They also tend to be carrying 5-25x more people than cars, so there are far fewer of them. They also tend to be driving in the city on the busiest streets, whereas arbitrary cars are more often driving on vacant streets and highways.

so try the comparison using (miles x flow) instead of miles if you're measuring safety, and if you're measuring worthwhile risk, multiply that by passenger-miles. If you're talking about pedestrian safety, you'll probably have to square the (miles x flow) because pedestrians are also less safe the busier a street is.