i think the biggest problem here is probably going to be the denominator. (andrew's already pointed out that the numerator could be expanded, too - mortality's the most salient for riders, and the most unambiguous to measure, but there are plenty of people who survive accidents that are nonetheless life-changing, career-ending, etc.).
so:
-we're measuring deaths per X. person-miles? person-hours? cyclists? outdoor cyclists? total population?
-i think rates would be preferable to absolute numbers here, since the popularity of cycling (and driving) has definitely gone up and down, and all else equal, more riders and more cars on the road would mean more accidents overall, if not relatively.
-if you wanted to calculate and odds ratio or relative risk, you'd probably want to get an 'exposed' group (cyclists) and a control group (non-cyclists), follow them for, say, a year, and then it's simple arithmetic to work out risk of death in both groups. unfortunately this doesn't get us very far, since we can assume that the risk of being hit by a car while cycling is zero for non-cyclists.
-so on that front i think number of deaths per km (or hour) ridden would be 'best'.
-in terms of controls, i think there would be a lot of things to account for:
-seasonality comes to mind - probably far more accidents overall in summer. on the other hand, with so few people riding mid-winter, the 'effect size' of a single fatality would be larger.
-gas prices (more cars driving more miles when gas is cheap)
-time of day
-maybe proximity to weekends/holidays would put more drunks on the road?
-geography? many experienced cyclists will tell you that 'the drivers are good' in this area but 'bad' in that one. is there anything to that?
-maybe timing w/r/t advertising campaigns, police safety blitzes, etc
-helmet laws by region
that's about all i've got at the moment. the damn thing with stats is that it's easy to gather a huge pile of data that actually tell you nothing, or at least don't tell you what you think they're telling you.
-mike
____________________________________
https://lshtm.academia.edu/MikeCallaghan http://howtobeswiss.blogspot.ch/