Killing or Letting Die, a series on decision-making when death is at the table:
Part 1 - Intending or Pretending?
Part 3 - Causation or Conflation?
Part 4 - The Priority of Authority?
I hesitated. What were we doing here?
The plan was to withdraw a ventilator from someone who had been dependent on it for over a year due to a degenerative neurological condition. After much deliberation, we agreed: today, one like any of the others for the past many months, would be the day. Sidled up alongside the feeling of arbitrariness was the feeling that we weren’t just withdrawing a ventilator; we were planning this person’s death.
Is there a difference between doing and allowing, between killing and letting die? James Rachels penned his famous paper “Active and Passive Euthanasia” in the New England Journal of Medicine in 1975 is an attempt to answer this question. The debate, like a restless sea, continues to surge and recede.
Rachels reflected on how some children with Down syndrome were treated after being born. If they had some lethal but easily correctable defect (e.g., bowel obstruction), they might be allowed to die. However, if they didn’t, then they lived on, shielded from any intent to actively end their lives. It would be wrong, presumably, to euthanize them. Rachels grumbled about the total moral irrelevance of the bowel obstruction. Either the child’s life is worth living, in which case they should get the surgery, or else the child’s life isn’t worth living, and why would it matter if their bowel was unobstructed? Since you would never allow an otherwise healthy child to die of an easily correctable bowel obstruction, we see that the primary moral criterion to determine a “life worth living” in this scenario is whether the child has Down syndrome.1 The moral calculus that leads people to “let die” should also permit them to “kill.”
To further refine his point, Rachels leads his reader to a pair of cases. In one, a man forces his young cousin’s head underwater in the bath to drown him for a grand inheritance. In the other, the boy slips, hits his head, and slides under the water. The adult watches on and does nothing, allowing the boy to drown. If there were some difference between killing and letting die, then we would say the first man did something worse than the second. We don’t. The behavior of both men was reprehensible. Using this logic, Rachels then argues that if we find withdrawing or withholding life-sustaining therapies to allow someone to die acceptable, we should also find euthanasia acceptable. There’s no moral difference.
Rachels takes for granted that the intention behind withdrawing or withholding treatment is to end someone’s life. This results in a situation where someone dies, maybe in agony, over the course of hours to days. A death from euthanasia would be quicker and less burdensome. If the intention is to end someone’s life, euthanasia is far more effective than withholding or withdrawing life-sustaining therapy.
Other authors have brought still more cases - of nephews engaged in clandestine ventilator disconnection to firemen trying to save people leaping from burning buildings. These cases are attempts to find, if there is one, any difference between acting and abstaining from action; between killing and allowing to die. It’s almost like ethicists have been working backward from an intuition that they are indeed different.
An intuition or a fantasy anyway. Let’s take a leap forward to 2010 when Franklin Miller, Robert Truog, and Dan Brock wrote a paper entitled “Moral fictions and medical ethics.” They draw our attention to these strange little entities, called “moral fictions,” which pepper the landscape of medicine. These are “false statements endorsed to uphold cherished or entrenched moral positions in the face of conduct that is in tension with these established moral traditions.” So, in medicine, we say first do not harm, and never kill your patient. But we also withdraw and withhold medical therapies which result in death all the time. The moral fiction we’ve created to assuage this dissonance is that letting someone die is somehow different from killing them. These authors bring forward their own pair of cases in which a paraplegic man requires a ventilator in one case and doesn’t in another. Both men wish to die. In the first case, the man can request to stop his ventilator and die. In the second, we say he must live because euthanasia is not available for him.2
Miller, et al.’s argument is nuanced but helpful. You should check it out. They help their reader to look closely at the ethical landscape, from intention to causation to responsibility, surrounding this issue. They eventually conclude that stopping a ventilator is an active intervention, that the clinician who does it causes the man’s death (that is, they kill the man), that what occurs is called suicide, indeed assisted suicide, that the clinician is morally responsible for this death, and that what they do is permitted. They conclude that euthanasia should be just as permissible as withdrawing a life-sustaining therapy.
In surveying these two papers, you might see that the devil’s in the details and definitions. How do we define “kill,” “let” (or “allow”), and “intend?” What are other important factors? If we can’t agree on our terms and emphases, we’ll forever talk past one another. But we can also use these discrepancies to uncover some intuitions that might make disagreement more productive.
For the next few issues, we’ll focus on this question of killing vs. letting die, which is just one facet of a bigger discussion (e.g., determining a life worth living, whether death can ever be a goal of care). We’ll start our ethical safari tracking the concept intention, and then see what else turns up along the way.
A couple of presuppositions before we head out: I believe the purpose of medicine is to sustain and restore health. I also believe that clinicians should never intend the death of their patients. You might disagree with me on those points. If you want to help me refine my thoughts on them, please venture over to those respective issues to leave a comment.
What do I intend?
Rachels overlooked a major facet of moral consideration in his paper: intention. Two actions can be the same in every respect but different intentions determine that one is right and the other wrong. I don’t shake your hand because I’ve been sick and I’m trying to avoid spreading contagion. Or, I don’t shake your hand to disrespect you. Same action, different intention, different evaluation. Daniel Sulmasy remarks, rightly I think, that “morally wrong intentions make an act morally wrong,” but he’s quick to add “there are plenty of other aspects of an act that are part of its moral evaluation.”
What is an intention?
Sulmasy argues there’s a difference between an intention and a desire, as any New Year resolver knows. You might intend to improve your diet even as you want to eat Girl Scout cookies every day. You can want those cookies without intending to eat them. Likewise, you may intend to clean up your kitchen even if you don’t desire it. There’s a complex interplay of desires, beliefs, and motivations that inform our intentions. An intention is that which we’re committed to achieving.
In healthcare, an incomplete list of factors that inform a clinician’s intention might be the law, their profession’s code of ethics, their personal experience, the standard of practice, their technical expertise, the patient’s goals of care, as well as any number of cognitive biases. The clinician relies on these, consciously and unconsciously, to form a commitment to a particular goal.3
Intentions are open to critique just like actions, but may not entirely mitigate disapproval. Consider the well-meaning quack. The quack doesn’t intend to deceive anyone. They earnestly believe in the treatments they’re offering, but their good will outstrips their competence. Their well-meaning intention isn’t sufficient to outweigh the fact that they should know better. This brings us back to Sulmasy’s point that intention is just one factor among many we use to evaluate an action.
But isn’t intention “multilayered, ambiguous, subjective, and often contradictory?” Yes, we can be ambivalent in our intentions and motivations. Even as we act, we may remain ambivalent about our goals. It doesn’t mean we dispense with intention entirely in how we evaluate action, only that we consider it given all the other evidence.
It’s also true that intention is only fully accessible to the person acting. I can tell you what I intend, but you need to believe me. In the example of the handshake above, it makes all the difference that I didn’t snub you but I actually cared for your well-being by withholding my handshake. You wouldn’t know that unless I told you though. Only I can know my true intentions (insofar as they can be truly known). Nevertheless, intention remains indispensable to determining the morality of an action. Criminal law depends on it, in whether we determine an act of killing as first degree murder or manslaughter. Most commonplace behavior depends on it, in whether I brought my wife flowers because I love her or if I brought them to flatter her for a request I’m later planning to make.
Do I intend all the consequences of my actions?
If someone answers, “No,” they might be worried that I’m able to evade accountability for some bad outcome. But we’ve already determined that a lot more is needed to evaluate an action than the intention alone. Some people still say that anything I foresee I also intend by undertaking a certain course of action.
I disagree. I think there are important differences in how we hold people accountable for the things they intend versus the things they foresee. How do you figure this out? “What am I aiming for, and how can I achieve it?” are the questions that accompany intention. You don’t deliberate on how you might bring about something foreseen.
Foresight is still important. We hold people accountable for things they should have foreseen but didn’t. As an expert in prescribing opioids, it wouldn’t help me to claim that I didn’t foresee that someone would have become sedated from high doses of morphine. We also hold people accountable for things they did foresee but didn’t appropriately weigh in their decision-making. If I prescribe those escalating doses of morphine for a hangnail until someone is unconscious, it won’t help my case to say that their sedation was a foreseen but unintended consequence of the prescription. Alison McIntyre argues that it’s quite a challenge to figure out whether a foreseen but unintended harm was proportional to the good I tried to bring about.
Who gets to decide whether death can be permissibly foreseen?
This issue of proportionality is a tough one. Clinicians are always weighing the burdens and benefits of their interventions. There’s no special math for this. The weighing is not only subjective, but might vary from day to day based on any number of factors (whether a patient is in pain, whether a patient’s spouse is with them, whether a patient is talking to a clinician they like, who the clinician is, etc.).
McIntyre points out one thorny challenge: whatever justifies causing harm as a foreseen side effect might also justify its use as a means for achieving a good end. So, for example, someone with respiratory failure needs a ventilator to live. We decide to stop the ventilator to remove a burdensome medical intervention. Death is a foreseen but unintended consequence of the decision to stop the ventilator. It is, therefore, a permissible harm, an accepted side effect. In a different case, we may not have so understood their death - e.g., if their disease was easily reversible, they were otherwise healthy, and we anticipated they’d be easily returned to their normal life.
But if we can permit death as a side effect, why can’t we use it as a means to achieving a good end? Instead of merely accepting death, we could put it to use! My response is that we shouldn’t ever intend the death of our patients. This would include intending to use death as a means toward some other end (e.g., relief of suffering). Death may still be a permissible side effect of one’s action though.4 Death is a harm, but there are other harms to consider. We need to take these into account, recognizing that death is inevitable. We can harm dying people in ways only the dying can be harmed and therefore our obligation to work to sustain life isn’t absolute.5 There comes a time to acknowledge when someone is dying. We can help them with that or leave them to it, but to force them in another direction might mean causing greater harm even as they die anyway. This means death may be permissible as a side effect but doesn’t justify its use as a means.
Another issue that McIntyre recognizes is if we’re able to foresee a harm, we’re obligated to minimize or avoid it. If there’s another course of action that can avoid this harm altogether, we should consider doing that instead. We shouldn’t shrug in the face of opioid-induced constipation. Nor, when considering a foreseen death, should we sedate someone so they can no longer eat or drink if it means that would shorten their life. We wouldn’t say these harms are intended, but we’d be careless not to mitigate them.
With those considerations in view, let’s get on with the question: who gets to decide whether death is permissible as a foreseen side effect? It can’t be the patient alone. There are plenty of patients who throw caution to the wind, begging their surgeons to put them under the knife. They don’t care if they “die on the table.” Any chance that surgery might succeed is acceptable to them, no matter the risk. But the surgeon has a say here too. The surgeon may believe that the therapeutic possibility of the operation is so low and the harm so likely that they can’t possibly, in good faith, intend any good from it. That fulfilling the patient’s mere desire is insufficient for the surgeon’s intention is telling. Clinicians aren’t tools in the hands of their patients.
Patients may lack the experience with these kinds of decisions to weigh the burdens and benefits. There is a huge existential difference between making a decision for your own health as a patient and helping a patient make a decision, but clinicians can lend their experience and expertise to the weighing and offer patients some guidance. Therefore, the permissibility of death as a foreseen side effect is negotiated among the patient, clinician, and broader society.
What am I intending if I “allow someone to die?”6
Sulmasy’s definition of “letting die” is helpful here. When we let someone die, we remove something that “forestalls or ameliorates a preexisting fatal condition,” or we refrain from forestalling or ameliorating that condition. Because we may or may not intend someone’s death by letting them die, all the aforementioned reflection on death as a foreseen but unintended consequence of one’s action needs to be brought to bear on these cases. I may withhold or withdraw a therapy that results in someone’s death but I may not intend their death by doing so. My intention isn’t fulfilled in their death. It may be fulfilled in their experiencing peace without the intrusion of burdensome medical technology. Nevertheless, the statement “allow to die” or “let die” acknowledges that I have some limited power over the timing and quality of death. This doesn’t seem in any way similar to intending someone’s death through euthanasia.
Who is capable of forming the intention to “allow someone to die?” Not just anyone. In order to allow something, I need the capacity and authority to act otherwise. If I don’t, it doesn’t make sense to speak of me intending to allow someone’s death, although I may desire it. That desire might earn me disapproval, but less than if I intended their death. I can only reasonably intend those things that I’m authorized and capable to achieve. I don’t know how to remove a patient’s inflamed appendix. I could, however, call a surgeon for help, or prescribe antibiotics. If I can’t even do those things, and the patient dies, I can’t be said to have let them die.
What if the patient intends to die?
A patient may intend their own death by stopping a life-sustaining therapy. In this case, the patient has revoked consent for the therapy. What can a clinician do?
They may share the patient’s intention: they’ll stop the life-sustaining therapy so that the patient will die. If the patient survives, the clinician will have failed.
They may not share the patient’s intention and suspect the patient lacks the capacity to revoke consent. The clinician will continue the therapy over the patient’s objection. This is a common practice in ICU delirium where patient’s are restrained to prevent accidental harm to themselves by pulling out a breathing tube or an IV line. If they believe the patient to be depressed and suicidal, though, the case becomes more complicated. Delirium might resolve with treatment of the underlying medical illness. There are no guarantees that treating the medical illness requiring a life-sustaining therapy will improve depression and suicidality, so there’s no limit on the amount of time that this patient may need to be treated over their objection. The burdens appear to weigh more heavily there.
Finally, the clinician may not share the patient’s intention but withdraw the life-sustaining therapy anyway. They can’t proceed without consent but they won’t intend their patient’s death. Death is neither the means (i.e.., toward the end of respecting the patient’s choice) nor the end itself. Their intention is to not further harm the patient by overriding their liberty, and to stop a medical therapy that likely has greater burdens than benefits for this patient.7 It may also be the case that the burdens of treating someone over their objection may be so great that it’s less harmful to allow their death (as in the case of considering indefinite life-sustaining therapy over the objection of someone who is depressed and suicidal).
With death in view, we might forget about dying. As Thomas Fuchs puts it, “The aim of killing is death, the aim of cessation of treatment is dying.” I’d replace that second “the” with “an,” but his point is well-taken: we can intend to help patients move from the sick role to the dying role by stopping a therapy that is only intended for sick people and not for dying people. This doesn’t mean we intend their death.
Why do we prioritize liberty over life in those cases when death is foreseen?
For these cases just mentioned, at some point liberty is prioritized even over the preservation of life. Curlin and Tollefsen shine some helpful light here. They argue that the good life requires us to make commitments to pursue those things that make it good. “Commitments are real,” they argue, “only if they are the person’s own commitments. … some human goods cannot be realized at all unless individuals are free to make and follow through on their own commitments.”
There’s a place for treatment over objection, and treatment to the standard of “best interests.” We’ll have to get to that discussion another time. In the vast majority of cases, though, clinicians don’t have the authority to treat someone over their objection. We require a patient’s consent to intervene, to even touch them. If they revoke their consent, we must revoke our intervention.8 Curlin and Tollefsen help us to see the deeper moral foundation to what’s well-established in American law. The patient herself is in the best possible position to determine how to situate her own health in the greater context of her whole life. Health, to be sure, is a good. It’s unique, even foundational, in that it’s the good through which we experience any other good.9 But it isn’t the greatest good. So a patient may prioritize time at home with their family, their finances, cake for dinner, or cigarettes more than their health.
To treat someone over their objection who has the capacity to revoke their consent (or decline to consent) is a malformed, even evil, attempt to live their life for them. A clinician might lament a patient’s behavior and a patient’s reasons for that behavior, but that clinician isn’t helping to force the patient to do otherwise.
Where does that leave cheerleading and negotiation? Patrick is twelve days post-op from a bilateral lung transplant. Things are not going well for him. He begs his care team to let him die. His care team, though, believes that with just a little more time, he’ll turn the corner and thank them for their efforts. So, too, with Lisa, who lost both her legs and one of her arms in a car accident. Her care team, having seen it before with other patients, believes she’ll adjust if just given a bit more time.
Cheerleading isn’t always right; it isn’t always wrong either. Blind deference to a patient’s expressed preference is too shallow to sustain clinical practice. Clinicians and patients negotiate their way to health.
Summary
So, is there a difference between killing and allowing to die? Maybe. At least there can be. We’ve seen that intentions matter. Wrong intentions are trump cards, making the actions that flow from them wrong. We’re responsible for those things we foresee accompanying our intended means in pursuit of our intended goals. We negotiate our intentions with other people, including our own patients.
Responsibility, authority, and causation matter too. We’ll turn to those next.
The analogy is Rachels’, not mine. People with Down syndrome have lives worth just the same as anyone else’s - that is, they’re beyond valuing.
This legal question (at least from the standpoint American jurisprudence) was settled in Vacco v. Quill, but the ethical question persists. The Supreme Court of the United States determined there were important differences between withholding or withdrawing life-sustaining therapy and intending to end a patient’s life. Therefore, patients could not appeal to the Fourteenth Amendment for a right to assisted suicide by arguing that since someone could end their life by stopping life-sustaining therapy, they should be able to end their life via other means if they don’t need life-sustaining therapy. To say there is no right is not the same thing as to say there is a violation of a different right, so there’s nothing in this ruling that prevents the creation of assisted suicide laws (which have blossomed all over the country).
You might see why it’s impossible for the patient’s preference to be the sole factor in determining the goal of a medical intervention and thus the clinician’s intention. More is needed for a clinician to form a commitment to act toward a particular goal (or abstain from acting, also with a particular goal in mind).
Self-sacrifice might be the most well-known non-medical example of this: someone sacrifices their life to save the life of another person. The intention wasn’t to die but to save their friend. Death was (probably) a foreseen, unintended, but acceptable side effect of their decision.
One way an absolute duty to preserve life comes into existence is by discarding any meaningful script of dying. Thus the response to impending death may take one of two polar forms: either a voracious clinging to medical technology to sustain the barest physiological functions (longevity at any cost), or an efficient death with as little dying as possible (euthanasia).
I take for granted that killing a patient is always wrong, so I’m not going to spend time on that here.
Well-functioning permanent pacemakers present a unique challenge to this rationale.
Usually. I don’t think a patient can claim that they’re continually consenting to a pacemaker remaining activated in their chest and therefore can revoke that consent at any time. Even if they were continually consenting to it, their revocation may not require a clinician to deactivate it for the reasons I discuss in the aforelinked pacemaker issue.
This means many try to remediate problems in other areas of life by appealing to a medical intervention. The management aberrant behavior with psychopharmacology and the history of eugenics provide two stark examples.
Thank you for a really thought-provoking, and helpful, article. It takes me back 30 years to my first year out of med school (UK). On call overnight looking after a ward of patients with heart failure and/or COPD, many of them on diamorphine pumps. And then you'd get a call by the nurse in charge asking you to increase the dose to relieve their distress, knowing full well that you were hastening their end. Back then we had no such considered discussions, which probably explains why we managed it with morbid and often inappropriate humour.
Another great post Joshua. I look forward to Part 2.
I'm struck that we both published posts using Caspar David Friedrich's paintings on the same day.