Given NHTSA's sloppy reasoning and screwy statistics WRT alcohol, they aren't a particularly good source of information.
An elderly man (a teetotaler) has a heart attack while driving and collides with a restaurant that serves alcohol. Fortunately, it is closed at the time so the only fatality is the driver. According to NHTSA's definitions, it is an alcohol related traffic fatality.
All of the stats you cite are such that no reasonable conclusion can be drawn. For example,
4 to 14 percent of drivers who sustained injury or died in traffic accidents tested positive for THC.
First, that's a pretty wide swing, can't they narrow it down if they have actual data? Answer, no because they extrapolated the data from a small (possibly cherry picked) sample.
Of that 4 to 14 percent, how many were currently high? You test positive for THC long after the high is gone.
What percent of drivers not involved in an accident tested positive for THC (or would have if anyone cared to test)?
Of that 4 to 14 percent, how many were concurrently drunk?
For the 3,000 Australians, were they at fault or were they 'at fault' because police found a roach in the ash tray and so decided they must have been at fault (quite common). I hear a question being begged.
Meanwhile, none of that has a single thing to do with overdose.