Earlier quoted context omitted.
Fatalities should be measured in terms of miles travelled. Not per mile of highway, or per number of cars, or per capita. Someone who drives five days a week, 60 miles per day is going to have a higher risk of getting into an accident that someone who drives only on weekends, 30 miles per day. This does have any bearing on how safe the roads are.
I agree with you that miles travelled is a better metric. But traffic density would be another interesting metric. Temporary speed limits on the autobahn are typically dynamically adjusted to the traffic situation.
Eventually, I see so many other factors that could enter the picture that I’m wondering if that comparison makes any sense at all (apart from proving the writer is trying to force his point?).