Friday, 29 February 2008

you can die from that?

The World Health Organization is an important agency with important functions. One of those functions is compiling lots of statistics from diverse sources into coherent, standardized formats. A friend of mine is using some of this data to complete a study and he made the mistake of showing me the file he was using--the WHO Mortality Database, a classification system for cause of death, with categories corresponding to dysentery or heart attacks. All of this is, quite literally, deadly serious. And that's exactly why none of what follows is remotely funny.

One might imagine that compiling a list of all the possible ways to die is grisly and horrifying work--I'm sure it is. In the four-character classification scheme, a letter indicates the basic category: A for viruses, B for bacteria, C for cancers, etc. There are dozens of pages of these, and it is disturbing to see all of the different places you can get cancer and die from it. But the classification job must also be surprising, given that there is a code for nightmares (F515). I don't mean that there is a code that can be stretched in such a way that it seems like nightmares might fit into it; rather, the complete description of that category is: "nightmares." I swear to god, if I die from the nightmares I have about dying from nightmares, I'm going to slap someone in heaven. (Um, plus, how exactly do we diagnose that one?) (image from the inimitable--and spookily timely--xkcd)

Next time you have a peek in your spam folder for a laugh at the Viagra ads, don't be so quick to be dismissive--there's a code for dying from "lack or loss of sexual desire" (F520).

If these codes seem absurd, consider F811, specific spelling disorder. Now, I understand F812, specific disorder of arithmetical skills; it has been stressed throughout my schooling precisely the sort of mathematical skills I will need to survive on a desert island, but spelling? Is this category reserved for LOST castmembers who spell out HLEP! on the beach?

Others, more or less absurd (some in name only):
F986: Cluttering. (Parents may want to cite this one for their messy kids)
J676: Maple-bark-stripper's lung
L671: Variations in hair color
L812: Freckles
R461: Bizarre personal appearance
R462: Strange and inexplicable behavior
V954: Spacecraft accident injuring occupant

It might be argued that it's easy for me to sit back, skim through the three-hundred page finding aid, and poke fun. Or that I'm making judgments about serious disorders that range from crude to offensive. To this I say: guilty as charged. Read on.

Perhaps the most amusing--er, "perplexing" series are the Vs and Ws, which seem mostly to correspond to what we might commonly call "accidents"--things like falls and vehicular collisions. One part of what's fascinating about this group is the level of specificity allowed by the third character: there are separate categories for falls in streetcars, falls from streetcars, and falls involving moving streetcars versus those involving stationary ones. Even this does not exhaust the depth of description--the fourth character is a location code. Take, for example, W067: Fall involving bed, farm. When you understand the coding scheme, you see that the last digit is a standard location code (4 for street or highway, 7 for farm), and it turns out improbable death codes are included for the sake of completeness. That's not to say that people never fall to their deaths from bed on a farm--but it's less likely they do so in a street or highway (other than mattress-racing fatalities, of course). Not that I checked. Even understanding the system, however, the juxtapositions are intriguing:

W527: Crushed, pushed or stepped on by crowd or human stampede, farm.
W654: Drowning and submersion while in bath-tub, street and highway.
W754: Accidental suffocation and strangulation in bed, street and highway.

And then there are the X-factors. Call in Sculley and Mulder for these ones; they're definitely the odd-balls: X05: Exposure to ignition or melting of nightwear (ouch!). Or take X52: Prolonged stay in weightless environment. None of the location codes work terribly well here, since I'm guessing this only happens in space.

It's very heartening to know that Y61, foreign object accidentally left in body during surgical and medical care, has about a billion subcategories. Did they leave in in there during endoscopic examination? Perhaps during dialysis? Or during the surgery itself?

One of the more disturbing codes is Y881: Sequelae of misadventures to patients during surgical and medical procedures. Misadventures? Really?

Scott Adams, creator of Dilbert, wants to get killed by frozen poop that's fallen from an airplane. (see here for associated puns). If I were the WHO, I would invent a new category for this.

On a more serious (or at least less crass) note, the implementation of a database for deaths is philosophically interesting for what it reveals about explanations of death. One of the canonical views in the philosophy of explanation is known as the pragmatic view, which basically says that our criteria for accepting explanations change depending on the circumstances (see van Fraassen's The Scientific Image for a more nuanced take). The pragmatic view has the virtue that it comports well with everyday explanations: if the question is, "Why did Bob die?" a coroner may explain that it was the "multiple blunt impacts" that did him in. A cop might explain that Bob "lost control of his motor vehicle while traveling in adverse conditions." A city planner might explain that Bob was killed because of poor roadwork design. All of these are acceptable answers, given the appropriate context. It is the job of the questioner to ask a question that provides the right context, either within the question itself ("What is the medical cause of death?") or in context-bearing ancillary statements ("Bob has driven this route every day for years, why did he die this time?") Often, the correct context is provided by setting up a contrast class, as in the last example (this time, instead of all the other times). Other contrast class-defining phrases include "why Bob in particular," or "why this location in particular?"

What strikes me about the Death List is that it is sometimes ambiguous (or, more likely, I don't know how to use it correctly). Death isn't a particularly straightforward event, as any episode of CSI demonstrates. Many factors contribute to death. A car crash is a cause of death, but so are the contusions sustained in the crash, or blood loss afterward. A heart-attack just before the crash might be the cause. But what if heart-disease caused the heart attack? It would be more than a little perverse to suggest that a diet rich in red meat caused an automobile crash, but the causal chain is (seemingly) there. At some point, though, tracing the causal chain backward just isn't what is being asked for.

There is a very interesting transition in the Death List. The list begins with the ABCs: infections, parasites, or cancers. D-N track diseases and disorders of various organs or systems. Then there are deaths due to developmental problems: O is for obstetrics, P for perinatal, Q for congenital malformations and deformities. At V we make the transition to "external causes of morbidity and mortality." In particular, V is for transport accidents; W for falls, drownings, and submersions; X for smoke, poisoning, or intentional self-harm; and Y for assaults. Between Q and V is a vast, four-letter chasm separating internal from external causes of death. In fact, there are no S, T, or U deaths. One letter stands in the mysterious gap between internal and external causes of death: R. R is the category for abnormal but inconclusive findings, including R960, Instantaneous death. This, it would seem, is neither a cause nor an explanation.


4ll4n0 said...

So it would seem that they compile there death list by allowing the possibility that every medical condition (including symptoms rather than diseases) could be fatal.

Just to feed your fear of nightmares I'm under the impression that complete lack of sleep can eventually be fatal. Therefore incessant nightmares that prevented sleep could in principal be fatal (if the nightmares could actually prevent sleep altogether).

I think one other key concept from philosophical debates about causation is the idea of overdetermination, when there are multiple seperate causes that would lead to the same outcome. Famous accounts of Rasputin's death suggest his end was heavily overdetermined.

I think in light of your concern about dieing from nightmares about dieing of nightmares we need a category for death by reflexivity (when an activity directs back on itself).

Keynyn said...

Say, are there categories for fork-stabbing injuries, or exsanguination caused by chocolate pudding, feather pillows, or aluminum foil?

I haven't died... yet... but I have injured myself with all of these things. :P

Isaac said...

4ll4n0: It does seem more like a list of symptoms, doesn't it? I didn't read the documentation, because that would likely make sense of everything and leave me with nothing to joke about, but it's entirely possible that the death reports are in fact lists of all conditions recorded by coroners anywhere in the world--and that could easily include things like freckles.

But nightmares? If I woman dies while having a nightmare, how does anyone else know about it?

You're right to bring up overdetermination, which I hinted at but neglected to mention explicitly.

As to reflexivity, let's call it recursive death. A recurse is like a curse, but again.

Isaac said...

keynyn: there is a whole category of death by self-inflicted injuries (and even a distinction between accidents and purposive incidents).

I don't remember a fork category, but there's one for pointy objects, I think.

I think we've all injured ourselves with feather pillows, but how did chocolate pudding make you bleed?

Or, ew... maybe I don't want to know. ;-)

Keynyn said...

Well, I'm being slightly misleading - it wasn't the chocolate pudding itself, but the plastic container it came in. The sharp-edged plastic container....