Marijuana and driving… plus bad reporting and bad science

Those of you who have followed this blog for awhile know that I’ve been very interested in the evidence regarding the dangers of marijuana and driving. This is an important issue — particularly when it comes to the topic of marijuana legalization/criminalization. Prohibitionists point to the supposed dangers of intoxicated driving as reason to oppose legalization (whether relevant or not) and also use the stoned-driver bogeyman to push for zero-tolerance driving laws that are based simply on the presence of cannabinoids — a sneaky back-door means to criminalize marijuana use.

Of course, there’s no doubt that marijuana use affects reaction and space/time perception. That’s pretty much a given, and anyone who has smoked pot will confirm this. So it’s reasonable to assume from that fact that marijuana and driving would be a very bad combination. And yet… and yet, the government has an exceptionally difficult time demonstrating that “fact.” They turn to discredited “studies” or isolated anecdotes about some tragic accident involving a stoned driver (where it usually turns out the driver was also high on a dozen other drugs).

The truth is, and studies have confirmed this (see U.S. National Highway Safety Administration (1993), Dutch study (1994), Australian study (1998), Transport Research Laboratory (2000 and 2001)), most stoned drivers are much, much safer than drunk drivers and are even safer than tired drivers or drivers talking on cell phones. It’s because pot makes people cautious (and sometimes a little paranoid). People who are stoned know that they are affected, and so they compensate by driving slower and focusing their entire energy on driving. Most pot smokers, when behind the wheel of a car, become old people. (Unlike many drinkers, who become reckless and believe themselves to be invincible.)

My own joke on the difference between a drunk driver and a stoned driver:

A drunk driver will speed through a stop sign without even noticing it, while a stoned driver will stop and patiently wait for it to turn green.

The fact that actual marijuana impairment is so hard to classify may be why there has been so little interest on the part of the government on researching a means of measuring a level of marijuana intoxication that would compare to Blood Alcohol Content levels. (That, and their desire to use cannabinoid presence to criminalize use.) Instead, the government has encouraged research into a variety of questionable drug-testing-in-relation-to-fatal-crash-statistic studies (many of which even fail to separate drivers from passengers).

So my attention was caught by this story from radio-canada that started out fairly dramatically:

U.S. drivers who tested positive for cannabis over a 10-year period had a 29 per cent higher risk of causing a fatal crash than motorists not taking the drug, a new Canadian-led study suggests.

Strong damning proof, or a mix of imprecise words and absolute rubbish?

Here’s what the sentence implies:

1. That drivers who tested positive for cannabis were somehow tracked over a 10 year period of their lives.
2. That certain drivers were identified as causing fatal crashes
3. That cannabis-positive drivers had a 29 percent higher rate than all other drivers.

I wrote to Michel BÚdard, author of the study, and asked for some clarification. Turns out #1 is a actually bad wording; #2 is bad reporting of limited scientific data; #3 is simply false.

Me: Should it read “U.S. drivers who tested positive for cannabis over a 10-year period…” or “Over a 10-year period, U.S. drivers who tested positive for cannabis…” … I’m guessing that it actually meant that the data covered 10 years, but that the instances of drug testing were relatively proximate to the crash.
BÚdard: You are correct, we used data from 1993 to 2003 and the presence of cannabis was in relation to a given crash.

Me: How was it determined that the person who tested positive actually CAUSED the fatal crash?
BÚdard: … we cannot claim that the driver “caused” the crash but rather that there is an association between the presence of cannabis in the blood and making an unsafe driving action.

Me: The 29 per cent — is that really a comparison between those who tested positive for cannabis and those who did not (which would include those who tested positive for alcohol but not cannabis), or is it a comparison between those who tested positive for cannabis and those who did not test positive for cannabis OR alcohol?
BÚdard: All drivers tested negative for alcohol. The comparison was between those who tested positive or negative for cannabis in the absence of alcohol.

He was also nice enough to send me information from the abstract (the full study isn’t yet available online).

We used a cross-sectional, case-control design with drivers aged 20-49 who were involved in a fatal crash in the United States from 1993 to 2003; drivers were included if they had been tested for the presence of cannabis and had a confirmed blood alcohol concentration of zero. Cases were drivers who had at least one potentially unsafe driving action recorded in relation to the crash (e.g., speeding); controls were drivers who had no such driving action recorded. We calculated the crude and adjusted odds ratios (ORs) of any potentially unsafe driving action in drivers who tested positive for cannabis but negative for alcohol consumption. In computing for the adjusted OR, we controlled for age, sex, and prior driving record.

So what do we have? First, all those who tested positive for alcohol were eliminated from the study, so the sample is dangerously skewed. Second, the best that they could do is identify drivers involved in fatal crashes that had been cited for something, with no information as to whether that related to the crash. If an alcohol impaired driver crashed into someone sober who used cannabis days earlier, and both drivers were cited for speeding, the drunk driver would not be included in the study, but the sober driver would be classified as positive for cannabis and having an unsafe driving action associated with a fatal crash. And even then, all we have is an association, not a cause. There could be many explanations that have nothing to do with cannabis impairment. The study tells us nothing useful, but it does feed the media feeding frenzy.

And while I appreciate Mr. BÚdard’s willingness to be candid with me about the study, he has contributed directly to misinformation through promoting the reults of a study with limited value, through his obvious bias, and through his imprecise and downright false communications with the press [from both the radio-canada article and here]:

“Those who tested positive for cannabis had 29æper cent more risk of having committed a driving action that led to the crash than those who did not,” BÚdard told CBC News.

“It tells us that cannabis is not a safe substitute for alcohol, and I especially mean that for young people,” said BÚdard. […]

What he found was that those victims where THC was involved, were at greater risk when involved in an accident.

“Compared to people who have not tested positive for cannabis, or THC in this case, which is the metabolite we’re looking at, people who did test positive have a 29 per cent higher risk.” […]

“The big thing is that cannabis is not a safe alternative to drinking when it comes to driving.”

And once again, the prohibitionists fail to find their smoking gun.