I prefer the death per people metric as I am more interested in how likely I am to receive some bad news than some metric based on distance.

Both measures have bias, the "per people" metric doesn't take into account when people are actually driving while the "per kilometer" metric puts too much emphasis on long distance driving, which is usually done on motorways where it is the safest. Maybe the best metric would be "per time spent on the road, including as a pedestrian on the sidewalk", but I guess it is harder to estimate.

Anyways the UK is doing well on both metrics.