As cannabis wafts back into mainstream America on growing winds of legal change, a west coast company says it has reached a technological milestone that is quickly becoming crucial to the justice system: a device police can use to detect recent marijuana use—and the amount consumed—in the air a driver exhales.

But don’t hold your breath.

Although many researchers and companies dream of fielding a roadside test for driving under the influence of cannabis, such a product may still remain years away from practical service use. Millions of dollars’ worth of speculative research will likely be needed to correlate data about consumption amounts with actual driver impairment. And the issue of stoned driving remains enormously complex, with experts still highly uncertain about marijuana’s precise effects on cognition and competence.

This fall a California-based company called Hound Labs announced it has developed a portable breathalyzer-like device called the Hound that could quantify recent cannabis use in the breath—the latest in a series of claims in the field. But like its predecessors, this product will have to run a tough gauntlet of medical and legal complexities. “There’s a little bit of hype,” says Nick Morrow, a retired Los Angeles County Sheriff deputy and court-qualified expert in drug symptomatology who has taught a drug recognition course for officers. “The good thing about the U.S. is if there’s a way to make a dollar, you can bet people will be lined up around the block to make the next best mousetrap. That’s what’s going on with cannabis testing. Breath, oral, fluid, saliva—all of them are competing to come out with the easiest, greatest devices,” Morrow says. “Maybe those devices aren’t ready for prime time yet, but everybody wants to get the patents. As far as having something that really works, defense and prosecutors can agree—the science is not there yet.”

New Fears

Fears of cannabis-impaired driving have spiked since states began legalizing adult recreational use of the psychotropic weed, beginning with Colorado and Washington in 2012. Oregon, Alaska and Washington, D.C., followed suit in 2014. This year California, Arizona, Nevada, Maine and Massachusetts could join them. Although marijuana-related arrests have reportedly plummeted between 80 percent (pdf) and 95 percent in those jurisdictions, law enforcement officials have publicly worried about a rash of high drivers hitting the road. Up to one in 10 weekend nighttime drivers in the U.S. may have cannabis’s main active ingredient, tetrahydrocannabinol (THC), in their bodies, studies suggest (pdf). Police in Colorado have said figures show marijuana-involved road fatalities doubling (pdf) in the state amid legalization.

Credit: Hound Labs

After cannabis is smoked or eaten, the presence of THC in the blood rapidly rises and then subsides after a few hours. While in the bloodstream the chemical’s effects can include delayed reactions, misperceptions of space and time, and distractibility, which is why some federal statistics (pdf) indicate that being high on cannabis effectively doubles a driver’s crash risk. Most available drug tests, however, can only measure marijuana’s metabolites—which are not known to cause any impairment by themselves but can linger in the body’s stored fat cells for weeks after THC’s befuddling effects have worn off. This leads to positive toxicology screens for a functionally sober person who may not have consumed weed for weeks. As cannabis laws loosen, the race is on to invent a device that is as effective and convenient as an alcohol breathalyzer, and is satisfactory to both police and public defenders.

Marijuana intoxication itself is pretty familiar to law enforcement officers. About half of Americans have reportedly tried pot at least once, and around 32 million use it regularly. According to retired California Highway Patrol Commissioner Dwight “Spike” Helmick, officers currently have a number of options to screen for stoned drivers. They primarily use a combination of sobriety checks and tests of blood, urine or hair for marijuana metabolites. This can be enough to convict impaired drivers who crash and/or hurt people, Helmick says.

Roadside marijuana detection technology is definitely advancing. For more than a decade police in Europe and in limited trials in the U.S. have used saliva swabs, which can quickly yield a positive or negative reading for active THC, but not its amount.

Hound Labs says it has gone much further. It recently announced successful field trials of its device, which the company asserts can both rapidly detect the presence of active THC in breath and measure the amount in picograms. Hound Labs developers also say their device is the first to have detected active THC in the breath of people who have recently eaten cannabis, not just smoked it. The company says the device will undergo more field testing in Lompoc, Calif., after a pilot trial with a law enforcement group they say they cannot name. “I think it’s extraordinarily needed,” Helmick says of such a product.

Hound Labs’ third-party product developer Joe Heanue, president of Triple Ring Technologies, says the device uses a combination of chemical materials science and optical scanning. Users blow into a disposable cartridge filled with a chemically active substrate that reacts and binds to THC molecules, effectively tagging them so they can be optically scanned by the device. “It’s using technology that’s commonly developed in various life sciences tests and point-of-care diagnostics tests,” Heanue says.

Hound’s CEO and co-founder, Mike Lynn, says peer review of the company’s data is “still another couple months away.” Also unknown is the device’s operational tolerance for the often harsh environments of the nation’s roadways. “It hasn’t been done in freezing cold or 100-degree heat, or rain or wind,” Heanue says.

And, more fundamentally, science still lacks data correlating the presence of THC and actual impairment. Getting such data could cost millions of dollars in human trials, and the effects of the drug vary tremendously between users. “There is understandably a strong desire by both lawmakers and the public to create legal limits for marijuana impairment, in the same manner as we do with alcohol,” said Marshall Doney, head of the American Automobile Association, in a landmark 2016 analysis. “In the case of marijuana, this approach is flawed and not supported by scientific research,” Doney added. “It’s simply not possible today to determine whether a driver is impaired based solely on the amount of the drug in their body.” The National Highway Transportation Safety Administration (NHTSA) Drugs and Human Performance Fact Sheet states that “it is inadvisable to try and predict effects based on blood THC concentrations alone, and currently impossible to predict specific effects based on THC-COOH concentrations.” (THC-COOH refers to carboxylic acid, a marijuana metabolite.) “It’ll take time to get all this correlated,” Helmick says.

Existing Science

The key word here is “tolerance.” Morrow, the drug symptomology expert, confirms that just 10 milligrams of THC—one standard Colorado dose—can have significant effects on an inexperienced user. But experienced users quickly develop a tolerance to the drug’s effects, Morrow says. “We need several good-quality research projects that take into account driving behaviors, field evaluations and real-time THC levels,” he adds. “The science we need to have has to be good enough for a majority of people out there.”

That’s because THC has a vastly different action in the body—or pharmacokinetics—than ethanol, the active ingredient in alcoholic drinks. Ethanol is unique in that its pharmacokinetics are rather linear—it is a nervous system depressant, and the more you drink the drunker you consistently get, leading to the widely accepted limit of about 0.08 blood alcohol content for driving. By contrast, THC is a nervous system modulator. It mimics the endogenous neurotransmitter anandamide, altering nerve cell signaling in ways for which experienced users compensate.

While drunk drivers tend to be over-confident, stoned drivers are the opposite, Morrow says. Some of them flat-out refuse to drive. The NHTSA fact sheet states that “Marijuana has been shown to impair performance on driving simulator tasks on open and closed driving courses for up to three hours.” But the agency has also noted that “some [THC-intoxicated] drivers may actually be able to improve performance for brief periods by overcompensating for self-perceived impairment.” One study pegged the crash risk of stoned driving at a level of 2, versus 1 for a sober driver, 7 for drunk driving—and 23 for driving while texting.

The Hound should be ready for sales to police departments by late next year at a target cost of $500 to $1,000 per unit, Lynn says, adding that he is confident there is a correlation between impairment and active THC in breath. Planned tests on human subjects, however, have been delayed while Hound Labs refines the device. “I hope we will make a difference,” he says.