How many deaths occur annually by telecommuting?
I’m sure if you look at the studies those numbers came from, they state all kinds of limitations. The groups studying this do the best they can with the data that is available, and if they’re very lucky they have information on how far off they might be from the real numbers. For example, maybe they do a pilot study comparing measured distance that people walk with their self reported distance.
Bus and train data is probably pretty good, because the agencies running them can report passenger miles and fatalities.
There are also going to be other differences, such as the type of person who decided to bike 5 miles instead of drive 5 miles might simply be a healthier type of person, because they will choose active things to do for leisure as well.
The problem is, diving into the rabbit hole of not being able to account for every little detail is a great way to avoid getting any useful information at all. For example, maybe the bicycle number is 10 instead of 11, and nitpicking about that ignores that fact that they’re far more dangerous than taking a bus.