This is untrue. Ohsfeldt and Schneider did not 'remove fatal injuries from life expectancy data', they used a regression analysis to adjust life expectancy values to equalise the number of fatal injuries. They have not actually dug deep to actually investigate whether fatal injuries does explain low life expectancies, only observed that there is a correlation.
This methodology is questionable, to say the least, and can potentially produce perverse results, and to my knowledge, they have not shown their working to show that they have not fallen into the many many pitfalls this method can generate.
This methodology is questionable, to say the least, and can potentially produce perverse results, and to my knowledge, they have not shown their working to show that they have not fallen into the many many pitfalls this method can generate.