This is a good example of a bad study. Or at least, one that I don't understand. "Cannabis doubles fatal crash risk" is the headline. The study says that
The French National Institute for Transport and Safety Research found evidence of cannabis use among 7% of drivers involved in fatal crashes.
Also, 2.9% tested positive for both cannabis and alcohol use. It is unclear from the article whether that 2.9% is part of the 7% or in addition to it. A doctor later in the article says "few" people tested positive for cannabis alone.
How do they draw this conclusion without having testing results for drivers who were not involved in fatal crashes?
Wait, they do claim that cannabis is found in 2.9% of all drivers. Where do they get that from? To my knowledge, the test for cannabis is a little more complicated than a roadside breathalizer, so I'm curious to find out where this statistic comes from. It seems like they would only be able to collect samples from people who already came into contact with law enforcement for some other reason, in which case their results would not be very good for a "normal" driver baseline.