Although deaths per km is pretty meaningless, deaths per hour or deaths per journey would better reflect the risk
Why?
Because comparing risk by distance when objects don’t travel similar distances gives skewed results. Let’s say I live 100K from work. If I drive I have a 0.00000028 chance of dying, 0.00000295 for cycling and 0.00000448 for walking. So cycling’s would appear to be 1.5 times safer than walking. But in reality people don’t walk 100K to work or cycle 100K to work, they choose a different form of transport. Let’s say the average uk commute is 30 minutes, you average 60Kmh in the car 20kmh on the bike and 5Kmh walking. So the distances travelled are 30km, 10Km and 2.5Km respectively. The death rates for the journeys are 0.000000084, 0.000000295, 0.000000112.
So taking into account the speeds people move at and the time they’re exposed to the risk you’ll see that cycling is 2.6 times more dangerous than walking, rather than the 1.5 times safer the base line statistics would suggest.
A moment’s thought will tell you that it’s patently absurd that walking is more dangerous than cycling. Try thinking of all the times that friends and work colleagues have regaled about all the near misses they’ve had walking into work rather than cycling.