by Jacob Sullum
Last year, during a congressional hearing on the threat posed by stoned drivers, a representative of the National Highway Traffic Safety Administration (NHTSA) was asked how many crash fatalities are caused by marijuana each year. “That’s difficult to say,” replied Jeff Michael, NHTSA’s associate administrator for research and program development. “We don’t have a precise estimate.” The most he was willing to affirm was that the number is “probably not” zero.
Michael knows something that grandstanding politicians and anti-pot activists either do not understand or refuse to acknowledge: Although experiments show that marijuana impairs driving ability, the effects are not nearly as dramatic as those seen with alcohol, and measuring the real-world consequences has proven very difficult, as demonstrated by a landmark study that NHTSA released last Friday. In “the first large-scale [crash risk] study in the United States to include drugs other than alcohol,” NHTSA found that, once the data were adjusted for confounding variables, cannabis consumption was not associated with an increased probability of getting into an accident.
Some news outlets accurately reported that result, and some did not, apparently because some reporters actually read the study, while others were content to skim NHTSA’s press release. Such carelessness misleads policy makers who are grappling with the issue of how to determine when people are too stoned to drive. It also aids pot prohibitionists, who cite the prospect of more blood on the highways as an important reason to resist legalization.
The NHTSA study included more than 3,000 drivers who were involved in crashes during a 20-month period in Virginia Beach, Virginia, plus 6,000 controls who drove in the same area during the same period but did not get into accidents. As usual, the study found that alcohol use was strongly correlated with crash risk. After adjustment for confounding, the crash risk for drivers with a blood-alcohol content (BAC) of 0.08 percent was twice the crash risk for sober drivers; it was six times as high for drivers with a BAC of 0.10 percent and 12 times as high at a BAC of 0.15 percent. But the picture for marijuana was quite different.
Over all, drivers who tested positive for active THC were 25 percent more likely to be involved in crashes. But once the researchers took sex, age, and race/ethnicity into account, the risk ratio shrank from 1.25 to 1.05 and was no longer statistically significant:
This analysis shows that the significant increased risk of crash involvement associated with THC and illegal drugs…is not found after adjusting for these demographic variables. This finding suggests that these demographic variables may have co-varied with drug use and accounted for most of the increased crash risk. For example, if the THC-positive drivers were predominantly young males, their apparent crash risk may have been related to age and gender rather than use of THC.
Further adjusting for alcohol consumption made the crash risk of cannabis consumers equal to that of drivers who tested negative for alcohol and all other drugs. In other words, the analysis, which NHTSA described as “the most precisely controlled study of its kind yet conducted,” provides no evidence that marijuana use increases crash risk. That result, the authors note, is similar to what the best-designed previous studies have found: a small or nonexistent increase in crash risk.
Several reporters understood this crucial point and communicated it to their readers. In a story headlined “Feds: No Link Between Pot and Car Crashes,” The Hill’s Jesse Byrnes reported that “marijuana use has not been found to increase the risk of car crashes, according to a new federal report.” Under the headline “U.S.: Pot Use Doesn’t Increase Crash Risk,” David Shepardson of The Detroit News reported that “a government study released late Friday found no evidence that marijuana use leads to a higher risk of getting into a traffic crash.” CBS News, Huffington Post reporter Matt Ferner, and Washington Post drug policy blogger Christopher Ingraham correctly noted the uncertainty about marijuana’s impact on highway safety, emphasizing that alcohol poses a much clearer and more serious risk.
By contrast, USA Today reporter Chris Woodyard summarized the Virginia Beach study this way:
[The study] showed that marijuana users are more likely to be involved in crashes. But it also points out that marijuana is smoked mostly by young men, the group with the highest propensity for accidents anyway.
That is a misleading way of describing the results, since controlling for demographic variables and alcohol use eliminated the association between marijuana and crash risk. Compounding the misimpression left by this gloss, Woodyard played up another study that NHTSA released last week: the National Roadside Survey of Alcohol and Drug Use by Drivers, which found that the share of weekend nighttime drivers who tested positive for alcohol fell from 12.4 percent in 2007 to 8.3 percent in 2013–14, while the share who tested positive for other drugs rose from 16.3 percent to 20 percent. Marijuana saw the biggest increase, from 8.6 percent to 12.6 percent.
Here is how the headline over Woodyard’s story summed up those results: “Study Finds New Driving Threat From Dopers, Druggies.” Since the case-control study showed (once confounders were taken into account) that “dopers” and “druggies” were no more likely to get into crashes than abstainers, the nature of the threat trumpeted by USA Today is rather hazy.
Washington Post reporter Ashley Halsey III made a similar error in a story headlined “Fewer People Driving Drunk, but Drug Use on the Road Is Rising”:
There are fewer drunk drivers on the road, but their place has been taken by people high on marijuana and prescription drugs….NHTSA conducted a second study to determine whether smoking marijuana increased the risk of crashes. They found that it did but, adding a caveat, said that pot smoking is most common among a group already at high risk for crashes: young men.
It is simply incorrect to say NHTSA’s case-control study found that “smoking marijuana increased the risk of crashes,” since the study found no such thing. Furthermore, it’s a mistake to describe the drivers who tested positive for any amount of THC as “people high on marijuana.” As NHTSA notes, “Drug tests do not necessarily indicate current impairment,” since “detectable blood levels may persist beyond the impairing effects.” And “whereas the impairment effects for various concentration levels of alcohol in the blood or breath are well understood, there is little evidence available to link concentrations of other drugs to driver performance.” Hence “specific drug concentration levels cannot be reliably equated with a specific degree of driver impairment.”
That observation is obviously relevant to the debate over how to define driving under the influence of marijuana. Colorado and Washington, the first two states to legalize marijuana for recreational use, settled on an arbitrary standard (five nanograms of THC per milliliter of blood) that criminalizes driving by many regular cannabis consumers, even when they are not impaired. Other states have gone further in the wrong direction, adopting a completely unscientific “zero tolerance” rule that effectively makes it illegal for pot smokers to drive, whether or not they are intoxicated.
The mistaken equation of “marijuana-positive” drivers with stoned drivers also plays a conspicuous role in the debate over legalization. A study reported in the journal Drug and Alcohol Dependence last year, for instance, found that “the proportion of marijuana-positive drivers involved in fatal motor vehicle crashes in Colorado has increased dramatically since the commercialization of medical marijuana in the middle of 2009.” Prohibitionists like former drug czar Bill Bennett and anti-pot activist Kevin Sabet latch onto such findings as evidence that legalization will result in more traffic fatalities. But as the authors of the Colorado study explained, “this study cannot determine cause and effect relationships, such as whether marijuana-positive drivers contributed to or caused the fatal motor vehicle crashes.” Since traces of marijuana can be detected in blood or urine long after the drug’s effects wear off, they added, “The primary result of this study may simply reflect a general increase in marijuana use during this same time period in Colorado.”
The fact that drivers commonly identified as “drugged” may actually be sober (along with marijuana’s relatively small impact on driving ability) could help explain why it is so hard to measure marijuana’s contribution to traffic accidents. If many or most of the drivers who test positive for THC (as in the Virginia Beach study) or marijuana metabolites (as in the Colorado study) are not in fact impaired, they may conceal the impact of those who are. “While the findings of this case control study were equivocal with regard to the crash risk associated with drug use by drivers,” NHTSA warns, “these results do not indicate that drug use by drivers is risk-free.” In other words, the fact that THC-positive drivers are no more accident prone, as a group, than THC-negative drivers (after adjustment for confounding) does not mean it is smart or safe to get behind the wheel when you are actually stoned.
At the same time, the differences between the way alcohol affects driving and the way marijuana does, confirmed once again by the Virginia Beach study, suggest that the concerns voiced by Bennett and Sabet may be entirely misplaced. To the extent that more cannabis consumption is accompanied by less drinking, legalization might actually reduce traffic fatalities, as indicated by a 2014 study of accident trends in states with medical marijuana laws. That happy outcome is by no means assured, since it depends on whether marijuana and alcohol, on balance, prove to be substitutes (meaning that more consumption of one leads to less consumption of the other), as opposed to complements (meaning that consumption of the two rises in tandem). But the possibility surely is worth considering with an open mind. Demonizing “dopers” and “druggies” as deadly threats to highway safety, whether or not they drive while impaired, is hardly conducive to that sort of rational inquiry.