For example, among sites where posted speed limits were raised, the crash rates increased at some sites but decreased at other sites after implementing new posted speed limits. Similarly, the direction of changes in crash rates varied among sites in the same state where speed limits were lowered.

The only clear conclusion they were able to draw was the obvious one: If you raise the speed limit, speeding citations decrease.

Increases in those likely correlate with an increase in severe injuries and those need to be counted too. I think it's safe to assume there are probably more of those than deaths.