A new B.C.-based study is examining the link between marijuana use and car accidents by testing the blood of injured drivers in hospital -- without their consent.

Dr. Jeffrey Brubacher says his study will include 3,000 drivers being treated at five separate emergency departments in the province, but since the data will be completely anonymous and not shared with law enforcement the patients' approval is not required.

The doctor says the study has undergone a rigorous ethical review, and will only use leftover blood drawn for the drivers' treatment.

"If drivers do not require bloodwork for clinical purposes, they will not be included in our study," Brubacher told ctvbc.ca in an email.

A 2010 government survey of B.C. drivers on the road between midnight and 3 a.m. found nine per cent tested positive for one or more drugs. Marijuana was the most commonly detected, accounting for 63 per cent of cases.

"There are almost 3,000 people killed in car crashes in Canada every year and many of these involved impaired driving," Brubacher said. "Certainly there is an urgent need to understand the factors that contribute to car crashes."

But there remains no conclusive research on the impact of cannabis on driving ability, according to Brubacher, who works in the Vancouver General Hospital emergency department.

Though similar research has been conducted in France and Australia, Brubacher says their results may have been skewed because it often took hours to collect blood samples from the subjects.

B.C. researchers will be able to collect blood much sooner in emergency rooms, however, which will more accurately reflect THC levels at the time of the crash. THC, or tetrahydrocannabinol, is the main psychoactive substance in pot.

It's possible the research will then lead to an established allowable THC blood content similar to the blood alcohol content tested using roadside breathalyzers.

Discussing his research in the BC Medical Journal, Brubacher said the evidence available suggests pot increases the risk of crashing but "this risk, and how it varies with cannabis dose, is not well quantified."

Simulator studies have shown drivers heavily impaired by marijuana were more likely to crash into sudden obstacles, Brubacher wrote, while "moderate doses impair highly automated tasks but leave complex functions such as interpretation and anticipation of traffic patterns relatively intact."