Friday, February 22, 2008

Does Latent Toxoplasmosis really increase risk of traffic accidents? The BIG Picture.

Road Deaths:
Population:


Do you notice a difference between the maps? Probably not considering there hardly is any discrepancy between the two. In the upper map, found on worldmapper.org , territory size shows the proportion of all road traffic accident deaths worldwide that occurred there. In the second lower map (which looks strikingly similar to the upper one to me), the territory size shows the proportion of the world population that lives there. If latent toxoplasmosis really increases the risk of traffic accidents, wouldn't we expect the countries with higher prevalence of this infection to have more road deaths? For example, rather than remaining relatively consistent with the proportion of the population that lives there, wouldn't we expect to see a higher amount of road deaths in France, Germany, the Netherlands, and Brazil where prevalence of toxoplasmosis infection exceeds 70 percent of the total population?







2 comments:

John Latto said...

Interesting. Visually striking though these maps are I'm not sure they are the best way to look at this data. A regression of traffic accident rate per thousand people against toxoplasmosis infection rate might be clearer. Lots of other things affect traffic accident rate so no-one would expect a huge effect - if any effect could be detected, even a small one that would be fascianting. Perhaps more sensitive would be to look at the US rate by state if the Toxoplasmosis prevalence data is available for all states. If anyone gets around to doing this before I do then please post what you find!

John Latto said...

I did a quick analysis of this and, as usual, learned that nothing is quick or easy. The way that traffic fatality rates are reported varies between countries - most report fatality rates per thousand population but a few, notably the US, insist on reporting them as per vehicle mile traveled. Since people travel longer distances here the latter makes US accident rates look better than the former. Anyway I did a quick multiple regression on 19 countries I found data on (using the US favored accident rate per million vehicle km traveled) and had Toxoplasmosis rates for. There was a significant effect of GDP - and this has been reported before for accident rates but Toxoplasmosis prevalence was not significant, nor was there a significant interaction. All but four of the countries I found data on were European (the exceptions were Australia, Japan, New Zealand and the USA). It would be good to include a bigger range of countries, 19 countries doesn't give much sensitivity - or do this on US data only. I found the accident rate by state but not the toxoplasmosis prevalence although I think it exists.