Among Drug Recognition Expert evaluation tests, the finger-to-nose test with 3+ misses was the single best predictor of cannabis impairment (87%+ accuracy), and combining four tests achieved over 96% accuracy.
Read this if you want to understand how cannabis impairment is detected in roadside evaluations and why blood THC levels may not be the best measure.
Combined behavioral tests achieved 96.7%+ accuracy for cannabis impairment
What the researchers found
Law enforcement uses the Drug Evaluation and Classification Program (DECP) to assess suspected drug-impaired drivers, but the reliability of specific tests for detecting cannabis impairment has been uncertain. This study compared 302 toxicologically confirmed cannabis-impaired drivers against 302 non-impaired controls.
The finger-to-nose test stood out as the single best predictor, with sensitivity, specificity, positive predictive value, and negative predictive value all above 87% when using 3 or more misses as the criterion. Eyelid tremors during the Modified Romberg Balance test also performed well (all metrics above 86%).
Combining at least 2 of 4 indicators (finger-to-nose misses, eyelid tremors, one-leg-stand clues, walk-and-turn clues) produced the strongest results, with all diagnostic characteristics above 96.7%.
Notably, the commonly debated 5 ug/L blood THC cutoff showed limited relevance. There were no significant differences in impairment indicators between drivers above and below this threshold.
Why it matters
Unlike alcohol, there is no agreed-upon standard for cannabis impairment testing. This study provides the strongest evidence to date for which specific roadside tests reliably identify cannabis impairment, and it challenges the usefulness of blood THC concentration cutoffs.
The numbers in context
Finger-to-nose test (3+ misses): all diagnostic metrics 87.1%+. Eyelid tremors: all metrics 86.1%+. Combined 2-of-4 criteria: all metrics 96.7%+. 302 confirmed cases vs. 302 controls. 5 ug/L THC cutoff showed no significant diagnostic value.
How the study worked
Retrospective analysis of 302 toxicologically confirmed cannabis-only DECP cases (blood THC 1+ ug/L) compared to 302 normative non-impaired controls. Evaluated pulse, blood pressure, pupil size, and performance on four psychophysical tests.
What this study cannot tell us
Cases were selected where Drug Recognition Experts successfully identified cannabis, potentially excluding cases where cannabis impairment was missed. The comparison to normative data rather than a matched control group introduces potential confounds. Polydrug cases were excluded.
How to read the evidence
Large sample with toxicological confirmation and systematic analysis of multiple diagnostic characteristics, though case selection may introduce bias.
When this study was published
Published in 2016. Cannabis potency and product diversity have increased, which may affect impairment patterns.
The bigger picture
As cannabis legalization expands, law enforcement needs validated tools for identifying impaired drivers. This study strengthens the evidence base for behavioral assessment over blood THC levels, which decline rapidly and correlate poorly with impairment.
Questions still open
- Would these tests perform as well in the field as in this analysis? How should these findings influence per se THC driving laws? Could automated or technology-assisted versions of these tests improve consistency?
Common questions
What is the best roadside test for cannabis impairment?
Does blood THC level predict impairment?
Read the original research
Drug Recognition Expert (DRE) examination characteristics of cannabis impairment.
Accident; analysis and prevention, 92, 219-29
Citation
Hartman, Rebecca L; Richman, Jack E; Hayes, Charles E; Huestis, Marilyn A. (2016). Drug Recognition Expert (DRE) examination characteristics of cannabis impairment.. Accident; analysis and prevention, 92, 219-29. https://doi.org/10.1016/j.aap.2016.04.012