http://www.helmets.org/walkerstudy.htm has an interesting analysis of that study, which notes some of the same issues I thought of, primarily the whole assumption that helmets are "the factor" - and also that 3 inches closer actually means more likely to be hit.
Our recent fatality was riding off the road on the shoulder, if my information is correct, and a phone-impaired driver swerved enough to take him out. So, most traffic was passing with all kinds of distance, but partially *because* that distance was enough to put him out of the "attention range," he's gone.
That article points out also that the roads are *very* different in the UK.