I might as well double down on my comments on this post on the differences in fatality rates based on education level, reporting on an academic study (from which I've stolen the graph).
Most of the commenters and the study itself focus on the differences in deaths among the different levels. But what struck me is the deviation in the relationships: simply put, the rate goes down between 1995 and 2010 for all groups, all except those who didn't graduate high school. I didn't see any explanation offered for why that group might have an increase in fatality rate. I could guess maybe a change in drug usage--meth and oxy usage is, I believe, up particularly in rural areas, but that seems unlikely to move the trend that much.
Chalk it down to a mystery.