Another badly flawed study

Jeff Brubacher from Vancouver General Hospital published “Cannabis use as a risk factor for causing motor vehicle crashes: a prospective study.”  The study has earned a great deal of press due to his finding that there was no increased risk of crash responsibility in the driver pool he studied for drivers with a blood THC level between 2 and 5 ng/ml.  There was an increased risk of crash responsibility for drivers over 5 ng/ml, but that increased risk was not significant.

Although the study appeared to be well designed and carried out, it suffers from a serious flaw that throws the conclusions into question.  The average time between a crash and taking a blood sample was 101 minutes, sufficient time that the laboratory determination of blood THC level would not represent the blood THC level at the time of the crash, or the blood THC level that would be determined forensically by a responsive police force.

As Hartman pointed out in 2016, the maximum blood THC level  dropped an average of 73% within the first 25 minutes after beginning to smoke a joint.  After 101 minutes, the level of THC remaining in the blood would be so low that in many occasional users it would not be detectable by today’s technology.  THC levels would remain elevated above 5 ng/ml only in marijuana addicts and other chronic users of marijuana.

The result of the time delay is that users with a THC blood level between 2 and 5 ng/ml at the time of the crash would most likely have undetectable levels of THC 101 minutes later and would therefore be pooled in with drivers who had no THC on board.  This would be expected to increase the rate of crashes for the “control” population, rendering any Odds Ratio calculation somewhat moot.

But what about the drivers who tested above 5 ng/ml in Brubacher’s study?  The only drivers who could have such a high level after 101 minutes of abstinence would be addicts and other chronic users.  Those drivers had an Odds Ratio of 1.74 higher than the “controls” but there were not enough cases to make that statistically significant.  

Only 20 cases out of 192 positive THC drivers tested at or above 5 ng/ml, a mere 10.4%.  Compare that with forensic results in Colorado where 55% of THC-positive drivers were above that mark in 2016 and 58% in 2017.  The Canadian laboratory results were far, far less meaningful than what would expect a responsive police force to deliver, and even further removed from what the THC levels were at the time of driving.

Based on these facts, Brubacher’s conclusions that drivers between 2 and 5 ng/ml THC do not present an increased risk to driving does not seem to be supported. 

But paradoxically, Brubacher does seem to find at least a slight increased risk for chronic marijuana users.  This is a somewhat unusual finding, one that was also found by Doroudgar et al. in their 2018 study.


DUID Articles