Measuring drug use among drivers: How accurate is self-reported use?

February 2020

Objective: There has been considerable work validating self-reported drug use with bioassays, but these studies have not been conducted with respondents on the roadway. The objective of this study was to assess the validity of drivers’ self-reported drug use. Method: This study analyzed 2007 and 2013–14 data from nationally representative roadside surveys of U.S. drivers. Using phi coefficients, the study examined the association between self-reported, past 24-hour drug use and bioassays (based on oral fluid and/or blood) for cannabis, opioids, cocaine, antidepressants, and benzodiazepines. Logistic regression models examined the association of various factors with the odds of reporting specific drug types. Results: Overall phi coefficients by drug type ranged from 0.17 to 0.34 in 2007 and 0.30 to 0.54 in 2013–14. The odds of reporting cannabis, cocaine, and benzodiazepine use were significantly higher among drivers who used these substances in 2013–14 (OR=1.92, OR=2.97, and OR=3.25, respectively), compared with 2007. Opioid and antidepressant reporting did not differ significantly among users of these substances across survey years (OR=1.01 and OR=1.44, respectively). Conclusions: Roadside surveys of drivers are an important method for gathering the types of data needed to monitor trends in drug use. The findings reinforce the importance of gathering multiple types of data to understand drug use among drivers. The many factors that influence self-reporting, as well as the limitations of biological measures, both pose challenges to accurately measuring drug use. Future studies should investigate ways to improve measurement of drug use in this population.