If you are pulled over on suspicion of drunken driving, the police officer is likely to ask you to complete three tasks: Follow a pen with your eyes while the officer moves it back and forth; get out of the car and walk nine steps, heel to toe, turn on one foot and go back; and stand on one leg for 30 seconds.
Score well on all three of these Olympic events, and there’s a very good chance that you are not drunk. This so-called standard field sobriety test has been shown to catch 88 percent of drivers under the influence of alcohol.
But it is nowhere near as good at spotting a stoned driver.
In a 2012 study published in the journal Psychopharmacology, only 30 percent of people under the influence of THC, the active ingredient in marijuana, failed the field test. And its ability to identify a stoned driver seems to depend heavily on whether the driver is accustomed to being stoned.
A 21-year-old on his first bender and a hardened alcoholic will both wobble on one foot. But the same is not necessarily true of a driver who just smoked his first joint and the stoner who is high five days a week. In another study, 50 percent of the less frequent smokers failed the field test.
As more states legalize medical and recreational marijuana, distinctions like these will grow more and more important. But science’s answers to crucial questions about driving while stoned — how dangerous it is, how to test for impairment, and how the risks compare to driving drunk — have been slow to reach the general public.
“Our goal is to put out the science and have it used for evidence-based drug policy," said Marilyn A. Huestis, a senior investigator at the National Institute on Drug Abuse. "But I think it's a mishmash."
A 2007 study found that 12 percent of the drivers randomly stopped on U.S. highways on Friday and Saturday nights had been drinking. (In return for taking part in the study, intoxicated drivers were told they would not be arrested, just taken home.)
Six percent of the drivers tested positive for marijuana — a number that is likely to go up with increased availability. Some experts and officials are concerned that the campaign against drunken driving has not gotten through to marijuana smokers.
"We've done phone surveys, and we're hearing that a lot of people think DUI laws don't apply to marijuana," said Glenn Davis, highway safety manager at the Department of Transportation in Colorado, where recreational marijuana use became legal on Jan. 1. "And there's always somebody who says, ‘I drive better while high.'"
Evidence suggests that is not the case. But it also suggests that we might not have as much to fear from stoned driving as from drunken driving. Some researchers say that limited resources are better applied to continuing to reduce drunken driving. Stoned driving, they say, is simply less dangerous.
Still, it is clear that marijuana use causes deficits that affect driving ability, Huestis said. She noted that several researchers, working independently of one another, have come up with the same estimate: a twofold increase in the risk of an accident if there is any measurable amount of THC in the bloodstream.
The estimate is based on review papers that considered the results of many individual studies. The results were often contradictory — some of the papers showed no increase in risk, or even a decrease — but the twofold estimate is widely accepted.
The estimate is low, however, compared with the dangers of drunken driving. A recent study of federal crash data found that 20-year-old drivers with a blood-alcohol content of 0.08 percent — the legal limit for driving — had an almost twentyfold increase in the risk of a fatal accident compared with sober drivers. For older adults, up to age 34, the increase was ninefold.
The study's lead author, Eduardo Romano, a senior research scientist at the Pacific Institute for Research and Evaluation, said that once he adjusted for demographics and the presence of alcohol, marijuana did not statistically increase the risk of a crash.
"Despite our results, I still think that marijuana contributes to crash risk," he said, "only that its contribution is not as important as it was expected."
The difference in risk between marijuana and alcohol can probably be explained by two things, Huestis and Romano both say. First, stoned drivers drive differently from drunken ones, and they have different deficits. Drunken drivers tend to drive faster than normal and to overestimate their skills, studies have shown; the opposite is true for stoned drivers.
"The joke with that is Cheech and Chong being arrested for doing 20 on the freeway," said Mark A.R. Kleiman, a professor of public policy at the UCLA School of Public Affairs.
Huestis also found that in laboratory studies, most people who were high could pass simple tests of memory, addition and subtraction, though they had to use more brainpower than sober people who passed the same tests. People who were drunk were much more likely to fail.
The deficits of being stoned really began to show up, she said, when people had to handle multiple tasks at once and were confronted with something unexpected.
"It's typical to see a young adolescent with three or four other kids in the car," she said of stoned driving. "He's aware he might be impaired, so he's driving carefully.
"But then he sees an old man in the middle of the street. All his senses say, ‘This guy is there but will be out of way by the time I get there.' But then the old man drops his keys and he's slower than the kid expected. By the time it takes to process a change in the situation, there's an accident."
Another factor is location. A lot of drinking is done in bars and clubs, away from home, with patrons driving to get there and then leaving by car. By contrast, marijuana smokers tend to get high at home.
There is a lot of debate about how best to prove that drivers under the influence of THC are too intoxicated to drive. Blood-alcohol content can be reliably tested on the side of the road with a Breathalyzer, and ample data link rising levels of blood alcohol to decreases in driving skills. The same is not true for marijuana.
THC levels must be measured from blood or urine samples, which are typically taken hours after an arrest. Urine tests, which look for a metabolite of THC rather than the drug itself, return a positive result days or weeks after someone has actually smoked. Yet most states have laws that equate any detectable level of THC metabolite in urine with detectable levels of actual THC in blood, and criminalize both. Only six states have set legal limits for THC concentration in the blood. In Colorado and Washington, where recreational use has been legalized, that limit is 5 nanograms per milliliter of blood, or 5 parts per billion.
The problem, Huestis said, was that studies from Europe suggested that this limit was far too high. Ninety percent of impaired-driving cases in Sweden would be missed at that level, she said.
The studies indicated that a better limit would be just 1 nanogram per milliliter, she said.