Many parents dread the moment they allow their teens to first get behind the wheel of a car. Even more so when the child is male — young men are nearly three times more likely to be in a fatal accident than young women. This fact is reflected in insurance premiums. You’d think it would be considered when Ivy League scientists research traffic safety, too.
But the sex difference was subtly ignored to frame cannabis as the culprit in a Harvard study on cannabis and driving risk. Published in Drug and Alcohol Dependence, the article unequivocally states in the title: “Recreational cannabis use impairs driving performance in the absence of acute intoxication.”
Just sentences into the abstract, the researchers (Mary Kathryn Dahlgren et al) begin their disingenuous caveats: “These results may be related to other factors associated with early exposure such as increased impulsivity.” In more direct language, the cannabis-using group was 82% male, while the control population was only 38% male. They found an association with risky driving because young men are more likely to use cannabis than women.
It’s hard to describe demographics without talking about sex or gender. Yet the Harvard researchers, led by Staci Gruber, never use the words “sex” or “gender” in the article’s body. Not once. They also avoid saying “male,” “female” or “man,” and only write the word “women” once (in reference to the definition of binge drinking).
Why are the authors avoiding this obvious topic? Did they really forget about sex? Or are they trying to cover up a flaw so large that it makes this study entirely irrelevant?
Instead, when describing their population, they contend that “[controls] and all cannabis users were well-matched for age, IQ, and alcohol use,” conveniently leaving out sex. Yet, sex has a greater impact on driving risk than any of the factors they account for. The sample size (16 controls and 28 users) is too small to control for gender statistically.
Regardless what the researchers hoped to study, they can’t provide compelling evidence for anything when such a major confounding factor is unaddressed. And from what is left unsaid, it seems that they know this.
Gruber’s team mistakenly claims that theirs is the first study on how sober cannabis users drive. Two such studies were published in October 2019 and November 2019 after this article was submitted for review, but weeks before it was revised and published. The other papers found that cannabis users drive slower and more consistently than non-users, in contradiction to this study (although these papers frame slow driving as dangerous, bearing the same biases against cannabis as Gruber and her collaborators).
What about the research itself? The Harvard scientists insinuate a dubious hypothesis that long-term cognitive impairment from cannabis use — which, in fact, has not been convincingly demonstrated — causes people to drive poorly, even when sober. The notion of long-term impairment is controversial and has not been established beyond an acute withdrawal phase, usually lasting up to a week, and the roughly month-long window when THC metabolites are being eliminated from the body.
Time and time again, diehard prohibitionist have argued that cannabis legalization should be forestalled because stoned drivers will cause fatal accidents. The Harvard study will provide more fodder for their feckless arguments. But closer inspection shows that blood levels of THC are not correlated with driving risk.
A handful of studies have noted that stoned drivers tend to drive slower than sober drivers. This observation has been twisted into ridiculous warnings that driving within the speed limit is a “public health concern” because of the association with THC.
According to aggregated research, stoned driving has a small but detectable impact on car crashes; driving while high appears to increase crash risk by 25-40%.1 Meanwhile, a blood alcohol content of 0.08, the legal limit in the UK and US, is associated with around 900% increase in crash risk — about 20-40 times more dangerous than driving after lighting up.
Stoned driving should be considered when assessing the public health impact of cannabis, but not disproportionate to the established harm. And scientists should pay more attention to how they’re recklessly driving public discourse about cannabis in the wrong direction.
Adrian Devitt-Lee, Project CBD’s chief science writer, is pursuing a PhD in math at the University College of London.
Copyright, Project CBD. May not be reprinted without permission.
- In the US in 2016, there were roughly 1.25 fatal crashes for every 100 million miles driven. If cannabis increases fatal crash risk by an average of 25%, and one in ten drivers in the US were driving high, you would expect 1.28 fatal crashes for every 100 million miles driven, an increase of 0.03 fatal accidents per hundred million miles. It is imperative to consider risks in the context of other factors that influence driving.
In 2017, Michigan’s government founded a commission to recommend policy on cannabis and driving, specifically how to set THC limits. Two years later, in March 2019, they released their conclusion: [T]he Commission finds there is no scientifically supported threshold of ∆9-THC bodily content that would be indicative of impaired driving (emphasis added).
Driving accidents cause about 60,000 deaths in the US each year, and nearly all of these are due to driver errors. Regulators are still trying to understand the risks associated with driving while high. Canadian epidemiologist recently analyzed data from the stoner holiday 4/20, finding that there was up to a 12% increase in the number of fatal accidents that that day. This is within typical daily variations, suggesting that, if there is an increase in crash risk, it is small.
Driving under the influence of cannabis (DUIC) is framed as a major risk associated with legalizing cannabis. Studies on cannabis and driving suggest that lighting up before getting behind the wheel increases crash risk between 25-40%. This is equal to the effect of having a single drink one hour before driving.