The Autopsy / Case File · Vehicular & Transit
Positive, Not Impaired
Every fatal-crash toxicology report that comes back positive for THC gets filed next to reports that come back positive for alcohol, as if the two tests measure the same thing. They don't. One clears the body in hours. The other can sit detectable in blood or urine for days or weeks after the high is gone — which means the clearest number this data can produce, a rising share of THC-positive drivers, is also the hardest number to actually interpret.
NHTSA's National Roadside Survey found nighttime drivers testing positive for THC rose from roughly 8.6 percent in 2007 to 12.6 percent in 2013-2014, with 20 percent of surveyed drivers positive for some potentially impairing drug by the later survey. Washington State's own three-wave roadside study around its 2014 legalization of recreational sales found daytime THC-positive drivers climb from 7.8 percent before legal sales began to 18.9 percent a year later — and fatal-crash toxicology data from the same state showed THC-positive drivers in fatal crashes rising from 8.8 percent pre-legalization to 21.7 percent by the fifth year after. What none of that data resolves is whether those drivers were actually impaired when the crash happened, a question the underlying crash-risk research answers inconsistently, ranging from a near-doubling of risk to no measurable increase at all once other factors are accounted for.
Join the discussion →
The country's traffic-safety system built its entire approach to impaired driving around one substance: alcohol, whose relationship between blood concentration and actual driving impairment is well established, consistent, and time-limited enough that a breathalyzer reading taken at the roadside is a reasonably direct proxy for how impaired a driver was minutes earlier. Cannabis doesn't behave the same way pharmacologically, and the country's drug-impaired-driving data was built, in large part, before most states had legalized it — leaving a widening gap between how often drivers now test positive for THC and how confidently anyone can say what that positive test actually means for crash risk.
What the National Roadside Surveys Found
NHTSA has periodically conducted National Roadside Surveys since the 1970s, stopping randomly selected drivers at roadside locations, mostly on weekend nights, and asking them to voluntarily provide breath, saliva, or blood samples in exchange for anonymity and, often, a small payment. The 2007 survey found 8.6 to 8.7 percent of nighttime drivers tested positive for THC. By the 2013-2014 survey — the most recent full national roadside survey completed, now more than a decade old — that figure had risen to 12.6 to 12.7 percent, and 20 percent of surveyed drivers tested positive for some drug with potential impairing effects. NHTSA's own reporting on that data is careful on a specific point: the survey measured presence, not concentration in a way that would support impairment conclusions, and did not attempt to evaluate whether any individual driver was actually impaired at the time of testing. It also didn't break results out by state, limiting its usefulness for measuring the effect of any one state's legalization specifically — which is exactly the gap Washington State's own follow-up research was designed to fill.
Washington's Before-and-After
Washington ran its own three-wave roadside survey timed specifically around the state's July 2014 launch of legal recreational marijuana sales: one wave immediately before legal sales began, a second six months after, and a third one year after, with data collected during both daytime and nighttime sessions at the same six sites each time. The daytime results showed the sharpest, most consistent movement: 7.8 percent of drivers tested THC-positive in Wave 1, rising to a statistically significant 18.4 percent in Wave 2 and 18.9 percent in Wave 3 — better than a doubling within six months that held steady a year out. A separate line of research using Washington's fatal-crash toxicology records found a similar trajectory playing out among drivers actually involved in fatal collisions: roughly 8.8 percent tested THC-positive before legalization, rising to about 18 percent in the first one to two years after, and reaching 21.7 percent by the fifth year post-legalization.
What a Positive Test Doesn't Mean
The prevalence numbers above are among the most solid figures in this entire research area, because they measure something straightforward: how many drivers, tested at a specific moment, had detectable THC in their system. What they cannot do is establish how many of those drivers were actually impaired while driving, because THC's pharmacology breaks the alcohol-era assumption that a positive test at the roadside reflects recent, active impairment. Alcohol clears the bloodstream at a predictable rate over hours; THC, particularly in frequent users, can remain detectable in blood for days and in urine for weeks after any intoxicating effect has worn off completely. A driver who used cannabis five days before a crash and a driver who used it five minutes before a crash can both test positive, and the roadside or post-crash toxicology screen alone cannot reliably distinguish between them.
A driver impaired five minutes ago and a driver who hasn't been impaired in five days can produce the identical positive test — and the crash-risk research built on top of that test disagrees with itself as a result.
That ambiguity shows up directly in the crash-risk literature itself, which has produced a genuinely inconsistent set of findings rather than a converging consensus. Some analyses have estimated marijuana-positive drivers face roughly 1.83 times the crash risk of an unimpaired driver. Others, reviewing the same broader body of evidence, have found estimated increased risk ranging anywhere from zero to three times that of an unimpaired driver depending on the study design and how well it controlled for other factors. And at least one large NHTSA-funded case-control study — using drivers involved in crashes matched against drivers who passed through the same location without crashing — found no statistically significant association between testing positive for marijuana and crash risk once age, sex, and alcohol use were accounted for as confounding variables. For comparison, the same body of research finds a driver at the 0.08 percent blood alcohol concentration legal limit is five to twenty times more likely to be involved in a crash than a sober driver — a gap in certainty between the two substances that the traffic-safety field has not closed, even as cannabis legalization has spread to most of the country.
Polysubstance use complicates the picture further. Alcohol remains the single substance with the strongest, most consistently replicated relationship to crash risk in this research, and a meaningful share of drivers who test positive for THC also test positive for alcohol at the same roadside stop or crash scene. The National Academies of Sciences, Engineering, and Medicine has specifically flagged confounding and effect modification with alcohol as a factor the field needs to account for more carefully before drawing firm conclusions about cannabis alone — because a crash involving a driver who used both substances together can get attributed entirely to whichever one the testing protocol happened to screen for first, or entirely to cannabis in a report that undercounts the alcohol involved. That's a distinct problem from the detection-window issue above: even a study design sophisticated enough to separate "recently used" cannabis from "used days ago" cannabis still has to separate cannabis's contribution to a crash from alcohol's, in cases where both were present, and the current research base is not yet built to do that cleanly at scale.
What Insurance Data Adds
A separate strand of evidence comes from IIHS-affiliated research using insurance claim frequency rather than toxicology, comparing states before and after legalization against neighboring states that hadn't legalized. That research found increased collision claim frequency in two of the three original legal-cannabis states studied, Colorado and Washington, with Oregon as the exception, alongside separately estimated increases in overall traffic fatalities equivalent to roughly one additional fatality per million residents in the years immediately following legalization — a real but modest effect that some subsequent analyses found receding somewhat over a longer trend line. That pattern, an initial measurable increase in crash-adjacent outcomes that isn't obviously explained away, sits alongside the toxicology ambiguity above rather than resolving it: it's consistent with cannabis legalization genuinely increasing crash risk to some degree, but the insurance data alone can't isolate how much of that increase runs through actual driving impairment versus other behavioral changes tied to legalization, like changes in overall driving patterns or alcohol-cannabis co-use.
A Fourteen-Year-Old National Baseline
The country is still, as of this writing, relying primarily on a national roadside survey conducted in 2013-2014 — before recreational cannabis was legal in the large majority of states that have since legalized it — as its baseline for national THC-positive driving prevalence. NHTSA has a new National Roadside Survey planned, the first full comparable survey since that 2013-2014 round, which would give researchers a current national prevalence figure to set against a legalization landscape that has changed dramatically in the intervening years. Until that data lands, the country's best national-level answer to "how many drivers are now driving with THC in their system" remains more than a decade old, even as state-level data like Washington's shows how much that number can move in just the first five years after a single state's legalization.
And Then You're Dead
A rising THC-positive rate is real, measured, and repeatedly confirmed across multiple independent surveys. What it adds up to — how much of it reflects a driver who was actually impaired behind the wheel, versus a driver who used cannabis several days earlier and would have tested the same way whether or not they ever got in a car — is a question the field studying it still can't answer with the confidence alcohol research reached decades ago.
That's it. That's the whole thing.
Sources
- National Roadside Survey data finding 8.6–8.7 percent of nighttime drivers THC-positive in 2007, rising to 12.6–12.7 percent in 2013-2014, with 20 percent of surveyed drivers positive for some potentially impairing drug in the later survey — NHTSA, "Results of the 2013-2014 National Roadside Survey of Alcohol and Drug Use by Drivers," DOT HS 812 118, and related NHTSA National Roadside Survey publications
- Washington State three-wave roadside survey around the state's July 2014 legalization of recreational marijuana sales, finding daytime THC-positive drivers rising from 7.8 percent (Wave 1, pre-legalization) to a statistically significant 18.4 percent (Wave 2, six months post) and 18.9 percent (Wave 3, one year post) — NHTSA, "Marijuana, Other Drugs, and Alcohol Use by Drivers in Washington State," DOT HS 812 299, and Washington Traffic Safety Commission Washington State Roadside Survey reports
- Washington fatal-crash toxicology data showing THC-positive drivers involved in fatal crashes rising from approximately 8.8 percent before legalization to approximately 18 percent within 12–24 months after and 21.7 percent by the fifth year after legalization — Washington Traffic Safety Commission, "Marijuana and Alcohol Involvement in Fatal Crashes in Washington," and related state fatal-crash toxicology analyses
- Crash-risk estimates for marijuana-positive driving ranging from an estimated 1.83 times unimpaired risk in some analyses to no statistically significant association after adjusting for age, sex, and alcohol confounding in a NHTSA-funded case-control study, compared with a 5- to 20-times increased crash risk at the 0.08 percent BAC legal limit — NHTSA, "Marijuana-Impaired Driving: A Report to Congress," DOT HS 812 440, July 2017
- IIHS-affiliated research finding increased insurance collision claim frequency in Colorado and Washington (not Oregon) following legalization, and an estimated increase in traffic fatalities of roughly one per million residents in early post-legalization years — Insurance Institute for Highway Safety / Highway Loss Data Institute legalization-effect studies
- Photo: "Sobriety checkpoint," East Haven, Connecticut, Versageek, CC BY-SA 3.0, via Wikimedia Commons
Comments
Loading…