Even for planes the stat is deceptive. Most people, hearing about the safety of flight, would think that if they replaced their car with a plane then they'd be far less likely to die in a crash, but that would be untrue because the real risk in flight is take off and landing - the distance traveled is near irrelevant.
But if I don't care about the destination, and only the risk, then I may simply choose somewhere close enough that the (risk per km * distance) from the other mode(s) of transportation is less than the risk from a flight.
I was thinking more along the lines of: I'm in Berlin, and it's possible for me to visit friends family in the UK by train* or by plane. If I want to visit them, I pick one of those options (plus the option of ferry or train for the channel).
If I want to go to the shops, my options are bus or walk*.
If I want to just go somewhere on holiday and I don't care where exactly, I can take the bus, trains, planes, etc., and pick my destination to taste, and at that point it's min( P(death or injury ∣ x ∈ destinations, y ∈ modes of transport) )
* not by car, I would have had to transfer my license within 6 months of moving here and I didn't, so it's not valid any more
Measures like accidents per hour or per trip skew the numbers, because air travel generates much more distance per hour and per trip than other forms of transportation.