I agree that fatalities can be expected to rise as miles driven rise, however these two data points are not so closely linked. For example, driving ten miles in western Kansas on I-70 at midday is not the same as driving ten miles in a metropolitan area filled with shopping centers, with the sun low on the horizon ahead of you and a dirty windshield.

There are many, many variables which complicate the study of traffic safety and my hat is off to any traffic engineers able to deliver meaningful results. It does happen, but it's far more complicated than tracking per capita miles driven (not that you were making this specific argument, just pointing out the obvious).