In this post, Andrew McGee (Queensland University of Technology) and Drew Carter (University of Adelaide) discuss their recent article in Journal of Applied Philosophy on the moral difference between withdrawing and withholding medical interventions.
Some health ethics writers and clinical guidelines claim that withdrawing and withholding medical treatment are morally equivalent: if one is permissible or impermissible, so too the other.
Call this view Equivalence. It is heir of a related view that has held sway in ethical and legal debate for decades, in support of the withdrawal of treatment that is no longer beneficial. The thinking was that if treatment no longer benefits a patient, then whether it is withheld or withdrawn does not matter – so there is no morally relevant difference between the two.
Equivalence goes beyond this. It applies to beneficial treatment, where two patients compete for one resource. The reasoning is: To save as many lives as possible, we would have no qualms about withholding a beneficial treatment from one person to give it to another who can benefit more. We should therefore have no qualms about withdrawing it either. In a recent article, we argue that Equivalence is false.
Here are two scenarios illustrating the dilemma from the COVID-19 crisis. In both scenarios, A and B both need the last remaining ventilator to survive:
Scenario 1: A arrives in the ICU and the doctors estimate that A has a 20-50 per cent chance of surviving with ventilation, so they decide to connect A to the ventilator. But before connecting A, B arrives. Doctors estimate that B has a 50-80 per cent chance of surviving with ventilation. Doctors must decide who to connect. Other things are equal.
Scenario 2: A arrives in the ICU and the doctors estimate that A has a 20-50 per cent chance of surviving with ventilation, so they connect A to the ventilator. B then arrives, and the doctors estimate that B has a 50-80 per cent chance of surviving with ventilation. Doctors must decide whether to withdraw from A to give the ventilator to B, or withhold the ventilator from B to keep giving it to A. Other things are equal.
We could compare the act we choose as most justified out of the two options in Scenario 1 with the act we choose out of the two options in Scenario 2. Suppose in Scenario 1 we choose to withhold from A and in Scenario 2 we choose to withdraw from A. We might then conclude this is evidence for Equivalence. This seems rational because all other things are equal between the two scenarios.
However, there is a problem. We also need to make other things equal between the two options within each Scenario (withholding from A or B in Scenario 1, withdrawing from A or withholding from B in Scenario 2). In each scenario, B can benefit more than A. This greater capacity to benefit may have led us to conclude that it is permissible to withhold from A in Scenario 1 and withdraw from A in Scenario 2. How do we know that this hasn’t overridden any moral difference between withholding from A and withdrawing from A? Once capacities to benefit are equal, it seems that it is impermissible rather than permissible to withhold from A in Scenario 1 and to withdraw from A in Scenario 2.
Now, it might seem that this conclusion does not affect Equivalence: we have switched our verdicts from permissible to impermissible in both scenarios, so no difference between withholding and withdrawing has emerged now that we have equalised capacities to benefit.
But why have we changed our verdicts now that other things are made equal? Because first-come-first-served has moral relevance.
And it is here that a moral difference between withholding and withdrawing emerges, despite reaching the same verdict of impermissibility in Scenarios 1 and 2. We cannot devise a scenario that involves the option of withdrawing from A or withholding from B, where some prior allocation decision favouring A has not already been made and put into effect.
Withdrawing from A always presupposes a prior allocation to A, because we can only withdraw what has been provided. To withdraw beneficial treatment is always to revisit an allocation already made. Whilst it is possible that the original decision was not made fairly, typically it will have been. This explains a greater reluctance to withdraw treatment once it has started.
We can also illustrate this with a third scenario. Again capacities to benefit are equalised:
Scenario 3: A and B both arrive in the ICU at the same time. Doctors estimate that both A and B have a 20-50 per cent chance of surviving with ventilation. Doctors must decide who to connect. Other things are equal.
In Scenario 3, A and B arrive at the ICU at the same time. It is not possible to have A and B arrive at the ICU at the same time, with A already being connected to the last ventilator, giving a withdrawal option. But we can have A and B arrive at the same time, perform a coin flip which A wins, connect A, and then consider whether we should withdraw from A and connect B. This, however, would, once again, involve revisiting a prior, fair allocation. And it is obvious that, when other things are equal, including capacities to benefit, there is no reason to revisit the decision. So whether A and B arrive at the same time or at different times, there will always be a prior decision to revisit when considering whether to withdraw treatment.
A decision to withhold treatment from A may also take place after a fair allocation decision, but this is not typically the case because dilemmas about withholding from A typically concern making the decision to allocate the ventilator in the first place. The option of withholding from B to keep giving to A does not presuppose a prior decision favouring B, but favouring A.
This helps us understand why medical practitioners may be more reluctant to withdraw treatment than withhold it. In the COVID-19 cases where the percentage capacities to benefit are different, doctors may be justified in overcoming this reluctance, since saving more lives might override first-come-first-served. But this should not be confused with the different claim of Equivalence.