When you go to the hospital, we rarely have the resources to treat everyone as promptly and thoroughly as we’d like. So we have to make choices about whose care to prioritize.
This makes sense. If grandma’s 110 and has a really low chance of returning to even a meager quality of life, maybe we should put the otherwise healthy 40-year-old on the last ventilator instead.
There are lots of ways hospitals try to make these judgments more “““objectively.””” (Get ready for lots of scare quotes.) One standard approach is the QALY framework, which stands for Quality-Adjusted Life Years.
There are lots of ways to calculate QALYs, but here’s one illustrative method:
On the left side of the scale are 10 years of Perfect Health.
On the right side are X years lived Blind.
How big does X have to be for you to consider the scale balanced?
12? 15? 30?
Say after surveying a bunch of people the average response is 20.
So 20 years Blind = 10 years in Perfect Health.
Therefore, each year lived Blind is only worth 0.5 Quality-Adjusted Life Years.
Great! Now that we have a stable conversion rate we can start doing a bunch of math. And wait a minute…it turns out we should systematically shift healthcare resources away from the Blind because their life years were way lower quality to begin with!
Oops! By trying to calculate our way through every moral dilemma we’ve outsourced our all decision-making to a bizarro formula built on super ableist preconceptions. It turns out that just averaging how bad folks think Blindness would be compared to “““Perfect Health””” is not a great method for figuring out how to prioritize care!
More technical problems: Blind people rate their own lives as going a hell of a lot better than half-rate, thank you very much! So what are you gonna do, only survey the Blind about their own condition? And how are you going to combine different conditions? If a year Blind is worth 0.5 and a year with IBS is worth 0.8, is a year with both only worth 0.5 x 0.8 = 0.4 years of Perfect Health? Are we just gonna assume they’re perfectly independent?
You might think I’m joking, but QALYs are an incredibly influential metric in real-life decision-making because they’re “““neutral””” and mathematically tractable. When anyone’s upset, you can always point em back to the spreadsheet.
But no matter how “““sophisticated””” you make a system like this it’s still built on really shaky, problematic grounds. And that’s what my first publication argued. It was a joint 3-page effort to shake the QALY framework to its core.
How cute.
Because it was an Open Peer Commentary, we had stunningly few words to point all of this out to three authors who had just published a paper trying to patch up the QALY framework.
So here was our overambitious argument:
Shelly Kagan draws a subtle distinction between
what’s good for you (your “well-being”), and
what’s good for your life (which he calls your “quality of life”)
On reflection, these aren’t quite the same. Taylor Swift’s generational success seems great when we consider how her life is going, where her projects keep bringing accolades and audiences. She brings so much joy to the world that her impact and legacy are just tremendous.
But is all that success good for her? Or is it enough to make anyone a bit isolated or detached or out-of-touch or self-obsessed or otherwise unwell? See, I told you this is a slightly different question! (And I can’t wait to receive your email that Taylor Swift has handled fame perfectly!)
Anyway, here’s Kagan:
a person's life seems to be broader and more encompassing than the person himself; it includes more within it…Accordingly, certain changes might constitute changes in the quality of a person's life without constituting changes in the person's level of well-being.”
If I ask what’s good for you, I organize my inquiry around you. What’s your mental and physical health like right now? How much cogency and agency and control do you exercise within your own life? How might you be doing if the surgery goes well (or not so well)?
But if I ask what’s good for your life, I organize my inquiry around something much bigger reflecting your choices and circumstances in the world. You’re still the star of the show, but I’m also going—What would be best for your projects? Your impact? Your legacy?
Suddenly we’re introducing way more stuff into our analysis. And we were hoping to boil all of this down to a single medically relevant number?
Quality of life is a slippery thing. If we’re really gonna chase it by maximizing QALYs, maybe we should sacrifice everyone else in the hospital to save our newest patient, Taylor Swift. Why? Because if we start handing out surveys, all the Swifties might say that a year lived without Taylor Swift is only worth 0.9 years of Perfect Health. Their lives are bound up with hers. I knew you were trouble when you walked in!
How are you gonna rule IBS in and Taylor out as medically relevant?
You might go, Ricky this is obviously cheating! Blindness and IBS are diagnosable maladies. And you’re only allowed to adjust life years in response to your patients’ diagnosable maladies.
But why’s that?
Remember, your basic unit of analysis is how many quality-adjusted life years your decision spits out. Patients only appear in your analysis as containers or locations where QALYs happen to be. So why restrict your analysis to your patients’ QALYs? Why can’t you go looking in the other containers?
And why only quality-adjust for diagnosable maladies? Once we’ve resorted to adjusting life years, why not adjust for everything? Why not survey folks about whether they’d like to live a world without Taylor Swift? If her work turns out to be worth billions of QALYs, wouldn’t that be medically relevant when she turns up needing a ventilator?
There are so many really deep problems here, and I didn’t know enough back then to avoid rustling too many jimmies all at once. I can also see now that I invited misreadings by bringing Kagan’s peculiar technical terms into this discussion in the first place.
But the basic unit of analysis is clear: We want doctors to treat patients, not lives.
And they ought to treat patients as moral equals, rather than focusing on QALYs as mathematically interchangeable! That doesn’t mean you have to treat them equally. But treating them as equals requires giving them the same sort of moral consideration. (Look ma, I’m heading off a boring misreading right now!)
So when tough choices have to be made, we want doctors to consider what’s best for their patients, NOT what would maximize QALYs. This amounts to a subtle but foundational critique that the QALY framework measures the wrong thing—a pretty classic case of value capture.
There’s a deep instability here that’s hard to fully flesh out in a three-page commentary, and we didn’t quite nail it. But boy did we try:
If this is right, then only some aspects of quality of life relate to personal well-being. This is why there is no medical need to treat the businessman just because his wife is secretly cheating on him. (Although if he finds out, he might need the medical attention of a therapist or cardiologist for his heart.) … To avoid diluting concerns central to treating patients, we should aim to treat them equally, not their lives, and certainly not their quality-adjusted life years.
In their response to us, the original authors mostly talked around our paper and when they did cite it once, appeared to misread us. (Again, I’ll take the blame for inviting misreadings.) In the last sentence before their conclusion, they went, okay maybe there are also “non-welfare values” like “the importance of recognizing patients…as persons worthy of dignified lives” that are worth considering, too. No big, throw that into the QALY soup if you like.
But treating patients as moral equals isn’t just an extra value we could simply tack onto the QALY framework. (As Kagan wonders, how are we supposed to weigh what’s good for me and what’s good for my life against one another?!)
Instead, treating patients as moral equals should structure how we reason about their good. In medical contexts, we don’t just care about a single quantity like QALYs. Even in the case of one patient, we care about a ton of different qualities (pain, autonomy, long-term prognosis...) that just aren’t commensurable. You have to judge individual patients’ good much more richly than that. So when we have to prioritize between patients, of course you can’t boil everything down into a single formula that loses sight of individuals altogether!
We failed to make this point as clearly as we could have. But I’ve learned to deal with misreadings much more proactively since then.
My next paper—the one on rioting—was rejected with an editor’s note that the position I was arguing for (political rioting might be justifiable without possibly leading to policy changes) was controversial and needed extra attention and argument.
But I had pretty good arguments already! So I just added one new paragraph that started, “Here is a defense of this controversial position.” and brought all the points I’d been making together in one place.
And the next place accepted the paper.
Causation, correlation, who can say? But you should be really clear about what you’re doing that’s unique and sell that.
You have to constantly signpost, or show your reader what you’re doing:
Here’s my distinctive contribution.
Here’s where I disagree.
Here’s something I’m definitely not saying!
And it still won’t always work!
You will be misread by both opponents and allies.
You will be misread by your referees (who may count among your closest readers).
You will be misread even when readers come in good faith, just because philosophy is so hard.
After all, it’s impossible to be perfectly clear. Language doesn’t work that way. So our arguments have to embed assumptions all the time about where we agree and what remains to be shown. Hopefully we mean the same thing by “equality,” but I won’t really know until I see how you misread me this time!
You can’t avoid being misread, and in my experience, trying to preempt misreadings by becoming hyper-clear and analytically precise usually just makes things worse.
But you can try to get others to misread you in more interesting ways, to use each round of feedback to cut off the worst or least interesting misreadings.
(That’s a lot of what this blog is for! Thanks for your weekly texts and emails.)
A misreading is an invitation to say things again more clearly.
The hope is that by highlighting your own distinctive contributions more and more, you can move things further along, so the misreadings that you do encounter are at least conversant with the basic moves you’re making.
And then it’s easier to pull the thread of the argument just a bit further and see just how much it unravels the latest misreading.
So how’d you misread me this time? =] Would love to hear it!