How bad is stoned driving, and what should we do about it?

Dishonest advocacy aside, what are the actual risks of stoned driving? The answer, from what seems to be a well-done case-control study, is that driving stoned is hazardous, but much less hazardous than driving drunk. (A relative risk of 1.83 - meaning that driving a mile stoned is about as risky as driving two miles sober - strongly suggests that cannabis-impaired driving is a problem, but also that it isn’t much of a problem; the relative-risk number for alcohol is over 13.) On the other hand, the same study shows that adding cannabis or other drugs to alcohol substantially worsens the odds: alcohol-and-something-else has a relative risk of 23.

Given those numbers, and the technical difficulty of identifying cannabis-impaired driving (because impairment doesn’t track cannabinoid levels in blood nearly as well as it tracks alcohol levels) I’d propose the following rule: anyone who tests positive for cannabis on a mouth swab (which detects use within the past few hours) should be considered guilty of impaired driving if that person’s BAC is detectably different from zero. All that means is that, if you’ve been toking and drinking, you need to wait as many hours as you’ve had drinks before getting behind the wheel.

Of course, if I were advising someone personally, I’d be much more cautious: driving within six hours of using cannabis is pretty damned stupid. But the same is true of driving after a sleepless night. The question isn’t what’s imprudent; the question is what’s hazardous enough to make a serious criminal offense. So far, the numbers I see about stoned driving (in the absence of alcohol) don’t bring it across that very high threshold.

Given the long latency of THC, the “zero-tolerance” rules now being passed in some states, which making driving with any detectable cannabis on board drugged driving per se, without evidence of actual impairment, are simply a backdoor way of recriminalizing cannabis use.

Footnote A case-control study - comparing a group of drivers responsible for accidents with a random sample of all drivers - is the only way to figure out what’s really going on. The observation that, as cannabis use spreads, more crash-responsible drivers have cannabis on board tells you precisely nothing. After all, if consumption of blintzes were to increase, more responsible drivers would test positive for ricotta. You neeed a denominator, not just a numerator.