Drivers with THC below 5 ng/mL showed no increased crash risk, while alcohol at 0.08% increased risk sixfold | Research

The takeaway

A prospective study of 3,005 injured drivers in British Columbia found no increased crash responsibility with THC below 5 ng/mL and only a non-significant increase at 5+ ng/mL (OR 1.74), compared to a sixfold increase for alcohol above 0.08%.

Traffic safety researchers, policymakers setting cannabis driving limits, and defense attorneys challenging per se THC laws.

THC <5 ng/mL: no increased risk

What the researchers found

No increased crash risk in drivers with THC <2 or 2-5 ng/mL. At THC >=5 ng/mL, adjusted OR was 1.74 (95% CI 0.59-6.36, non-significant). By contrast, BAC >=0.08% showed OR 6.00 (95% CI 3.87-9.75, p<0.01). Other recreational drugs: OR 1.82. Sedating medications: OR 1.45.

Why it matters

This is one of the most rigorous studies comparing cannabis and alcohol driving risk using prospective data and objective blood levels. The findings suggest THC-based per se driving laws (particularly at low thresholds like 2 ng/mL) may be penalizing drivers who are not actually impaired.

The numbers in context

3,005 injured drivers tested. THC detected in 8.3%. Alcohol in 14.4%. Other drugs in 8.9%. Sedating meds in 19.8%. THC <5 ng/mL: no increased risk. THC >=5 ng/mL: OR 1.74 (NS). BAC >=0.08%: OR 6.00 (p<0.01).

How the study worked

Prospective responsibility analysis of injured drivers at British Columbia trauma centres. Blood from clinical samples tested for broad-spectrum toxicology. Police reports analyzed for crash responsibility. Adjusted for age, sex, and other impairing substances.

What this study cannot tell us

Only non-fatally injured drivers included, which may underestimate risk for the most severe crashes. THC blood levels reflect timing of use but not consistent impairment levels across individuals. The 5+ ng/mL group was small, limiting statistical power.

How to read the evidence

Rated strong because this is a large prospective study with objective blood level measurements, police crash reports, and appropriate statistical adjustments.

When this study was published

Published in 2019, before many jurisdictions finalized cannabis driving laws.

The bigger picture

Cannabis driving laws in many jurisdictions use THC blood level thresholds modeled on alcohol BAC laws. This study suggests the relationship between THC blood levels and impairment is far weaker than for alcohol, with implications for how driving under the influence of cannabis should be enforced.

Questions still open

Should per se THC limits for driving be reconsidered? Is a 5 ng/mL threshold too low, too high, or the wrong metric entirely? Could performance-based impairment testing replace blood level thresholds for cannabis?

Common questions

Does cannabis impair driving?

This study found no statistically significant increase in crash responsibility at any THC blood level tested, though drivers with THC above 5 ng/mL showed a trend toward increased risk (OR 1.74). The risk was far lower than for alcohol.

How does cannabis driving risk compare to alcohol?

Alcohol above the legal limit (0.08%) increased crash responsibility sixfold (OR 6.00). The highest THC group showed only OR 1.74, which was not statistically significant.

Are current THC driving limits appropriate?

The findings suggest THC levels below 5 ng/mL are not associated with increased crash risk, raising questions about per se limits set at lower thresholds.

Read the original research

Cannabis use as a risk factor for causing motor vehicle crashes: a prospective study.

Addiction (Abingdon, England), 114(9), 1616-1626

View on PubMed ↗Publisher’s paper ↗

Citation

Brubacher, Jeffrey R; Chan, Herbert; Erdelyi, Shannon; Macdonald, Scott; Asbridge, Mark; Mann, Robert E; Eppler, Jeffrey; Lund, Adam; MacPherson, Andrew; Martz, Walter; Schreiber, William E; Brant, Rollin; Purssell, Roy A. (2019). Cannabis use as a risk factor for causing motor vehicle crashes: a prospective study.. Addiction (Abingdon, England), 114(9), 1616-1626. https://doi.org/10.1111/add.14663