I'm not sure that makes much sense as a safety metric, bicycles obviously travel less distance on average per instance used, so the 'rate of death' would not have the same scaling as distance traveled
Well, when I'm considering biking or driving my car, the destination is fixed. I can drive 2 miles to the store or I can bike 2 miles to the store. I'm not choosing between, should I take an average bicycle trip right now, or should I take an average car trip right now. So the danger per mile does seem like the relevant metric there.
You are correct in your consideration, but above 15 kilometers bicycling is not an option while using a car is, so the cars have a hugely larger average drive distance which skews the metric making it effectively useless.
