"Lives" are the Wrong Measure

When thinking about triage situations, it's common for people to assume that saving lives (as many of them as possible) should be our moral goal.  But this is wrong, for the straightforward reason that some deaths are vastly more tragic than others.

It's worth bearing in mind that lives can't be saved, but only extended.  So "saving lives" is not even a coherent goal.  You can aim to maximize the number of lives extended (for any period whatsoever), but we can now see that this is akin to trying to blindly maximize the number of patients treated.  By ignoring how much the patients gain from different treatments, you're clearly neglecting what actually matters -- the underlying health benefits that are the whole purpose of medical interventions in the first place.  Willfully blinding yourself to the magnitudes of different interests will lead to predictable injustice: you might foolishly prioritize two patients' papercuts over another's spreading gangrene, for example.  Raw numbers helped is not the important thing.  Moreover, this principle is as true of life-extending treatments as it is of any other.  (This is most obvious if you imagine a treatment that will extend life by mere minutes.)  I don't see how any remotely sensible person could possibly deny this.



This has important practical implications. For example, reasons of utility and justice alike should lead us to view an extra five decades for one 20-year-old as morally more important to secure than an extra five years each for as many as ten 80-year-olds.  This is a drastic counterexample to the idea that we should simply prioritize "saving the most lives".  The indiscriminate number-chasers neglect how much more is at stake for the 20 year old. Their entire adult lifetime should not so easily be sacrificed for the sake of bonus years to others who have already lived more than most people ever will.

Shifting focus to (quality-adjusted) life-years lost or gained is a better approximation, but still not exactly right. (Amongst other problems, it would mistakenly lead one to view the above choice situation as a moral tie.  I think that's still mistaken, but at least a major improvement over the raw life-numbers view.)

Weyma Lübbe objects that maximizing life-years entails that a sixty-year old with a 70% chance of 20 years more survival only if treated (and zero otherwise) "should give way to a 20 year old" for whom treatment merely enhances their chances of an extra 60 years of survival from 70% up to 100%.  She continues [Google translated]:

The deferred patients and their relatives would be expected to give up considerable chances of survival in favor of augmenting the chances of survival of people who already have substantial chances of survival without this solidarity (or what should one call it?). Why should they do that? And how does it happen that something like this can be presented as a result of "ethical reflection"?

They should do it because the 30% risk of death to the twenty-year old is the graver harm.  Perhaps this isn't immediately obvious because our brains are bad at probabilistic reasoning: the move away from fixed-size increments to vague talk of "substantial chances" is a red flag here.  A 30 percentage-point increase in survival chance to a particular individual has equal value regardless of their baseline survival rate (between 0 - 70). Call this the Fixed Value of Probability (FVP) principle.

FVP implies that the above-described choice scenario is morally equivalent to one in which the 20-year old has a baseline survival rate of only 45%.  In that case, the extra 30% boost from treatment brings them up to a 75% chance of survival.  The treatment thus shifts their prognosis from "likely death" to "likely survival" -- a big difference, according to commonsense thinking!  Such a big difference, for a young person with so much to lose from an early death, could reasonably be prioritized over treatment that boosts the 60-year-old's chances of survival from (say) 5% to 75%, given that her loss is so much less. (Still a huge loss, of course, but I'm talking comparatively here: the loss to the 20-y/o is beyond huge.)

So the intuitive reasoning Lübbe appeals to depends upon implicitly rejecting the FVP principle, and treating fixed-size probability increments as having variable value depending upon the starting probabilities.  But that's clearly just a mistake: the value of an extra fixed-percentage-point chance of survival should not vary in this way.  (It should vary depending upon other factors, such as the expected duration of your subsequent survival.  But it should not vary depending upon your starting chance of surviving.  You don't need to know where you currently lie, within the range of 0 - 70% chance of survival, in order to know how important or good for you it would be to increase your survival chance by an extra 30 percentage points!)  Our intuitions overweight the differences between 0-1%, 49 - 51%, and 99 - 100%, because we use intuitive categories that correspond to "no chance", "un/likely", and "certain", but it's the underlying spectrum that really matters.

If we're to respond appropriately to real-life emergencies, it's vital to understand the ways that these cognitive biases and demonstrably inapt concepts can distort our moral understanding.

[See also my suddenly highly-relevant 2016 Bioethics paper, "Against 'Saving Lives': Equal Concern and Differential Impact."]