Cannabis is in the spotlight in the United States due to increased levels of acceptance of its use for medical treatment, and for recreational use. Increasingly, states are proposing changes to their laws through legislative action or voter initiative to decriminalize and legalize its use. One of the major concerns shared by both opponents and proponents of greater access to cannabis is its impact on driver performance and relationship to adverse effects on traffic safety. While the exact relationship between cannabis use and increased risk for crash involvement remains unclear, both sides recognize that the cognitive and psychomotor effects of cannabis use in the period immediately after use can impact vehicle control and judgment and present some risk for deterioration in driving performance. These concerns have led to a strong desire among lawmakers and traffic safety advocates to consider laws that criminalize cannabis-involved driving including laws that set a quantitative threshold for concentration of delta-9-tetrahydrocannabinol (THC), the primary active component of cannabis, in a person’s blood. This threshold would constitute an offense per se in an effort to discourage cannabis-impaired driving. What that threshold should be is a subject of much debate, and this study was undertaken to determine whether data from the Drug Recognition Expert (DRE) program consisting of physiological indicators of drug use, and performance in roadside cognitive and psychomotor tests, supported any particular quantitative threshold for a per se law for THC. Data from two sources were evaluated: 602 drivers arrested for impaired driving in which only THC was present, along with a sample of 349 drug-free controls, in which full records of the subjects’ performance in the DRE exam were available; and 4,799 drivers arrested for impaired driving who tested positive for one or more cannabinoids (THC, hydroxy-THC, and carboxy-THC), and for which demographic information and comprehensive toxicology testing results were available. Evaluation of indicators from the DRE arrestees compared to drug free controls indicated poorer performance in the psychophysical tests for impairment (walk-and-turn test, oneleg-stand test, and finger-to-nose test). On the walk-and-turn test, 55.5 percent of drug free subjects were able to complete the test without errors while only 6.0 percent of the cannabis-positive subjects were able to do so. In the one-leg-stand test, 67.2 percent of drugfree subjects were able to complete the test with no errors, while only 24 percent of the cannabis-positive drivers were able to do so. On the finger-to-nose test 49.2 percent of the drug-free subjects performed the test without errors, compared to only 5.2 percent of the cannabis-positive subjects. Indicators of red, bloodshot and watery eyes, eyelid tremor, lack of convergence and rebound dilation all showed significantly greater (p<0.001) incidence in the cannabis-positive subjects. Cannabis-positive subjects were also more likely to have higher systolic blood pressure and higher pulse rates. Having established differences in these parameters between cannabis-positive and negative subjects, we evaluated the relationship between blood THC concentration and performance on tests for impairment. We performed a bivariate correlation analysis of the indicators as a function of blood THC concentration. Neither the walk-and-turn, nor one-leg-stand tests showed increasing rates of error as a function of THC concentration across the range 1 to 47 ng/mL. Only the finger-to-nose test showed that subjects with higher THC concentrations made a greater number of misses than the subjects with lower THC concentrations. A chi-squared analysis of the same data was conducted considering whether indicators of impairment differed between subjects with blood THC concentrations above or below 5 ng/mL, the threshold for per se driving under the influence of cannabis adopted in Colorado, Washington, and Montana. No differences were found in performance in the walk-and-turn, or one-leg-stand tests, according to whether subjects were in the higher (>5 ng/mL), or lower (<5 ng/mL) THC groups. The number of misses on the finger-to-nose test was higher in the elevated THC group. We evaluated through logistic regression analysis whether the physiological, cognitive and psychomotor indicators from the DRE exam could predict THC concentration above or below a 5 ng/mL threshold and they could not. Additionally, assuming the validity of a 5 ng/mL threshold as defining impaired versus non-impaired subjects, we evaluated whether performance on any of the physiological, cognitive or psychomotor indicators correctly assigned the subject to the impaired or non-impaired group. None of the indicators met the 80 percent sensitivity threshold for correctly predicting impairment status. Analysis of the sensitivity, specificity, and accuracy of various THC concentration threshold suggested the concentration threshold associated with the best sensitivity (80.4%) and accuracy (77.0%) was 1 ng/mL, which also had the lowest specificity (70.2%). Higher THC concentration values reduced sensitivity but increased specificity. The distribution of THC concentrations in this large arrest population (4,799 subjects), indicated a median THC concentration of 4.0 ng/mL, which was telling in itself indicating that 50 percent of these subjects placed under arrest based on evidence of suspected impaired driving had blood THC concentrations of 4.0 ng/mL or below, significantly below the proposed or enacted THC per se threshold in some states. The population showed considerable combined alcohol and other drug and cannabis use, with only 23 percent of these DUI drivers being positive only for cannabinoids. Alcohol was present in 59 percent, and other drugs in 33 percent, of these cannabinoid-positive subjects. Of the subjects positive only for cannabis (N=1,117), the median THC concentration was 7.8 ng/mL, and the mean was 5.6 ng/mL. Applying different proposed per se thresholds to this group of drivers positive only for cannabis, 49 percent of drivers would be 5 ng/mL or greater, while 79 percent would be 2 ng/mL or greater, and 91 percent would be 1 ng/mL or greater. Considering the larger population of all subjects arrested for DUI with evidence of cannabis use, only 30 percent would have THC concentrations above a 5 ng/mL threshold. There is no evidence from the data collected, particularly from the subjects assessed through the DRE exam, that any objective threshold exists that established impairment, based on THC concentrations measured in specimens collected from cannabis-positive subjects placed under arrest for impaired driving. An association between the presence and degree of indicators of impairment or effect from cannabis use were evident when comparing data from cannabis-positive and cannabis-negative subjects. However, when examining differences in performance in these parameters between subjects with high (>5 ng/mL) and low (<5 ng/mL) THC concentrations, minimal differences were found. There was no correlation between blood THC concentration and scores on the individual indicators, and performance on the indicators could not reliably assign a subject to the high or low blood THC categories. Analysis of the sensitivity, specificity, and accuracy of various per se thresholds suggested the highest sensitivity was found at 1 ng/mL: 80 percent of drivers who demonstrated impairment on the SFST had THC concentrations of 1 ng/mL or greater. However, 30 percent of drivers who did not demonstrate impairment on the SFST also had THC concentrations of 1 ng/mL or greater. Finally, among both samples of drivers who came into contact with law enforcement and were subsequently placed under arrest for DUI, only 30-49 percent would have been considered impaired under a per se standard set at 5 ng/mL, depending on whether alcohol or other drugs are detected and taken into consideration. Based on this analysis, a quantitative threshold for per se laws for THC following cannabis use cannot be scientifically supported. (Author/publisher)
Abstract