included among these studies, 1137 of whom received some form of palliative sedation. Quantitative meta-analysis of pooled data was not possible owing to variation in the methodology of each study, so the outcomes of each article were assessed individually. Among these, none of the 13 studies measuring survival time demonstrated a statistically significant difference between sedated and non-sedated groups. The second and third articles1 3 were both similar and identified 11 studies each, all of which were observational, with four prospective and seven retrospective in each case. They both drew the same conclusion as Beller et al.2 ,ii Importantly, there was a significant overlap in the primary studies between the three review articles; they all referred to some combination of the same 16 observational studies.
However, there are dire shortcomings in this body of evidence, many of which are openly acknowledged by their authors. Most obviously, none of the primary studies were experimental (ie, RCT) designs, meaning that it is tenuous to draw any conclusions about the presence or absence of a causal relationship (rather than just the presence or absence of a correlation). The lack of randomisation also introduces marked selection bias.1–3 Hence, there is a high chance of there being an imbalance of confounding variables between the control and palliative sedation groups, which could skew the results considerably. Furthermore, the patients, doctors and nurses could not be blinded as to whether or not they were receiving/administering palliative sedation, introducing a high risk of performance bias and outcome measurement bias.1 2
It is nonetheless possible to predict some confounding variables, identified in the baseline characteristics table of each primary study (eg, gender, age, type of terminal illness, prognosis, severity of symptoms). These characteristics can then be balanced by ‘matching them’ between control and palliative sedation arms, even in observational study designs. However, only 1 of the 16 primary studies5 attempted to match these characteristics. Despite this attempt, Beller et al 2 reflect, ‘there was a significant difference between the groups in symptoms at admission, with the sedated group having more uncontrolled symptoms, as expected.’ This highlights the fundamental problem in this body of evidence: the indication for palliative sedation (rather than other palliative interventions) is symptom severity, and this cannot be ethically randomised. It would be terribly unethical to provide a placebo to a dying patient with intractable symptoms when palliative sedation is required, so an RCT or an observational study matching control and palliative sedation groups according to severity of symptoms could never be ethically conducted. Although far from a complete solution, Beller et al 2 note that severity of symptoms in an observational study could be matched via statistical methods after the time of data collection; however, this was not conducted in any of the primary studies.
The next major shortcoming in these 16 primary studies is what is actually being measured. In every case, average (mean or median) survival time in both the control and palliative sedation groups was measured from the time of admission to a palliative care inpatient facility, or time of referral to community palliative care services, until death. They did not measure the time from commencement of palliative sedation until death. For example, a patient might be admitted to a hospice on day 1, have a gradual clinical decline with worsening symptoms, have palliative sedation commenced on day 30 and then die on day 35, in which case his/her ‘survival time’ would be measured as 35 days, not 5 days. This surrogate measure bears little resemblance to the question we are interested in. These studies all found that there was no significant difference in average survival time from admission/referral until death between patients who did or did not eventually receive palliative sedation at some time during the course of their illness, and then tenuously extrapolate that palliative sedation itself does not shorten survival time. Beller et al 2 recognise this problem, commenting, ‘the use of time from admission to death in comparative groups may be a weak measure of any effect of palliative sedation shortening life.’ However, they also reflect that there is no feasible alternative measure. The control group does not feature any specific event in time (eg, giving a placebo drug) that could be used as a comparison time point with the commencement of palliative sedation, so the only common time point between the two groups is the time from admission/referral.
Another major shortcoming was that all 16 trials measured a mix of different types of palliative sedation in their intervention arms (ie, intermittent vs continuous, mild vs moderate vs deep). Given that the most extreme form of palliative sedation, CDPS, is presumably the most likely to shorten survival time, any effect that it does have on survival time has been obscured by the other forms of palliative sedation mixed in with it, essentially diluting and underpowering its effect. No primary study has independently compared CDPS with a control.3
Considering all of the above, I think it is underappreciated just how poor the quality of this body evidence is. The ethical limitations on conducting high-quality research should not be accepted as an excuse that justifies us accepting the conclusions from the low quality of research that has been conducted. We really do not know whether or not palliative sedation, in particular CDPS, shortens survival time, so admitting agnosticism on this matter is the correct position to adopt.
On a more philosophical note, although not explicitly stated, all studies defined death according to the ‘whole brain’ criteria, meaning that death has occurred once the whole brain (including the brainstem, which regulates the respiratory and cardiovascular systems) has irreversibly ceased functioning.6 On the alternative ‘higher brain’ definition of death, where death has occurred once a person has irreversibly lost the capacity for consciousness (but may still have brainstem function and may therefore still be breathing and circulating blood through his/her body), then CDPS actually is a state of death and therefore clearly shortens survival time.
Turning briefly to Sykes and Thorns’4 systematic review of survival time in palliative opioid use, it is clear that the methods, conclusions and shortcomings of included primary studies are essentially the same as for those regarding palliative sedation. They identified four retrospective observational studies, meaning that none were randomised experimental studies, and none even included prospective data collection. None were blinded, and they all used odd surrogate measures of survival time as per above. None found a significant difference in survival time between control and opioid arms, although for the same reasons as before, a position of agnosticism is the correct position to adopt.
The ethical claim that the ATAOs, and palliative sedation, are ethically distinguished from euthanasia
Given our agnosticism, suppose it were actually the case that the ATAOs and palliative sedation do hasten or bring about death. They could still be ethically distinguished from euthanasia via appeal to Aquinas’ DDE, which states that an action with two effects—one good and one bad—is permissible if and only if 7:
The action is not bad in itself.
Only the good effect is intended, that is, the bad effect is merely foreseen.
The bad effect is not the means to the good effect, that is, the bad effect is coincidental.
The good effect outweighs the bad effect.
Call these the intrinsic, intention, means and proportionality criteria, respectively. I will now argue that although the DDE is a useful ethical tool, it is not ultimately sound, and even if it were, it is unclear whether or not the ATAOs and palliative sedation meet its four criteria.
The claim that the DDE is a sound ethical principle
The DDE is not a single homogenous principle. It is a strange hybrid of four principles. First, I will describe how the intention and means criteria are consistent with each other, and then I will argue that the intrinsic and the proportionality criteria are not consistent with each other.
The intention and means criteria are closely related and consistent with each other. This is because an action which directly brings about a bad effect (ie, it is an instrument to that bad effect) cannot reasonably claim to merely foresee that bad effect; however, an action which directly brings about a good effect but with a coincidentally associated bad effect can reasonably claim to merely foresee (rather than intend) that bad effect. To illustrate this, I will borrow from a discourse between Foot8 and Thomson.9–11 Consider the Bystander at the Switch scenario where you need to decide whether or not to pull a switch to divert a runaway mine trolley away from a track with five people on it, but onto a track with one person on it. Whether or not the one person was on the other track has nothing at all to do with whether pulling the switch would save the five initial people (after all, the one person being on the sidetrack is just a coincidence), so it can be reasonably claimed that this coincidental bad effect is merely foreseen but not intended. Compare this with the Fat Man scenario where a runaway mine trolley on a track with five people on it could be stopped by pushing a fat man into its course, saving the five people at the expense of the one fat man. The fat man’s presence is important (not coincidental) in this instance; the presence of the fat man is essential because he is the instrument by which the five people are saved. Therefore, it cannot reasonably be claimed that the bad effect of the fat man’s death was unintended and merely foreseen, given that he was the mechanism by which the good effect of saving the five people was achieved.
However, the intrinsic and proportionality criteria do not sit well together and are not internally consistent. The intrinsic criterion is derived from deontology, the predominant form of which is championed by Kant and is grounded on the ‘categorical imperative’. For Kant, an action is permissible if and only if it passes the contradiction test of the categorical imperative; otherwise, it is impermissible.12 There are different formulations of the categorical imperative which are supposedly equivalent (4:436), although to add to the confusion it is contentious as to how many formulations there actually are. If these formulations really are equivalent then any of them can be chosen to evaluate a particular ethical problem because they will always yield the same result. Kant’s Formula of Humanity (the second formulation) is the most applicable within the context of the DDE, which states, ‘Act in such a way that you treat humanity, whether in your own person or in the person of any other, never merely as a means to an end, but always at the same time as an end’ (4:429). Applied to the aforementioned scenarios, pushing the fat man in front of the mine trolley uses him merely as a means, so it contradicts the Formula of Humanity and is therefore an intrinsically bad action which is impermissible without exception. Conversely, pulling the switch does not use the one man on the sidetrack as a means, so it is not prohibited by the Formula of Humanity. At face value, on this account the intrinsic criterion seems consistent with the intention and means criteria.
However, on deeper inspection, on the thoroughgoing Kantian understanding of the intrinsic criterion the DDE is incoherent. On this understanding, the intrinsic criterion contradicts the DDE’s fundamental premise that outcomes can be good or bad; for Kant, only the intrinsic features of actions are the subject matter of ethical evaluation. Ethics is about right and wrong actions, not about good and bad outcomes. By virtue of this the intrinsic criterion also contradicts the other three criteria; because it is incoherent to speak of good and bad outcomes, good or bad outcomes cannot be intended or merely foreseen, cannot be instrumental or coincidental and cannot be weighed against each other. Even if there were good and bad outcomes, the intrinsic criterion makes the DDE circular and does all of the explanatory work, rendering the other criteria superfluous. The DDE states that for an action to be permissible it must be intrinsically not bad, but for Kant an action just is permissible if it is intrinsically not bad; hence, the DDE states that for an action to be permissible that action must be permissible. If this intrinsic criterion is met then the DDE is circular and has already determined that that action is permissible; the other three criteria play no ethical role. There is the additional problem that the intrinsic criterion may be manipulated by construing the same action in different ways. For example, consider the intrinsic criterion in relation to killing in self-defence. If the action is construed as ‘defending oneself’ then the intrinsic criterion is met, whereas if the action is construed as ‘shooting someone in the chest’ then the intrinsic criterion may not be met.
The proportionality criterion is derived from consequentialism, the predominant form of which is the ‘classical utilitarianism’ championed by Mill. For Mill,13 happiness (or pleasure) is the currency of ethical evaluation. This includes the ‘higher’ pleasures, for example, enjoying poetry, and the ‘lower’ pleasures, for example, enjoying ice cream, although I will turn a blind eye to this probably unjustified elitism. According to the ‘principle of utility’ or ‘greatest-happiness principle’, an action is permissible (perhaps even obligatory) if it produces a net increase of happiness, impermissible if it produces a net decrease of happiness and ethically neutral if it produces neither a net increase nor decrease of happiness. The calculation incorporates all those affected by the action rather than just the individual agent. The classical utilitarian responses to the Bystander at the Switch and Fat Man scenarios discussed earlier are that the actions of pulling the switch, and pushing the fat man, both save a net of four lives (five minus one) and therefore produce a net increase in happiness; hence, acting in both of these scenarios is permissible, perhaps even obligatory, and on the dedicated utilitarian account there is no ethical difference between acting in the two scenarios.
On the thoroughgoing Millian understanding of the proportionality criterion the DDE is also incoherent. Despite being consistent with the DDE’s premise, the proportionality criterion contradicts the intrinsic criterion. All actions are good or bad by virtue of their producing good or bad outcomes; it is incoherent to speak of intrinsically bad actions. The proportionality criterion also makes the DDE circular and does all of the explanatory work, rendering the other criteria superfluous. The DDE states that for an action to be permissible its good outcome must outweigh its bad outcome, but for Mill an action just is permissible if its good outcome outweighs its bad outcome; hence, the DDE states that for an action to be permissible that action must be permissible. If this proportionality criterion is met then the DDE is circular and has already determined that that action is permissible; the other criteria play no ethical role.
Given that Kantian deontology and Millian consequentialism are fundamentally incompatible, it is unsurprising that the DDE’s attempt to combine the two runs into difficulties. Both Kant and Mill would have been horrified at the thought of attempting to mesh their theories together. The thoroughgoing Kantian and Millian conceptions of the intrinsic and proportionality criteria, respectively, are estranged principles artificially forced together by the DDE. They need to be sacrificed for a compromising position where both the intrinsic features of actions and their outcomes form the subject matter of ethical evaluation. The two somehow need to be reconciled, and how this should be done is elusive. However, this is a problem that every attempt at doing pluralistic ethics inevitably encounters, including Beauchamp and Childress’14 principlism which is often applied in the medical literature. Admittedly, the DDE is respectable as a genuine attempt at doing pluralistic ethics. It is an excellent example of ‘mid-level theory’ as described by Arras15; it distils principles from ‘high theory’ (ie, purist deontological and utilitarian theories) and combines them as necessary but individually insufficient criteria for an action being ethical, applied in the context of actions with simultaneous good and bad effects. By deriving criteria from deontology and consequentialism, as well as from the fully intended versus merely foreseen and instrument versus coincident distinctions, the DDE reflects the central tenet on which pluralistic ethics is grounded: ethical principles can converge on agreed-upon conclusions despite their coming from irreconcilable starting points.16
However, this all-inclusive combination is in some ways ad hoc. The DDE has been criticised as being ‘a piece of complete sophistry’,8 as being merely a proposition which requires—although lacks—supporting evidence,7 and as being ‘artificial and indeed Jesuitical’.17 However, it has one major advantage: it is safe. In biostatistical language it is specific but not sensitive; there will not be many ‘false positives’ (unethical actions which are erroneously deemed ethical) but there may be many ‘false negatives’ (ethical actions which are erroneously deemed unethical). In requiring that all four criteria are met, the DDE sets a high standard for an action’s (with a good and a bad effect) being ethical. Consider the Bystander at the Switch scenario: pulling the switch meets all four criteria so it is a safe bet that pulling the switch is permissible. However, despite being useful, the DDE is ultimately unsound because it is too safe.
We can prove this by devising a scenario in which an ethical action is deemed unethical by the DDE. Suppose that Bystander at the Switch were altered so that there was a piece of additional track behind the five on the first track, and behind the one on the second track, which are then joined together in a loop, connecting the two tracks. Call this the Loop Variant scenario (figure 1) as described by Thomson.10 Pulling the switch so that it hits the one will kill him/her but will stop the trolley, preventing it from continuing on its course and hitting the five from behind. This makes the one person the instrument to saving the five (like in the Fat Man scenario), so the means criterion and subsequently the intention criterion are both failed, hence the action is unethical. However, I invite you to agree with me that a small piece of additional track which the trolley will never even run on irrespective of whether the switch is pulled or not cannot possibly be ethically significant (ie, if the switch is not pulled, the trolley stops after hitting the five, and if it is pulled, the trolley stops after hitting the one). If pulling the switch is ethical in Bystander at the Switch then it must also be ethical Loop Variant; the two cannot have different ethical statuses. If the DDE says that an action is ethical (eg, acting in Bystander at the Switch) then it probably is, but if it says that an action is unethical (eg, acting in Loop Variant) then there is still a reasonable chance that it is ethical. I think that euthanasia may fall into this latter category.
The claim that the ATAOs, and palliative sedation, meet the DDE’s four criteria
Now suppose for the sake of argument that the DDE were sound (and therefore a successful way of ethically distinguishing euthanasia from the ATAOs/palliative sedation). Also grant that the supposedly bad outcome of death arising from the ATAOs and palliative sedation really is a bad outcome (although some have challenged this, such as Len and Leslie Doyal18 19). We now need to ask whether the ATAOs and palliative sedation meet the DDE’s four criteria.
First, consider the intrinsic criterion through some examples. Suppose that Hamish requires you to administer him opioids for cancer pain, and that Kathryn requires you to administer her temporary sedation to unconsciousness for a wisdom tooth extraction. Acting is not unethical in either of these scenarios. However, if Hamish’s pain could be relieved without opioids or Kathryn’s tooth extracted without temporary sedation, those might be better states of affairs. Hence, although administering opioids or temporary sedation in these scenarios is not intrinsically bad tout court (‘without qualification’) it might be intrinsically bad pro tanto (‘to an extent’). Put simply, administering opioids or temporary sedation might always occupy the ‘cons’ column by virtue of their intrinsic nature, although sometimes those cons may be outweighed by ‘pros’ such as relieving cancer pain or extracting a wisdom tooth. If this is the case, these actions may fail the intrinsic criterion (depending on whether it applies to pro tanto bads or only to pro court bads); these actions are intrinsically bad to an extent but not so much that they cannot be outweighed by more pressing concerns. The ATAOs and palliative sedation (but not CDPS) may also fall into this camp.
In addition, CDPS might be intrinsically bad without qualification; being permanently sedated to unconsciousness is indistinguishable from being killed from the patient’s perspective. By definition, if a patient is permanently sedated to unconsciousness then s/he does not and will not ever have any conscious experience again, in the same way that a dead person does not and will not ever have any conscious experience again. There might be a significant ethical difference for the patient’s family, who can spend some additional last moments with their sedated loved one for their own emotional benefit, but at a metaphysical level and from the patient’s perspective there is no difference here. If killing is bad without qualification then it is reasonable to say that CDPS is as well, therefore violating the intrinsic criterion.
To investigate the intrinsic criterion further we can attempt to directly apply Kant’s categorical imperative, in particular his Formula of Humanity which we have seen bares immediate relevance to the DDE. Are the ATAOs and palliative sedation, including CDPS, permissible according to the Formula of Humanity? In other words, do palliative opioid and sedative use at the end of life, for the purpose of respecting autonomy and relieving suffering (even when those drugs shorten survival time), use the patient as merely a means to an end, or also at the same time as an end? It is difficult to understand exactly what this question means, apart from recognising that it is clearly related to the intention and means criteria. Furthermore, to answer this question it is important to know whether the suffering is purely physical or whether it also has a psychological component.
First, consider purely physical suffering. The ATAOs can combat physical symptoms (eg, pain) directly, and palliative sedation can combat the awareness of physical symptoms (eg, the awareness of pain) directly, independent of whether or not they shorten survival time. However, supposing that they do shorten survival time, then death itself can also be considered a means to relieving physical suffering; you cannot suffer from pain if you are dead. Call this an indirect means (‘indirect’ because the ATAOs or palliative sedation brings about death, which in turn brings about the relieving of suffering, in a longer and therefore ‘indirect’ chain of events). Considered in this way, the ATAOs and palliative sedation are sufficient for relieving physical symptoms in a direct manner, but death is also sufficient for relieving physical symptoms in an indirect matter. Neither are necessary independently (the direct means will achieve the relief of suffering regardless of the indirect means, and vice versa), making each sufficient but not necessary, and therefore resulting in a state of overdetermination. Now, the direct means does not seem to violate the means or intention criterion (or the Formula of Humanity); however, the indirect means does violate all of these in a similar way to in the Fat Man scenario discussed earlier. The overall problem is that we are speaking as if there are two separate actions here, but there are not; there is only one action being considered in two different inseparable ways (a direct and indirect way) or with two different inseparable components (a direct and indirect component). Whether or not the action as a whole violates the means and intention criteria is unclear, because the means criterion has not been crafted to accommodate situations of overdetermination. For the same reason, it is also not clear whether the action as a whole contradicts the Formula of Humanity.
However, if there is also a psychological component to the suffering (eg, existential suffering, or the suffering associated with the loss of independence, mobility, memory and the capacity to engage in meaningful pursuits) then there will be instances in which the ATAOs and palliative sedation (save for CDPS) simply cannot relieve all of that suffering directly, and that it can only be relieved indirectly after death has been brought about. For example, suppose that in the face of death a patient reflects on his/her life and has deep regrets about major life decisions, or views his/her life as meaningless or unfulfilled, or is preoccupied with thoughts of bereavement or failed relationships, or is upset that s/he will never be able to do his/her favourite activities again or watch his/her grandchildren grow up. In this case, no amount of opioid analgesia or intermittent sedation is able to circumvent this suffering; psychotherapy and some other drugs such as antidepressants may play some role but surely cannot always be effective. In this instance, death (or CDPS) is necessary and sufficient for the suffering to be relieved, so the means criterion (and therefore also the intention criterion) are clearly violated, and along with them the Formula of Humanity.
Finally, consider the proportionality criterion. For it to be fulfilled, the relieving of suffering (the good outcome) has to trump the hastening or bringing about of death (the bad outcome). If this is the case then not hastening or bringing about death cannot be of insurmountable ethical importance, that is, it can be trumped by other ethical considerations and is therefore a pro tanto bad rather than a pro court bad. This is not a problem in-and-of itself, although it is worth noting that it contradicts the Doctrine of the Sanctity of Human Life (including when formulated in a non-theistic way), which claims that human life is inviolable; ending human life is a pro court bad. In order to maintain that the ATAOs and palliative sedation satisfy the proportionality criterion and can therefore be justified by the DDE, the Doctrine of the Sanctity of Human Life has to be surrendered. Either human life is of the utmost ethical significance, in which case the proportionality criterion of the DDE cannot be satisfied, or human life is not of the utmost ethical significance (and can be outweighed by other considerations such as pain relief), in which case the proportionality criterion can be maintained but the Doctrine of the Sanctity of Human Life has to be abandoned. Many opponents of euthanasia (and hence proponents of its distinction from the ATAOs and palliative sedation) will not like this sacrifice, given that it is one of the fundamental reasons—although of course not the only reason—for being opposed to euthanasia in the first place.
In summary, I have argued firstly that we should be agnostic as to whether or not the ATAOs and palliative sedation shorten survival time, secondly, that although being a useful ethical tool, the DDE is ultimately unsound by virtue of setting the bar too high and erroneously deeming some ethical actions unethical (of which euthanasia may be an example), and finally, that even if the DDE were sound, there are a number of serious doubts as to whether or not the ATAOs and palliative sedation satisfy its four criteria, and they cannot do so unless the Doctrine of the Sanctity of Human Life is abandoned. Although this essay does not establish any definitive proofs, it provides arguments that weaken the often-claimed ethical distinction between euthanasia and the widely accepted mainstream practices of palliative opioid/sedative use.
The author thanks Dr Michaelis Michael and Dr Markos Valaris for their academic guidance during the researching and writing of this essay. He also thanks his family and friends for their support and feedback during this time.
↵ i I have deliberately not defined the term ‘euthanasia’ since there is considerable disagreement among interlocutors as to what the correct definition of the term actually is, and as to whether or not certain non-paradigm cases constitute euthanasia. This is the topic of another essay. For the purposes of this essay it will suffice to focus on the aforementioned paradigm case.
↵ ii It is worth noting that another study20 (not identified in our search strategy) demonstrated a survival benefit among patients with metastatic non-small cell lung cancer receiving palliative care services introduced earlier than in the control group. This study evaluated the effect of palliative care services as a whole (including a range of ‘active’ and ‘passive’ practices), although did not specifically evaluate the ATAOs or palliative sedation compared with a control group that did not receive those interventions. This study is therefore not immediately relevant to our research question; it does demonstrate that early, good symptom control introduced soon after diagnosis can prolong life, but it does not address whether or not the ATAOs and palliative sedation at the end of life shorten survival time. Furthermore, the control group that had a shorter survival time turned out to receive more aggressive palliative care interventions at the end of life, which may actually weigh against the evidence from the other aforementioned studies.
Contributors The author conceived, researched, wrote and edited this essay.
Funding The author has not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.