The rate of traffic fatalities, measured by the number of deaths per mile driven, has spiked almost 20% over what it was in 2019. A number of theories have tried to explain why this might have happened. One proposes that marijuana, which is now legal or decriminalized in 30 states, has led to greater consumption and more accidents.

But connecting the dots between marijuana use, legalization and road injuries has been a tricky public health puzzle. A 2019 study in the journal Addiction found that after marijuana was legalized in Washington, Colorado and Oregon, there was only a slight, temporary increase in traffic accidents in those states and their neighbors. Other studies saw no correlation at all, whereas a paper in the Journal of Studies on Alcohol and Drugs that analyzed 10 years of crash data from 2009 through 2019 found that legalization was associated with a 6.6% increase in injury crash rates.

For a January 2022 study in The New England Journal of Medicine, one research team went directly to emergency rooms. They tested blood samples from patients who’d had a car crash, with about 3,500 of those samples coming from a period before cannabis legalization and 700 from after  legalization. In the post-legalization set, researchers found that the blood levels of tetrahydrocannabinol (THC)—the psychoactive ingredient in marijuana—had risen sharply, roughly double the levels in earlier drivers.

Even this isn’t a smoking gun, however. “The correlation between THC level and impairment varies widely depending on the person,” says Jeffrey Brubacher, one of the authors of the NEJM study and director of the road safety and public health research lab at the University of British Columbia. While some of those with THC in their blood may have had critical impairment, that might not be true for all. “An occasional recreational cannabis user might be pretty impaired after smoking a joint. But someone who uses cannabis all day every day could have high THC levels with only subtle impairment,” he says. Other mitigating factors might include how the person consumed the cannabis and whether or not they did so on an empty stomach.

While states have set limits about how much THC drivers may legally have in their blood, such factors make it hard to draw a straight line between the THC levels and actual intoxication. alternate forms of testing may also miss the mark. The Standardized Field Sobriety Test, the most common roadside test for behavioral impairment in the United States, was designed for alcohol and may miss most cases of cannabis impairment. Another roadside test, the Drugs Evaluation and Classification program, does much better, but it requires specialized training to administer, and the test takes an hour.

Without good measurements for cannabis intoxication, it’s difficult to gather public health data about the role that cannabis may play in car accidents. But a few innovations are looking to make that task easier.

Law enforcement has generally relied on blood or urine tests to measure levels of THC, and neither can be done quickly during a traffic stop. While drivers suspected of impairment can be taken to a clinic or police station to have their blood drawn, that may happen hours after a traffic stop or an accident when THC levels are likely to have fallen.

A new tool called EPOCH, created in the lab of Hahko Lee, director of the Biomedical Engineering Program at Massachusetts General Hospital, could provide a quicker method. A swab is held in a driver’s mouth for one minute, then a smartphone attachment measures the THC level in saliva from the swab. Results appear in about five minutes.

Although this test represents a significant improvement, THC levels in saliva may not tell the whole picture, Brubacher says. “Saliva levels of THC may be higher than what’s in the blood or brain,” he says. One option is to pair these tests with a newer generation of cognitive test.

Being stoned causes clear cognitive shifts—slower reaction times and difficultity in staying vigilant, estimating distances and coordinating motions. All of these affect safe driving, and all can be targeted in tests. One being developed by the company Cognivue, which created the first digital test approved by the Food and Drug Administration for measuring cognitive decline,  could fit the bill for roadside use.

Lee envisions that drivers would take such a cognitive test on a smartphone or tablet while the EPOCH test was being processed. “The combined result of the saliva THC level and the cognitive test may produce a more accurate result of whether someone is under the influence,” Lee says. That combination would also generate research data that could help show how cognition and THC levels are correlated, and how they relate to accidents on the road.

But saliva tests on their own might become more accurate and useful as part of a database of cannabis-infused spit, says Lee, who envisions accompanying data on how often drivers use cannabis and whether they’ve recently eaten. “That would really help facilitate this type of research,” he says, and could help researchers begin to answer crucial questions about the  effects of cannabis legalization on auto accidents.