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?
We return to the paper by James Rachels with two infamous uncles.
He described two cases to help make his point that there’s no moral difference between killing and letting someone to die. In one case, a man forces his young cousin under water to drown him. In the other case, the cousin slips, hits his head, and slides under the water. The man watches on and doesn’t lift a finger. Rachels argued that if there were a true difference between killing and allowing to die, we would claim the man in the first case behaved worse than the man in the second. We don’t, and therefore Rachels concluded there’s no difference.
In the first case, the uncle clearly caused the boys death; he would be alive if not for that murderous man. In the second case, though the uncle didn’t act to drown the boy, he could have easily saved him. We can’t say that he caused the boy’s death in the same way, but we can blame him for it. We do this by reflecting on the uncle’s duty, which we explored in the Part 2 of this series. However, some, like Rachels, claim that if one is permitted to allow a particular outcome, one should also be permitted to act to bring that outcome about.
Let’s think through this a bit.
How does understanding causation help us evaluate an action?
This is surprisingly not straightforward.
Alicia trips and pushes Liam into traffic. Alicia might escape blame because it was an accident, even if she was the cause of harm to Liam. In contrast, the second uncle wasn’t involved in the chain of events that lead to the boy’s death but is still blameworthy for reasons we’ve already discussed related to intention and duty.
Cause is one way of assigning responsibility. If you caused a particular outcome, it may be the case that you’re responsible for it some blame- or praiseworthy way.1 As we saw when we explored intention, someone might act to intend one outcome, knowing that another undesired outcome is foreseen but not intended. However, they still cause that latter, undesired outcome. How that causation impacts the evaluation of the action depends on a number of things, chief among them how the actor weighed the proportionality of their action against other possible actions, and the good outcome against the bad outcome. It wouldn’t make sense to weigh these things if the bad outcome was going to happen no matter what, nor would it make sense to hold someone accountable for an outcome outside their control.
Understanding a clinician’s role in causing a particular outcome is one facet of evaluating their action, but it doesn’t settle the matter - just like with intention and duty.
Does a clinician cause the patient’s death when they kill them?
The answer is obviously yes. A patient is alive, and would still be alive (for however short a time) had the clinician not ended the patient’s life (which is the most straightforward understanding of what it means to “kill”).
For most of human history, a clinician who killed their patient would be guilty of murder. Now there are special allowances in some countries for euthanasia. The evaluation doesn’t change based on determining a different cause; in either case, the clinician is indisputably the cause of the patient’s death.
Does a clinician cause the patient’s death when they withdraw or withhold life-sustaining therapy (LST)?
Here’s the trickier question.2
Returning to the paper by Franklin Miller, Robert Truog, and Dan Brock, they argue that “withdrawing life-sustaining treatment, when followed shortly by the patient’s death, is a life-terminating intervention. … the power to sustain life by technological means goes hand-in-hand with the power to end life when these means are withheld or withdrawn.” They don’t see a difference between killing and allowing to die when they’re reckoned on causation (or a number of other factors). Clinicians really do kill their patients when they withdraw LST. To claim otherwise is a moral fiction.
Is that so?
Let’s approach this question with the help of Joachim Asscher, whom I introduced in the essay on duty. Asscher was careful to argue that someone might have responsibility for an outcome if they created the circumstances that allowed that outcome to occur, even if they weren’t directly involved in bringing the outcome about. So, Alicia may bear no responsibility for tripping into Liam and making him fall into traffic. However, if she were tossing a frisbee with a friend along the sidewalk, stumbled, and pushed Liam into the street, Asscher would say Alicia has “surrounding responsibility” for what happened because of her carelessness. She contributed to the circumstances that led to the accident.
In a medical context, this surrounding responsibility belongs to clinicians who manipulate the body for any number of reasons. You might be hard-pressed to justify that a surgeon caused a post-operative blood clot in a leg. However, the surgeon created the circumstances that allowed for this to happen; they have this “surrounding responsibility.” The patient wouldn’t have developed the blood clot but for the operation the surgeon performed. The patient also shares this surrounding responsibility (probably to a lesser degree) because they wouldn’t have developed the blood clot but for their consent to the operation.
Asscher argues that there may not be a difference between killing and letting die because when a clinician has surrounding responsibility for a patient’s life-threatening situation, there may be no extra responsibility to take for killing someone compared with letting them die. The clinician is responsible for the situation in which the patient will die if LST is withheld or withdrawn, so, from this standpoint, it’s just the same as if they killed the patient. “Where medical staff have a high degree of surrounding responsibility, it is inappropriate to appeal to the distinction to justify letting die.” It doesn’t make sense to claim that letting die is different from killing some situations because clinicians have this “surrounding responsibility” while a patient’s life is in their hands.
With that in mind, let’s look at “letting die” a bit more closely.
When someone lets something happen, they’re indirectly involved in contributing to an outcome that was already going to occur. This is different from surrounding responsibility. “Let” or “allow” implies both the capacity and authority to intervene to produce a different outcome. One actively chooses to let something happen. So, if I let my lawn overgrow, I could have mown it, and I had every right to mow it if I wished. I didn’t cause my lawn to overgrow in any direct sense (watering it, fertilizing it, helping the plants draw nourishment from the soil), but because I withheld an action, the lawn overgrows. It only makes sense to implicate me at all because of my capacity and authority to act. Neither my dog nor my neighbor can be said to have let my lawn overgrow. When clinicians claim they “let someone die,” it also implies they had the capacity and authority to act otherwise.3
But also note that the claim that I let my lawn overgrow implies something about my action, namely, that my inaction had an overgrown lawn in view. There’s nothing in “let” that helps us discern whether I intended or merely foresaw an overgrown lawn. It also tells us nothing about my competing priorities - perhaps I wanted to spend time with family instead. All “let” tells us is that I had the capacity and authority to act, but I didn’t, and I knew a particular outcome was possible because of this allowance. Likewise, for clinicians, “letting die” implies they have the capacity and authority to prevent death, but it says nothing about whether they intend or merely foresee that death will occur.
It does suggest that preventing death may not be the top priority in all circumstances. For example, more harm than good may come from mechanical ventilation for a dying person. This doesn’t suggest death should be intended though. We’ll explore this in greater detail in a moment.
This helps us recognize that clinicians should take responsibility for those things they allow. Samuel Sheffler argues that when someone allows something to happen, it isn’t as passive as it sounds. “It requires marshaling the full resources of one’s agency, including one’s capacities for deliberation, choice, and action. …if one sees oneself as subject to a standard of responsibility and holds oneself to that standard, then one sees oneself as responsible for exercising a kind of overall regulative control of one’s conduct, and the exercise of such control is itself a full-fledged expression of one’s agency.”
So far, we’ve seen that clinicians can have surrounding responsibility by virtue of their power over the circumstances in which a patient has fallen. They also have relevant capacities and authority such that if they don’t exercise those capacities and authority, they allow what they might otherwise be able to prevent (or delay). This allowing is a “full-fledged expression of one’s agency.” The whole picture here is that the clinician is actively involved, whether they’re allowing or doing, letting die or killing. Clinicians need to justify themselves, no matter what.
That doesn’t mean there’s no difference between allowing and doing but Sheffler is pretty pessimistic about the whole endeavor: “…I do not believe that there is any way of specifying the distinction between doing and allowing that could render the norms immune from controversy. No candidate specification of the distinction, however precisely drawn, and however firmly rooted in the theory of action or causation, could either produce consensus about how the balance should optimally be struck or allay the ambivalence people often feel in hard cases.”
Sheffler might be right, but that hasn’t stopped authors like those we’ve surveyed from attempting to build a bridge from the permissible to justify what was once impermissible. Killing certainly has death in view. The link between killing and death is necessary. By definition, killing entails death.
On the other hand, the link between allowing and death is not necessary. We can speak of all kinds of things clinicians allow: what kinds of food a patient might eat, whether a medication adverse effect is permitted in pursuit of treatment, whether to leave this or that issue undiscussed in a clinical encounter, and so on. For every action there are gradations of inaction that action implies; for every “yes” there is at least one (and probably many) related “no”s. What the link between allowing and death is doing is establishing that death should be the outcome by which the clinician’s action is judged. We consider something like “killing” and look around the clinical landscape: what is like killing? Ah, yes - other dilemmas in which death might be likely, even almost guaranteed, as a result of clinical decisions.
At the risk of understatement, death is an important outcome. But it isn’t an outcome that should control all decisions at all times. There are other examples where people prioritize other things above even their very lives - e.g., a soldier sacrificing themselves to save other people. When ethicists contrast killing and letting die, they imply either that death should be first among all considered outcomes, or that clinicians should have such control over their agency that death should always be controlled. There may be other meaningful outcomes, even outcomes that require clinicians to abstain from prolonging life. We’ve already discussed, when surveying duty, an interest in protecting the liberty of vulnerable patients.
What the claim of “letting die” is doing is drawing our attention to the outcome of death and asserting, “Hey, you’re responsible for that death. What’s your justification?” Clinicians should have a good justification for decisions that indirectly result in death, or for creating circumstances in which death is a possible outcome (i.e., weighing risk). But that doesn’t mean a clinician need shoulder that responsibility in exactly the same way as if they were killing their patient, because to directly kill the patient is to do something entirely different. It can only have one end in view.
So, that’s a long-winded way of saying, yes, clinicians indirectly cause a patient’s death by withholding or withdrawing LST. However, just like almost any other verb, there are justified and unjustified, right and wrong ways to allow something. My homeowner’s association may want to know why I’ve let my lawn overgrow, and will fine me or not based on my response.
Which brings us to our next big question.
Is there a threshold upon which “letting die” becomes “killing?”4
Both Asscher and Miller et al. argue that “letting die” is actually “killing” in certain circumstances. The implication here is if withdrawing or withholding LST is permissible, then killing by euthanasia should be permissible too. There’s nothing morally distinguishable between these actions.
“Kill” is a morally neutral verb that describes the action of ending life. To say you killed a tree, killed a canary, or killed a human doesn’t provide enough information for moral evaluation, even if you’re identified as the cause of the creature’s death. This is made clearer by acknowledging that we can claim things that aren’t moral agents can kill. Hurricanes, lava flows, and cyanide can all kill but without bearing moral status in themselves; they aren’t blameworthy in the same way humans are. Contrast this with a term like “murder” which is a morally and legally laden term implying wrong-doing. We don’t claim that hurricanes, lava flows, and cyanide commit murder.
To identify killing certainly raises a flag. The act of ending life is so serious it requires further justification. We’re attuned to ask more questions about killing than about most other actions. Likewise, as we’ve already explored, “let” or “allow” is also a morally neutral verb that can’t justify a moral evaluation on its own, even when we “let die.” From a purely causal standpoint, you could describe cases that are so similar so as to blur the concepts of “letting die” and “killing” into one. Deactivating a ventricular assist device (VAD) in a patient who isn’t otherwise dying can feel like this. This person would be alive an hour from now but for the clinician turning off the VAD.
So if it’s permissible to turn off a VAD, which feels very much like killing, why wouldn’t it also be permissible to euthanize a person who doesn’t have the expeditious option of VAD deactivation? This is where I worry that relying on causal rationale alone can make us stumble. Some clinicians may be tempted to conclude that because some forms of letting die may qualify as killing, and because they’re justified to let die in some circumstances, they would also be justified to kill in other ways (e.g., euthanasia). I don’t think that follows. The devil’s in the details.
Sheffler warns against this, claiming it’s an attempt to “optimize causal instrumentality” (philosophers and their $10 words, right?). That is, if allowing and doing are the same, and allowing is commendable, you should always do instead. If it’s good, in a certain circumstance, to let someone die, and letting die is the same as euthanizing, you should euthanize instead. I’ve painted the picture rather starkly there, skipping over a few steps like consent, but I want to show Sheffler’s point about what it means to optimize causal instrumentality. The better path is “doing” if you’re trying to optimize an outcome that results from either allowing or doing. With this rationale, euthanasia would always be preferable to letting die. It may even be the obligatory option between the two.5
You might detect the faint scent of something we’ve explored before: efficiency. The efficient mindset about which I’ve written before might contribute in pushing clinicians and patients toward euthanizing if it’s just the same as letting die. Efficiency crowds out other values. So, if some determine that letting die and killing are the same so as to establish that killing via euthanasia is permissible, the evaluations of efficiency push them to never let die when they could otherwise euthanize. This ironically and paradoxically reinforces the differentiation between the two ideas but in the opposite direction of what we’ve traditionally maintained!
One additional challenge facing the project to erase the distinction between killing and letting die: it would re-label innumerable clinical decisions as “killing” which we probably appreciate are most certainly not killing. Consider a patient with advanced cancer who may experience little, but not zero, prolonging of their life with additional cancer-directed therapy. Whether their oncologist withholds that additional therapy or starts it and then withdraws it, can we say the oncologist killed their patient?
“No,” you might object. “This patient will die whether or not they receive the therapy. The therapy isn’t life-sustaining, only death-delaying.”
If a patient dies because of an action by a clinician, whether they die tomorrow or a year from now, this is killing using the causative rationale we’ve been interrogating. The patient would be alive (or alive for longer) but for the clinician’s decision to withhold or withdraw a therapy.
Maybe you can see how this shatters into absurdity. Any clinical decision that is not in direct pursuit of longevity can be implicated in a causal chain of events that may result in someone’s death and be labeled as killing. That antihypertensive you discontinued five years before a patient died of a myocardial infarction? Killing. That insulin you opted not to start seven years before a patient developed lethal renal failure? Killing. The oncologist who stops further cancer-directed therapy because their patient is too weak? Killing. This is just silliness.
“But the act and the outcome are so far removed from one another, with so many intervening events - it’s hard to call that killing.”
Is it? If Wayne hired someone to hire someone to hire someone to kill someone, would he be less responsible for the assassination? In cases of obvious murder, it doesn’t matter how labyrinthine the chain of causality. Why would it matter here?
Thankfully, it doesn’t, because there’s more that matters than mere causation. We shouldn’t call these actions killing, even if they’re somehow related in the causal chain of events that result in death. Death is not a necessary result of “allowing” in the same way it is for “killing,” the logic of efficiency would push us toward more and more euthanasia, and relabeling these allowances as killing results in definitional absurdity.
I don’t think it’s helpful to call any form of “letting die” killing, even if the strictest, barest sense of the terms, there may be instances of “letting die” that could qualify as “killing.” If an instance of “letting die” was unjustified, it may rise to level of murder or negligent homicide, which are clearer terms to describe the action because of the additional justification they require.
Help set the tone for future issues of Notes from a Family Meeting! An example of what I’m talking about can be found at the end of this issue.
Although the word “responsibility” can also mean “duty,” in this case I’m referring to it in a sense of culpability or ownership of an action.
The rabbit trail examining the difference, if there is one, between “allowing” and “doing” leads into a deep, dark forest where sharper folks than I have failed to chart a reliable course. Readers with more philosophical experience will see that, in an attempt to address my narrow question, I’m probably hobbled by a lack of familiarity with the broader conversations about this distinction between “allowing” and “doing.” Alas. We’ll avoid any discussion of trolleys.
We’ll cover authority in Part 4. However, just a note: in this case, the clinician has the formal authority to act otherwise (they are authorized by the state to intervene on the human body in certain ways) but they may lack the actual authority to act otherwise (this individual patient has either not given or they’ve withdrawn consent for LST).
Let’s acknowledge the terrain here: what we’re debating is the definition of terms, something notoriously difficult. We’re sifting through presuppositions and intuitions. So, maybe what I’m about to write will be unsatisfying to you. It might be unsatisfying to me a year from now!
To the extent any pro-euthanasia folks would disagree with this, it would be to that extent we both agree other lines of inquiry are required beyond mere causation to properly morally evaluate an action (as I’m doing in this series).