Tuesday, 4 January 2011

Yes, Virginia, epidemiology is really hard

I had been trying to get away from these road safety rants of mine, I really had.

But this was too interesting to pass up: An interesting, amusing (as in "funny hmmm....") and self-aggrandising article from the Courier Mail: Fast-aid cuts road toll to lowest since 1952 (January 04).

What does? I thought it was speed cameras! Also: Where, dear deity, did that hyphen come from? "Fast-aid" isn't mentioned elsewhere in the article so it's not a brand or other name, and clearly it's not a compound adjective - "cuts" is a verb. It's bloody stupid, is what it is.

The point of that article, as expressed concisely and almost well in the lead, is:

BETTER, faster ambulance crews helped Queensland achieve its lowest road toll last year since accurate records began in 1952.

Again: I thought it was speed cameras!

In the midst of a health system SNAFU (it's probably so long-term it doesn't deserve to be called a crisis anymore) where staff aren't paid and emergency beds have waiting lists, hearing that any aspect of the ambulance system has been improved is fantastic.

What interest me here, however, is that it is a beautiful, nay, elegant illustration of the fact that reality is really complicated and, therefore, epidemiology is quite hard.

There are two factors affecting the road toll: How many people are injured badly enough to be fatal and; how many are given adequate treatment. The first factor can be broken down into: How many people are involved in potentially fatal accidents and; how well they are protected by the car and fate.

The road toll is therefore affected by road and traffic conditions, the judgement of drivers, the road-holding and handling of their cars, the crash protection and passive and active safety features of those cars, and the response times and skill of all arms of the emergency response services.

This is why I keep describing the road-safety debate as puerile, shallow and all but a waste of time.

The point of this article is that: The road toll is lower because more people are being prevented from dying.

Yes, there are numerous other factors, short- and long-term. But I would really, really love to see the figures reported as: total accidents/potentially fatal accidents/fatal accidents. Give us some perspective.

There is also this:

In 2010, there were 5.5 road deaths per 100,000 Queenslanders, which met a target to virtually halve the rate from a decade ago.

Note halve. There are more many more people using the same road network, meaning higher traffic densities, meaning greater risk, and yet per capita roughly half as many people are dying. If you ever look at the annual traffic toll and observe the sameness of the figures from year to year, ponder what that means in terms of a growing population. It means that it has never been safer to be driving. Well, not since cars were limited to walking pace and they attracted gawkers wherever they went.

Well, they're using a historical low for comparison purposes, which will make any statistician beat their head against their desk, but you see the point.

Whether it's due to stick approaches like cameras, or better driver training (although I have my doubts) or the simple fact that cars are improving in capabilities and tank-like levels of occupant protection out of sight (and bikes get the occasional nice feature like ABS or traction control) while the roads and speed limits are staying the same, we have seen real death tolls plummet.

I'm not sure which is worse: The publicly portrayed understanding of causality or the public portrayed understanding of mathematics.

No comments:

Search This Blog