I'm not sure if I see how fatalities per 100k vehicles is a better metric than per 100k inhabitants: the composition of vehicles is very different from country

I'm not sure if I see how fatalities per 100k vehicles is a better metric than per 100k inhabitants: the composition of vehicles is very different from country to country. Given the example above, in Indonesia I would expect a very considerable portion to be scooters which while more dangerous for the driver themselves are safer for others. Meanwhile in the United States, cars are the big majority, being safer for the driver but more dangerous to others.