08/07/2026
The office cannabis policy just got 5 new findings. One exam takes a penlight. One takes a question.
Convergence, rebound dilation, nystagmus, finger to nose, and the teach-back.
Here is why more findings is the whole point. Only 4 of the 5 appear in the numbers below, because nystagmus and the teach-back have no sensitivity figure in the literature.
In the roadside drug recognition data, rebound dilation on its own runs 70.9% sensitivity. Convergence on its own, 78.8%. Finger to nose on its own, 87.1% or better on every measure. Either eye test positive, 92.7%.
The combination outperformed every single test. So a chairside decision gets better as independent findings accumulate, and the policy now carries more of them.
THE EYE EXAM. 25 seconds, a penlight.
Convergence. Pen tip slowly toward the bridge of the nose. Normal is both eyes tracking inward and crossing, holding to about 2 inches. Abnormal is one or both breaking outward before that. 78.8% of cannabis cases, 10.9% of controls.
Rebound dilation. Penlight in the eye, watch 15 seconds. Normal is the pupil constricting and staying constricted. Abnormal is a brief constriction followed by a steady re-opening that never comes back. 70.9% of cases, 1.0% of controls.
Nystagmus, which runs the other way. Cannabis doesn't cause it. Vertical gaze nystagmus was found in 0 of 302 cannabis cases. So if it's there, something else is on board, and that matters in the minute before a patient gets nitrous.
THE COGNITIVE ASSESSMENT.
Finger to nose. Eyes closed, head back, arms out. 6 trials, record the misses. 3 or more out of 6 ran 87.1% or better on every measure.
The teach-back. "Tell me back in your own words what we're doing today, and name one thing that could go wrong." Every other finding tells you a drug is probably on board. The teach-back tells you whether the patient can understand, appreciate, reason, and choose, which is the accepted standard for capacity and the actual decision in front of you.
WHAT NONE OF IT DOES.
A positive eye finding suggests a drug is on board. It does not establish impairment and it does not establish inability to consent. More elements make the pattern defensible. They don't make any single one decisive.
And the source data has a flaw worth naming. Those 302 controls were police officers and academy cadets. In a placebo-controlled trial in JAMA Psychiatry, certified drug recognition instructors classified 49.2% of the sober placebo group as impaired.
So the screen tells you how carefully to proceed and whether consent holds. It never tells you a patient is high. Chart the findings, not the conclusion.
The policy template and the one-page chairside algorithm are rebuilt and free: https://ce-dojo.com/cannabis-office-policy-five-new-findings/