Interesting stats ARE compiled.
Accounting for population, there were about 44% more accidents in 1990 vs. 2009. So overall accidents per capita are down. (2600/100K vs. 1800/100K)
BUT. . . this site shows different stats than Dana has:
Only 38% of deaths are alcohol vs. 60% in 1982.
Somewhere, the #'s are fudging. (The NHTSA alcohol-related accidents # includes drunken passengers, drunken VICTIMS of accidents and drunken pedestrians hit by sober drivers.)
This is an interesting mental exercise. I'd like to know the # of drunk driving accidents per year 1990-2009. They track deaths, not accidents. :(