How do we judge worse than the worst?
The Italian film industry are quite weird. In particular, they are curiously inventive when they translate the original title of foreign movies. Here are some of my favourite examples.
|The eternal sunshine of the spotless mind||Se mi lasci ti cancello||If you leave me I'll erase you|
|The break up||Ti odio, ti lascio ti...||I hate you, I leave you, I...|
|The Cosby show||I Robinson||The Robinsons|
|The Shawshank redemption||Le ali della libertà||Freedom's wings|
But here’s another nice example: when they came to visit us a few months ago, Ale and Teresa mentioned that they have watched this French movie, which apparently had been a hit in Italy (I don’t think it was released in the UK). The original title is Intouchables, (obvious English translation: untouchables), but the Italians have thought better to re-title it “Quasi Amici” (literally: “Almost Friends”). They said it was really good, albeit a bit sad (and right they were!), so we watched it. Then, before I went to San Diego, I was looking for a book to bring with me and Marta suggested this (I thought it’s a good book, although a bit girly).
Although not identical, the film and the book have a lot in common since they both tell the story of a very rich man who has an accident and becomes a quadraplegic, and his carer (a bloke in the movie, and a she-doesn’t-know-it-yet-but-she’s-really-hot girl in the book). I thought both dealt quite tastefully with this delicate issue and the struggle of the protagonists. As if it wasn’t enough, I was reading the newspapers the other day and found this article, about a “locked-in syndrome” victim who has been denied the right to commit assisted suicide by an English court.
So all this got me thinking a bit with my health economics hat on (you may think that this is not what you should typically thinking of, while you’re away celebrating your birthday; and you may probably be right! But there you go…). I haven’t really thought hard enough about all this, so I don’t even know for myself whether I’m pro choice (but I think I might just be), or not. But this opens a potentially huge loop-hole in the very foundations of health economic evaluation.
Typically, NICE (and consequently nearly everybody doing cost-effectiveness analysis) evaluates competing health interventions in terms of the “cost-per-QALY”, a QALY being a measure of “quality-adjusted life years”, supposed to account not only for survival (due to a given intervention), but also for quality of life in the extra years granted by the intervention. The use of this clinical outcome effectively makes different interventions (or even diseases) comparable in economic terms.
I heard several times Karl Claxton, arguably one of the brightest minds we have in health economics, speak about QALYs. His point is that: yes, they’re not perfect; but they’re probably the best we’ve got. And yes: it is difficult to compare interventions and diseases that are so different; but we still need to do it, if we are to put a monetary value on health.
More importantly, the idea behind QALYs is that it is somehow possible to give a utility score to every possible health status. Typically, “perfect health” is valued 1, while the worst possible status (usually “death”) is valued 0. Any other status is given a value in between these two extremes. But what if someone experiences a health status that is even worse than death? Then, leaving all moral considerations aside, assisted suicide becomes cost-effective because it saves money (given that patients with these extreme conditions are generally very costly just to be kept alive) and improves the patient’s quality of life.
This is an old theme in health economics; the estimations for the QALYs to be associated with the health status have been based on a sample of the UK general population. In particular, these are healthy people, who are only thinking hypothetically about diseases or conditions they don’t even have. Of course, to ask people with the disease how they feel about it may over-estimate the gravity and thus it is certainly not ideal either.
Perhaps it would be good to have even more complex data collection mechanisms than the one currently used (some ideas here; I also discuss them in chapter 1 of the book). For example, to sample both healthy people _and _patients suffering from some condition or disease. May be hierarchical models could be applied to account for potentially different (latent) data generating process, in terms of their preferences. And may be procedures such as post-stratification, a topic that Andrew Gelman discusses a lot (for example, here; or in his book), could be used to re-proportion the observed preferences and opinions.