You estimate from survey data how many more people are taking several hundred mile road trips instead of flying, then apply known traffic fatality statistics how many wound up dead from that decision.
See https://www.schneier.com/blog/archives/2013/09/excess_automo... for a reference.