This originally appeared at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3163816

Mark A.R. Kleiman BOTEC Analysis, LLC; New York University Marron Institute of Urban Management

Tyler Jones Jones BOTEC Analysis, LLC

Celeste Miller BOTEC Analysis, LLC

Ross Halperin New York University Marron Institute of Urban Management

Date Written: June 29, 2018

Abstract
THC is the most commonly detected intoxicant in US drivers, with approximately
13% of drivers testing positive for marijuana use, compared to the 8% that show a
measurable amount of alcohol (NHTSA, 2015). Because cannabis use remains
detectable for much longer than alcohol, and also for long after the driver is no longer
impaired, the difference in rates does not show that stoned driving is more common
than drunk driving. Nonetheless, cannabis intoxication while driving is on the rise
and has been shown to impair reaction time and visual-spatial judgment.
Many states, including those where cannabis sales are now permitted by state law,
have laws against cannabis-impaired driving based on the drunk-driving model,
defining criminally intoxicated driving as driving with more than a threshold amount
of intoxicant in one’s bloodstream—a per se standard—as opposed to actual
impairment. That approach neglects crucial differences between alcohol and
cannabis in their detectability, their pharmacokinetics, and their impact on highway
safety.
Cannabis intoxication is more difficult to reliably detect chemically than alcohol
intoxication. A breath alcohol test is (1) cheap and reliable; (2) sufficiently simple
and non-invasive to administer at the roadside; and (3) a good proxy for alcohol in
the brain, which in turn is (4) a good proxy for subjective intoxication and for
measurable driving impairment. In addition, (5) the dose-effect curve linking blood
alcohol to fatality risk is well-established and steep.
None of those things is true for cannabis. A breath test remains to be developed. Oral-
fluid testing can demonstrate recent use but not the level of impairment. A blood test
requires a trained phlebotomist and therefore a trip to a medical facility, and blood
THC levels drop very sharply over time-periods measured in minutes. Blood THC is
not a good proxy either for recency of use or for impairment, and the dose-effect
curve for fatality risk remains a matter of sharp controversy. The maximum risk for
cannabis intoxication alone, unmixed with alcohol or other drugs, appears to be more
comparable to risks such as talking on a hands-free cellphone (legal in all states) than
to driving with a BAC above 0.08, let alone the rapidly-rising risks at higher BACs.
Moreover, the lipid-solubility of THC means that a frequent cannabis user will always
have measurable THC in his or her blood, even when that person has not used
recently and is neither subjectively intoxicated nor objectively impaired. That
suggests criminalizing only combination use, while treating driving under the
influence of cannabis (however this is to be proven) as a traffic offense, like speeding.