Yes, but what you are missing is that behaviour has changed! We have that level of net road deaths today, given that kids are being driven everywhere.

So you can't state from your data that it has got safer: you need the child deaths/km or deaths/hr walking.