Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jan 6.
Published in final edited form as: Traffic Inj Prev. 2019 Nov 1;20(SUP2):S148–S151. doi: 10.1080/15389588.2019.1662256

Correlation of EEG Biomarkers of Cannabis with Measured Driving Impairment

T Brown 1, M McConnell 2, G Rupp 2, A Meghdadi 2, C Richard 2, R Schmitt 1, G Gaffney 3, G Milavetz 4, C Berka 2
PMCID: PMC8733968  NIHMSID: NIHMS1540971  PMID: 31674856

Abstract

Objective

To use EEG biomarkers derived from a short, easily administered neurocognitive testbed to determine acute cannabis intoxication and its effect on driving performance in a driving simulator.

Methods

The data analyzed were from a study examining the relationship between psychomotor task performance, EEG data, and driving performance in a simulator. EEG data was collected using STAT® X-24 EEG Wireless Sensor Headset, which was worn during the psychomotor and driving tasks. Driving data were collected for segments of consistent driving environments including urban driving, urban curves, interstate, interstate curves, dark rural, and rural straightaways. Dependent measures included measures of lateral and longitudinal vehicle control.

Results

There was a significant relationship between impaired driving performance as indicated by increased standard deviation of lane position and EEG power in slow theta band (3–5 Hz) in parietal and occipital areas.

Conclusions

These results, combined with our prior reported results, suggest that EEG and ECG acquired concurrent with neuropsychological tests hold potential to provide a highly sensitive, specific and dose dependent profile of cannabis intoxication and level of impairment.

INTRODUCTION

As cannabis use becomes more widely accepted socially and as legalization becomes more prevalent, there is growing interest in its effects on brain function. While some of the effects of cannabis are well-established, such as memory impairment1–4, it is unclear how cannabis use may impact daily functional activities such as driving, operating machinery and other safety-related tasks. Unfortunately, there are currently no validated methods for quantifying acute impairment from cannabis intoxication. To ensure public safety, it is essential for employers, policymakers and law enforcement officials to fully understand the effects of cannabis on all aspects of performance and to have validated methods measures to assess and quantify these effects.

Previous work5 by the University of Iowa and Advanced Brain Monitoring Inc. has demonstrated impaired performance in a driving simulator during acute cannabis intoxication. Furthermore, when acutely dosed with cannabis, subjects were shown to have a reduction in slow theta (3–5 Hz) power during resting state eyes closed, and a decrease in the amplitude of the Late Positive Potential (LPP) during a sustained attention task and verbal memory task. Additionally, across all tasks heart rate was significantly increased in the dosed condition.

The present work further elucidates these findings and attempts to establish a robust, dose-dependent neurophysiological profile of cannabis impairment obtained via a short neurocognitive testbed that is easily delivered by an iPad with concurrent EEG acquisition.

METHODS

Data was collected during a multi-visit study completed at the National Advanced Driving Simulator (NADS) at the University of Iowa to evaluate the relationship between performance of psychomotor tasks, EEG data, and driving performance. Subjects were consented and completed a screening visit to ensure eligibility. All subjects were required to be licensed drivers between the ages of 18 and 40, healthy, and drive a minimum of 5000 miles per year. Subjects were excluded if they reported taking any medications that are known to cause driving impairment. Subjects were also excluded if they used cannabis less than once in a three-month period or more than four times per week. A total of 18 people were enrolled in this study; however, two failed to meet study criteria during the screening visits, two withdrew between screening and the first dosing visit, and three people withdrew from the study due to nausea/sickness. This resulted in 11 participants who completed the full study protocol.

The study, approved by the University of Iowa’s Biomedical Institutional Review Board (FWA00003007), consisted of three visits: a screening visit and two dosing visits. At the screening visit, subjects were provided with a description of the study and provided informed consent prior to completing screening procedures. The screening involved a urine drug screen and (if female) pregnancy test, measurement of height and weight, a physical and psychological evaluation, a demographic and driving history questionnaire, viewing a study overview presentation, a simulator drive, and a wellness questionnaire. Eligible subjects were then scheduled for their dosing visits and third-party transportation was arranged. On the dosing visit days, subjects provided a urine sample to test for illicit drug use and pregnancy to verify continued eligibility. Subjects provided information about their sleep and food consumption to verify 7–9 hours of sleep the preceding night and limited caffeine intake. Upon completion, they were given a questionnaire about their current sleepiness level. Subjects were fitted with the STAT® X-24 EEG Wireless Sensor Headset (Advanced Brain Monitoring, Carlsbad, CA). Subjects were administered a single dose (500 mg) of Cannabis (6.7% THC) or Cannabis Placebo (0.09% THC) from NIDA-provided plant material via inhalation of vaporized material, then rested for 30 minutes. Subjects then completed a 70-minute Cannabis Impairment Detection Application (CIDA) that included the following tasks: Resting State Eyes Open (RSEO), Resting State Eyes Closed (RSEC), Three Choice Vigilance (3CVT), Standard Image Recognition (SIR), Verbal Memory Scan (VMS) , and a modified Trail Making Task, all with concurrent EEG and ECG acquisition6,7. The subject was then escorted to the simulator and completed a 45-minute drive. Blood draws were completed before dosing, and before and after driving. The order of the visits for dose administration were counterbalanced. The nighttime driving scenario consisted of three approximate 10-minute segments of urban, interstate and rural driving followed by a ten-minute straight rural driving environment, providing a variety of different speeds and environments 8–10. The drive focused on vehicle control; no crash imminent events were included. Three equivalent versions of the scenario were used in a counter-balanced manner to provide coverage of all three scenarios across the conditions and to minimize learning effects.

Vehicle, environment, and driver data, were collected at 60 Hz and reduced to provide summary data for each driving segment. The dependent measure in this analysis was standard deviation of lane position (SDLP). EEG and ECG data were collected at 256 Hz.

EEG data during resting state were filtered between 1 and 40 Hz. Artifact decontamination was performed using independent component analysis (ICA) and removing components labeled as artifact using EEGLab11 and IClabel12 toolboxes. Power spectral densities were computed using Fast Fourier Transform with Kaiser Window on one second windows with 50% overlap. The total power in each frequency band was computed for each epoch and were averaged across all the epochs during each session. Power-spectral density (PSD) analyses were completed for RSEO, RSEC, and the driving simulator. For event-related-potential (ERP) tasks, data were filtered between 0.1 and 50 Hz and epoched relative to stimulus presentation onset. ERP analysis pipeline included baseline adjustment (−100 ms pre stimulus onset), thresholding (+/− 100 uV), ICA artifact decontamination, and rejecting outlier epochs. Trials were averaged for each stimulus type. Late positive potential (LPP) component for each trial type at each channel was measured by averaging ERP waveform amplitude within a window of 400–800 ms post stimulus onset.

RESULTS

The analyses of the effect of acute cannabis use on driving performance and heart rate and the relationship between the EEG measures of theta power and LPP and driving performance showed significant effects. Within subject analysis showed that both heart rate and SDLP were significantly increased during active sessions (dosed) compared to placebo. On average, SDLP was increased by 6.29 cm (paired t-test, p = 0.0004, df = 9) and heart rate was increased by 7.88 beats per minute (paired t-test, p =0 .02, df = 10) as shown in Figure 1. Both EEG measures of interest (theta power in resting state and LPP component in VMS ERP task) showed a negative correlation with SDLP only during dosed sessions. Figure 2 shows theta (3–5 Hz) power at POz channel versus SDLP (Spearman’s correlation coefficient rs = −0.7, p = 0.04) in dosed session. The LPP component in VMS test versus SDLP (Spearman’s correlation coefficient rs = −0.35, p = 0.36) in dosed session is shown in Figure 3.

Figure 1.

Figure 1.

Open in a new tab

SDLP (a,b) and heart rate (c,d) were significantly higher during driving in dosed sessions compared to placebo (6.29 cm increase in SDLP) and (7.88 bpm increase)

Figure 2.

Figure 2.

Open in a new tab

Theta power in resting state EEG was inversely correlated with SDLP, only in dosed session (a) and not in placebo (b)

Figure 3.

Figure 3.

Open in a new tab

LPP measure in VMS test was inversely correlated (Spearman rank order correlation; r = −0.38, Pearson’s r = −0.35) with SDLP only in dosed session (a) and not in placebo (b). This correlation did not reach significance level.

DISCUSSION

The decriminalization of cannabis has resulted in increased use, decreased public perception of the risks of use and an alarming rise in reported driving under the influence of cannabis as well as more positive post-crash tests. Although Δ9-THC levels can be measured in blood, saliva, urine and hair, the correlations with impairment are weak and there is a growing unmet need for validated methods for quantifying acute impairment from cannabis intoxication. Our findings show several strong biomarker candidates associated with cannabis ingestion can be derived from EEG and ECG acquired during a brief neuropsychological testing protocol. Our results reveal a relationship between these biomarkers and impaired driving performance in a simulated driving scenario.

Specifically, the EEG biomarkers (theta power and amplitude of the event-related late positive potential LPP) were correlated with the standard deviation of lateral position (SDLP) a standard metric shown to be among the most important performance measures for evaluating impairment from medication use on driving performance. These data in combination with our prior reported results suggest that EEG and ECG acquired concurrent with neuropsychological tests delivered using an iPad hold the potential to provide a highly sensitive, specific and dose dependent profile of cannabis intoxication and level of impairment.

ACKNOWLEDGEMENTS

This research was funded by the National Institutes of Drug Abuse (NIDA) under contract HHSN271201800015C/N43DA-18-1218. This manuscript reflects the findings of the authors and not necessarily NIDA.

REFERENCES

  • 1.Gruber SA, Sagar KA, Dahlgren MK, Racine M, Lukas SEJPoAB Age of onset of marijuana use and executive function. 2012;26(3):496. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Gruber SA, Sagar KA, Dahlgren MK, Racine MT, Smith RT, Lukas SEJFip Splendor in the grass? A pilot study assessing the impact of medical marijuana on executive function. 2016;7:355. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Ramaekers JG, Kauert G, van Ruitenbeek P, Theunissen EL, Schneider E, Moeller MRJN. High-potency marijuana impairs executive function and inhibitory motor control. 2006;31(10):2296. [DOI] [PubMed] [Google Scholar]
  • 4.Wesnes K, Annas P, Edgar C, et al. Nabilone produces marked impairments to cognitive function and changes in subjective state in healthy volunteers. 2010;24(11):1659–1669. [DOI] [PubMed] [Google Scholar]
  • 5.Smith S, Stone B, Meghdadi A, Spurgin A, Brown T, Berka C. Neurophysiological Indices of Cannabis and Impairment. Society for Neuroscience; 3–7 November, 2018; SanDiego, CA. [Google Scholar]
  • 6.Berka C, Meghdadi A, Smith S, et al. Mapping the trajectory of neurodegenerative disease with neurophysiological biomarkers acquired during sleep and waking. Paper presented at: The 14th International Conference on Alzheimer’s & Parkinson’s Diseases; 26 March, 2019; Lisbon, Portugal. [Google Scholar]
  • 7.Waninger S, Berka C, Meghdadi A, et al. Event-related potentials during sustained attention and memory tasks: Utility as biomarkers for mild cognitive impairment. Alzheimer’s Dementia: Diagnosis, Assessment Disease Monitoring. 2018;10:452–460. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Brown TL, Milavetz G, Gaffney G, Spurgin A. Evaluating drugged driving: Effects of exemplar pain and anxiety medications. Traffic Injury Prevention. 2018;19(sup1):S97–S103. [DOI] [PubMed] [Google Scholar]
  • 9.Brown T, Spurgin A, Milavetz G, Gaffney G, Johnson R. Do Drowsy Driver Drugs Differ? 8th International Driving Symposium on Human Factors in Driver Assessment, Training, and Vehicle Design; June 22–25, 2015, 2015; Park City, UT. [Google Scholar]
  • 10.McDonald AD, Schwarz C, Lee JD, Brown TL. Real-Time Detection of Drowsiness Related Lane Departures using Steering Wheel Angle. Paper presented at: Human Factors and Ergonomics Society 56th Annual Meeting2012; Boston, MA. [Google Scholar]
  • 11.Delorme A, Makeig SJJonm. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. 2004;134(1):9–21. [DOI] [PubMed] [Google Scholar]
  • 12.Pion-Tonachini L, Kreutz-Delgado K, Makeig S. ICLabel: An automated electroencephalographic independent component classifier, dataset, and website. arXiv preprint. 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES