Cannabis presentations to the emergency department after motor vehicle crashes in the era of legalization for recreational use

June 2021

Esther Choo Oregon Health & Science University

Daniel Nishijima University of California, Davis

Stacy Trent University of Colorado, Denver

Angela H. Eichelberger Insurance Institute for Highway Safety

Yu Ye Alcohol Research Group

Ariane Audett Karen Brasel Steve Kazmierczak Oregon Health & Science University

Cheryl J. Cherpitel Alcohol Research Group

This manuscript has been accepted by the Journal of Safety Research.

Contents

ABSTRACT ...... 3

INTRODUCTION ...... 4

METHODS ...... 6 Study design ...... 6 Study population ...... 6 Instruments ...... 7 Data analysis ...... 8

RESULTS ...... 9 Acute and alcohol use within 8 hours of injury ...... 9 Biosample sensitivity analysis ...... 10 Crash characteristics and self-reported cannabis and alcohol use within 8 hours of injury ...... 10 Biosample sensitivity analysis ...... 11 Types of cannabis use before injury ...... 11 Biosample sensitivity analysis ...... 12 Description of past-year use of cannabis, alcohol, and other drugs ...... 12

DISCUSSION ...... 13 Limitations ...... 15 Conclusions ...... 16

ACKNOWLEDGEMENT ...... 17

REFERENCES ...... 18

TABLES ...... 21

APPENDIX ...... 26

2 ABSTRACT

Introduction. There is limited research on the impact of cannabis liberalization laws on health-related behaviors and healthcare utilization. Our objectives were to examine cannabis and alcohol use among adult injured drivers presenting to emergency departments (EDs) in cannabis-legal states in order to capture an expanded profile of cannabis use in this population and evaluate differences in crash characteristics among those using cannabis alone and in combination with alcohol. Methods. This was a cross-sectional study of visits by drivers involved in motor vehicle crashes (MVCs) who presented to the ED following the event. Event-related and usual drug and alcohol use information was obtained from consenting ED patients presenting immediately after the MVC using a detailed interviewer-administered computerized questionnaire (RedCap). Blood and breathalyzer data were obtained and the electronic medical record reviewed. We examined frequency and types of acute cannabis and alcohol use, crash mechanisms and characteristics, and frequency and types of past-year use. Our primary method of determining cannabis and alcohol use was self- report; as a sensitivity analysis, we used measurements from biological samples (biosamples). Results. Overall, 8% percent of drivers reported cannabis use in the 8 hours prior to MVCs across three cannabis-legal states, either cannabis alone or in combination with alcohol; however, a higher proportion (18%) were positive by biosample criteria. There was a high level of heterogeneity in types of used, with , spliff, or use by only 24% of the population reporting cannabis use in the 8 hours before presentation and 48% of the population reporting past-year use. High-risk crash features (moderate- to high-speed collisions, rollover, vehicle ejection, intrusion, single vehicle) were common in MVCs associated with cannabis, as they were for alcohol use and co- ingestion; however, while patients injured seriously enough to require admission were less likely to report cannabis use (7% vs. 9%), they were more likely to report alcohol use (17% vs. 10%). Conclusions. Cannabis use was common among patients presenting to the ED after MVCs in this sample of cannabis-legal states. The differences between self-report and biosample data for cannabis and alcohol use in this population were significant and supports the need to use both means of assessing acute use in future studies, particularly when done in this context, where legal and medical concerns may affect reporting. However, the limitations of laboratory data when used as an objective measure must also be acknowledged. The heterogeneity of use captured supports altering our standard screening processes. Differences in reported drug patterns across sites supports the influence of contextual factors in use and health outcomes.

3 INTRODUCTION

Motor vehicle crashes (MVCs) are a common mechanism of injury and fatalities in the

United States, with more than 6.7 million crashes, 2.7 million injuries, and 36,750 deaths per

year (Insurance Institute for Highway Safety [IIHS], 2018; Thies & Register, 1993). Based on

the National Roadside Survey, cannabis is the most common recreational substance used among

drivers (National Highway Safety Traffic Administration, 2016); data from crashes suggest that

cannabis is the most common drug used in fatal MVCs, whether alone or in combination with

alcohol (Governors Highway Safety Association [GHSA], 2018). There is some evidence that when using cannabis alone, drivers may recognize and compensate for driving irregularities

(Hartman et al., 2016); however, many studies also demonstrate the detrimental effects of

cannabis on numerous parameters of driving performance (Bondallaz et al., 2016; Hartman &

Huestis, 2013; Sewell et al., 2009). Further, combining cannabis with alcohol may remove any

ability to compensate and appears to have an overall additive, if not synergistic, effect on driving

behaviors (Chihuri et al., 2017; Dubois et al., 2015; Hartman et al., 2015).

Epidemiological studies have demonstrated the risk of alcohol use with MVC occurrence

and the risk of MVC-related injuries and fatalities after the use of alcohol (Cherpitel, 2007).

Alcohol has been shown to have a dose-dependent relationship with the risk of traffic-related injury and a broad range of other injury types (Cherpitel et al., 2019; Kuendig et al., 2008;

Taylor et al., 2010). This risk is influenced by community factors, including the restrictiveness of policies around alcohol use (Cherpitel et al., 2019). This foundational research established the

basis for public health measures to affect outcomes through individual and community-level

interventions, including establishing lower legal limits for blood alcohol level in work and

recreational settings. Such studies also established the emergency department (ED) as a location

4 for screening to identify alcohol and other drug use, both for clinical care implications and to establish the role of substance use in a wide range of injury-related and noninjury-related health problems (Cherpitel, 2007; El-Guebaly et al., 1998.

However, existing literature on cannabis use alone and in combination with alcohol and the association with injury is limited (Cherpitel et al., 2017). There are little data on nonfatal crashes, and most of the published literature has been performed in experimental settings, looking at a range of poor driving behaviors, rather than actual crashes or injury. The steady increase in the liberalization of cannabis use has made this research more imperative. As of

December 2020, 15 states and the District of Columbia have legalized cannabis for recreational use (IIHS, 2020). Existing research has associated the legalization of recreational cannabis use with more ED visits (Roberts, 2019) and hospitalizations and more injured patients with concomitant cannabis use (Delling et al., 2019). There is still little research on the impact of such laws on behaviors, public health, and healthcare utilization.

Our study objectives were to examine cannabis and alcohol use among adult injured drivers presenting to EDs in cannabis-legal states. We wanted to capture an expanded profile of cannabis use in this population and to evaluate differences across demographics (including age, gender, race and ethnicity) and crash characteristics among those using cannabis alone, cannabis in combination with alcohol, and alcohol alone. We hoped to contribute to a better understanding of how changes in community use with liberalization might be reflected in the injured driver population requiring emergency care.

5 METHODS

Study design

This was a cross-sectional study of visits by drivers involved in an MVC who presented

to the ED following the event. Event-related and usual drug and alcohol use information was

obtained from consenting ED patients presenting immediately after the MVC using a detailed

interviewer-administered computerized questionnaire (RedCap) of about 30 minutes in length.

Blood and breathalyzer data were also obtained and the electronic medical record reviewed.

Study population

Research assistants recruited English-speaking patients ages 18 and older who presented

to the EDs of three hospitals, one each in Denver, Colorado; Portland, Oregon; and Sacramento,

California. Patients were recruited within 8 hours of being involved in an MVC as the driver. At

the Portland site, research assistants recruited 24 hours a day, 7 days a week from April 2017 to

November 2019; at the other two sites, recruitment occurred between 7:00 a.m. and 12:00 a.m.

from October 2018 to October 2019 (Denver) and November 2018 to November 2019

(Sacramento). Colorado legalized cannabis for recreational use in 2012, Oregon legalized in

2014, and California legalized in 2016 (IIHS, 2020). The three study sites are all urban,

academic, Level 1 trauma centers. However, they represented a range of annual patient volumes

and some differences in physical space and patient flow through the department. For example,

the Denver site has an urgent visit portion of the ED that sees lower acuity visits and was used to

recruit patients.

Excluded were prisoners and patients who were unable to consent due to nondrug-related

purposes, such as psychiatric illness. Patients who were felt to be clinically intoxicated on

presentation and thus unable to provide immediate consent were followed until (1) the patient

6 became clinically sober and regained capacity to consent for themself (based on a mini-mental exam), or (2) a legally authorized representative (LAR) provided consent for the patient. For patients unable to consent, consent via an LAR acting on the subject’s behalf was obtained.

Patients for whom an LAR could not be located were followed for the duration of their stay in the hospital and periodically assessed for capacity to consent. Participants were provided a $30 gift card as a token of appreciation for completing the study.

This study was approved by the Institutional Review Boards of the following institutions:

Oregon Health & Science University; University of California, Davis; and University of

Colorado. Given the sensitive nature of the study topics, a National Institutes of Health (NIH)

Certificate of Confidentiality was obtained.

Instruments

Blood samples for (THC) and its metabolites and a breathalyzer reading for alcohol were collected after arrival at the ED and prior to consent, for timeliness, but were discarded if no consent was obtained. Plasma samples for testing were frozen at −80o C until analysis by High Performance Liquid Chromatography/Tandem Mass

Spectrometry.

We interviewed participants about the mechanism of the MVC, drug and alcohol use

prior to the MVC, context of use, and past-year drug and alcohol use. In asking about cannabis

use, an expanded instrument was used to capture a wide variety of use beyond common forms

(like joints) that are the focus of traditional instruments but may miss the wide variety of products that have become readily available in states where cannabis is legal. For example,

asking specifically about hash reveals some cannabis use that would be missed by asking about

7 marijuana alone. We also measured cannabis dependence with the

Identification Test (CUDIT) (Adamson & Sellman, 2003).

Research assistants abstracted information from the electronic health records, including

documented crash characteristics, disposition, and measurements from biological samples

(biosamples) obtained for clinical use, using a standardized data form.

Data analysis

Cannabis and alcohol use on the day of the ED visit were defined in two ways. As with

previous studies (Asbridge et al., 2014) for the main analysis, an affirmative answer for cannabis

use within 8 hours of the MVC was considered as evidence of drug use contemporaneous with

driving, and an affirmative answer for alcohol use within 8 hours of the MVC was considered as

evidence of alcohol use contemporaneous with driving. The self-reported cannabis and alcohol

use that occurred before the MVC event (i.e., acute use) were cross-classified into four

categories: cannabis use only, alcohol use only, both cannabis and alcohol use, and use of neither. Prevalence of acute cannabis-alcohol use was examined by sociodemographic characteristics. Bivariate analysis between crash characteristics and acute cannabis-alcohol use was also performed. In addition to self-reported acute use, sensitivity analysis was conducted using a plasma blood sample with THC levels of > 0.5 ng/mL (the lowest detectable limit) and a positive 11-OH-THC (the acute metabolite of THC) result as evidence of likely acute use prior to driving (Asbridge et al., 2014). Similarly, any positive level on blood or breathalyzer testing was considered as positive for alcohol in the sensitivity analysis. Last, past-year cannabis use

(frequency and type of use), risk perception, and harmful use were examined together with past-

year use of alcohol and other drugs.

8 RESULTS

During the study period, 1,867 patients presented after MVCs; 83 were missed and could not be fully assessed for eligibility; 1,784 were approached to determine full eligibility; and

1,405 patients were eligible. Reasons for refusal included a lack of interest in participating

(56%), feeling too unwell for study activities (13%), being concerned about privacy (10%), being worried about the time required to participate (21%), not wanting to provide biosamples (8%), and other reasons (36%). Of the eligible patients, 817 consented to the study. Sufficient interview information was obtained on 711 patients to include in our data analysis (Table 1). The mean age was 43 years. Sixty-seven percent identified as White non-Hispanic, and 40% identified as female. Patients in Portland were older than those in Sacramento and Denver, and they were more often White and male. The differences in racial and ethnic representation across sites generally reflected the demographics of the populations served.

Acute cannabis and alcohol use within 8 hours of injury (Table 2)

Overall, 18% reported cannabis and/or alcohol use (8% cannabis: 14% alcohol, alone or in combination). Self-reported cannabis and alcohol use was highest in the youngest age group

(ages 18–29, 25%) compared with the older age groups. Cannabis use among drivers was also associated with lower educational and income level and lack of health insurance. Cannabis and alcohol use did not significantly vary across racial or ethnic groups.

Some site differences in cannabis and alcohol use prior to MVCs were noted. The Denver population reported the highest cannabis-only use prior to MVCs (8%) of the three sites and lower alcohol-only use (9%, with 4% co-ingestion) compared with Portland, where alcohol only

(14%) and co-use prior to MVCs (4%) were more common than cannabis only (3%). Cannabis and/or alcohol use prior to MVCs was most common in Denver (20%) and Portland (21%), with

9 only 9% reported in Sacramento. None of the Sacramento patients reported co-use (this finding

was corroborated by biosample results).

Patients admitted to the hospital reported alcohol use more often than patients who were discharged (17% vs. 10%). On the other hand, cannabis use was reported in a smaller proportion

of the population requiring admission (7% vs. 9%). Among patients admitted to the intensive

care unit (ICU), there was little difference in self-reported cannabis-only use compared with

those not admitted to the ICU (4.2% vs. 4.7%).

Biosample sensitivity analysis

The study population, using a positive biosample for cannabis and alcohol (as defined in

the Methods section) as an indication of use prior to driving (Table A1 in the Appendix), was

also examined. These criteria suggested more acute cannabis use (18%, alone or in combination

with alcohol) and less acute alcohol use (11%, alone or in combination with cannabis) compared

with self-report. Although cannabis and alcohol use did not significantly vary across racial/ethnic

groups, we saw a higher prevalence of cannabis use with biosampling than self-report across all racial/ethnic groups, with the largest absolute difference among Black participants (29% vs. 7%; compared with 17% vs. 8% for White participants, 15% vs. 7% for Hispanic participants, and

13% vs. 8% for other racial/ethnic groups).

Crash characteristics and self-reported cannabis and alcohol use within 8 hours of injury (Table 3) We examined the likelihood of specific self-reported driving behaviors and crash characteristics in relation to whether the driver reported cannabis use only, alcohol use only, or both. Moderate speeds (16–40 mph) at the time of the crash were more common among those reporting cannabis- or alcohol-only use compared with those reporting both or neither, and high speeds ( > 40 mph) were more common among those reporting both cannabis and alcohol use

10 compared with others. Vehicle rollover and ejection from the vehicle were more common among those reporting cannabis and/or alcohol use compared with neither. Front-end impacts were most common across the MVCs, and more so with cannabis or alcohol use than without; those reporting cannabis or alcohol use were also more frequently in single-vehicle MVCs. Cannabis and alcohol use were not significantly associated with the type of vehicle driven (car/truck/van vs. motorcycle or motor scooters vs. other). Alcohol only and combined use were most often implicated in evening/nighttime driving, but cannabis-only use was common in daytime and evening MVC’s, suggesting different norms and contexts for use.

Biosample sensitivity analysis

When cannabis and alcohol use were defined by biosamples, as opposed to self-report, there were some differences in the crash characteristics data (Table A2 in the Appendix). Seat belt use was less common with cannabis and/or alcohol use (74%, vs. 81% based on self- reported data). Vehicle rollover was no longer significantly associated with cannabis and/or alcohol use, with a much smaller occurrence of rollover among those with cannabis use only

(22% vs. 41% based on self-report). Moderate speed remained more common in MVCs with cannabis only or alcohol only, whereas high speed was more common among those with combined cannabis and alcohol. Vehicle ejection and single-vehicle MVCs remained characteristic of MVCs with cannabis and/or alcohol involvement, although the pattern with cannabis was not as strong compared with the results from self-reported use. Time-of-day patterns were consistent.

Types of cannabis use before injury (Table 4)

Among the 53 patients who reported use in the 8 hours prior to their MVC, there was

heterogeneity of use captured in the extended cannabis instrument. The majority of use was

11 marijuana that was smoked/inhaled through a joint, spliff, or blunt); bong; or bowl. However,

participants also reported the use of edibles and other forms of marijuana as well as use of hash concentrates. In terms of the context of use, the most common location was in a private home

(71%), and participants reported using alone (54%) more often than with others. A dispensary

was the most common location of purchase (44%); acquisition from dealers was rare (6%). The

co-use of other drugs was reported by only 6 patients (11%).

Biosample sensitivity analysis

Biosamples were positive for likely acute cannabis use in 76% of those who self-reported

use. Biosample results suggested more cannabis use, both alone and in combination with alcohol,

than self-report (Table A3 in the Appendix). Self-report captured 43% of use captured by

biosamples; however, self-report also captured some cases that were negative by biosample.

Description of past-year use of cannabis, alcohol, and other drugs (Table 5)

Forty-four percent of participants reported past-year cannabis use, ranging from 38%

(Portland and Sacramento) to 58% (Denver). Of participants reporting past-year cannabis use,

72% reported at least monthly use, 59% reported at least weekly use, and 37% reported daily or near-daily use. Again, the extended instrument captured heterogeneity in forms of cannabis use; in addition to traditional forms of smoking and inhalation, edible products were popular, particularly among women (27% vs. 11% in men). Thirty percent of cannabis users reported that they only used joints, blunts, or spliffs (not shown in table), and 14% reported hash use. Twenty- four percent of cannabis users met criteria (by CUDIT score) of hazardous use, while 21% had a possible cannabis use disorder. Only 11% of cannabis users felt that monthly cannabis use could result in moderate or high risk of physical or other harm, but 34% felt that daily use was risky.

12 There were no gender differences in cannabis-related risk perception, frequency of use, or dependence scores.

There was evidence of regional variation in the popularity of specific types of cannabis

use. For example, edibles were more prevalent among the participants at the Denver site (29%

vs. 11% Portland and 10% Sacramento), while inhalation from pipes was more prevalent at the

Portland site. Among those who reported using hash, oil was used more often in Portland (69%)

than in Denver (15%) or Sacramento (22%). The average CUDIT score in the Denver population

(8) met criteria for cannabis use disorder, although scores across the sites were close (Portland

7.2, Sacramento 7.7).

DISCUSSION

In this study of drivers involved in MVCs presenting to EDs in three cities where

cannabis was legal for recreational use, several observations are notable. First, self-reported use

in the period prior to the MVC was relatively low, while biosamples suggested a much higher

rate of acute use. Second, usual past-year use and cannabis dependence were high compared with

prior ED studies and compared with national samples. Many high-risk crash features (moderate

to high speeds, rollover, vehicle ejection, intrusion, single vehicle) were common in MVCs

associated with cannabis use, as they were for alcohol use and co-ingestion; however, while

patients requiring admission were less likely to report cannabis use, they were more likely to

report alcohol use, suggesting there may have been cannabis-related driving behaviors that

contributed to MVCs but mitigated against more serious harms. Finally, an expanded cannabis

use instrument captured heterogeneity of drug use in this population.

High rates of self-reported usual (past-year) cannabis use and a high level of problem and

harmful use were observed. The prevalence of past-year use (44%) in this study population was

13 higher than in previous ED studies (Walsh et al., 2005) but concordant with more recent

observations of increasing national rates of cannabis use, particularly in states that have legalized

medical marijuana and recreational marijuana use (Cerdá et al., 2020; Palamar et al., 2021). EDs

continue to be an important place to monitor epidemiologic trends in drug and alcohol use and

identify individuals at high risk for health consequences from substance use.

Self-report is the standard for measuring drug use and has demonstrated concordance with objective measures in a variety of settings (Asbridge et al., 2014; Kristensen et al., 2005;

Smith et al., 2018). However, in the current study, we found a large difference between rates of use as divulged by self-report and acute use suggested by biosamples. The biosample might have detected cannabis use that fell outside of the 8-hour window during which we felt use was most likely to influence driving; further, the accuracy of the cannabis levels likely varies depending on the route of use and variable absorption times (especially with edibles). However, when asked about past-year use, which would not carry as much social or legal implications, particularly in a cannabis-legal state, respondents reported high levels of use, making extremely low levels of acute use seem unlikely.

An additional explanation for the difference between self-reported and biosample data might be the specific circumstances of our study: we asked about drug use in relation to an

MVC, an event in which drivers might be concerned about the legal implications of divulging information related to their culpability in the event, despite the increasing normalization of drug use, and drug use reporting, over time (Eichelberger & Kelley-Baker, 2020). In particular, the discordance between self-report and biosample by race supports this possibility, as racial minorities may reasonably fear harsher assessments of culpability and sentencing (Mitchell &

Mackenzie, 2004). The medical establishment's lack of earned trust among racial minority

14 populations likely also played a role; study participants may have been reluctant to report drug use due to social desirability bias and concerns of how such information may impact their clinical care. These findings argue for the necessity of careful protections for privacy and legal and clinical care consequences and relaying such protections to patient populations, not only for

the accuracy of self-reported variables, but also for success recruiting participants into studies. It

also supports the value of the triangulation of data from multiple sources.

With an expanded cannabis use instrument, we captured a wide variability in types of

use, including hash concentrates, edibles, tinctures and lozenges, and other forms of use that

would not be captured on traditional marijuana questionnaires. For example, the Cannabis Abuse

Screening Test (CAST) (Legleye et al., 2007), the Cannabis Problems Questionnaire (CPQ)

(Copeland et al., 2005), and the Marijuana Craving Questionnaire (Heishman et al., 2001) ask

only about smoked cannabis and/or refer only to joints, without mention of any other of the

numerous forms and modes of delivery now readily available for commercial sale. In this study,

asking about cannabis in a limited fashion would have missed significant amounts of information

about drug use. This underscores the need to update traditional means of asking about drug use

and use disorders in healthcare and research settings.

Limitations

The study, by definition, excluded less severe MVCs for which drivers would not be

transported to the hospital and the most severe MVCs with fatally-injured drivers. Recruiting

drivers involved in collisions may have created a bias in reporting of cannabis use due to fear of

legal or medical consequences. Acute use within an 8-hour period does not necessarily translate to impairment. There may have been other factors, including unmeasured confounders, contributing to crash occurrence. There was heterogeneity in the MVC population (Tables 1, 5)

15 across study sites: the percentage of study participants who were full or “911” traumas ranged

from 9% (Sacramento) to 33% (Denver), and the percentage of participants discharged from the

ED ranged from 17% (Portland) to 83% (Denver). These differences likely reflected differences

in the medical centers themselves. For example, the ED in Portland receives a large volume of

trauma transfers from across the state with serious injuries, and given its geographic location,

does not see very many walk-in visits. In contrast, the Denver site had a substantial volume of

patients recruited from its urgent care area, who would be walk-in patients of lower acuity. In

addition to differences in site characteristics, heterogeneity in the MVC population may also

indicate site-to-site sampling bias, which could skew our overall results.

Some heterogeneity in drug use captured across the sites may have been due to local or

regional factors. The absence of reported co-use of cannabis and alcohol in Sacramento, for

example, may have been due to local factors such as more medicinal use in Sacramento; more

recent legalization and thus, more reluctance to self-report use; or tougher prosecution of

impaired driving.

We used concurrent THC levels of > 0.5 ng/mL and a positive 11-OH-THC result to select drivers with acute use. Rarely, those with chronic cannabis use might have detectable 11-

OH-THC, even days after their last use (GHSA, 2018).

Conclusions

This research study suggested a high prevalence of cannabis involvement in MVCs across three cannabis-legal states. The differences between self-report and biosample data for cannabis and alcohol use in this population were significant, and supports the need to use both means of assessing acute use in future studies, particularly when done in this context, where legal and medical concerns may affect reporting. However, limitations of laboratory data when

16 used as an “objective” measure must also be acknowledged, particularly given the myriad forms of cannabis and their varying absorption times. The heterogeneity of use captured supports the need to alter our standard screening processes.

ACKNOWLEDGEMENT

This study was funded by the Insurance Institute for Highway Safety.

17 REFERENCES

Adamson, S. J., & Sellman, J. D. (2003). A prototype screening instrument for cannabis use disorder: The Cannabis Use Disorders Identification Test (CUDIT) in an alcohol- dependent clinical sample. Drug and Alcohol Review, 22(3), 309–315. doi:10.1080/0959523031000154454

Asbridge, M., Mann, R., Cusimano, M. D., Trayling, C., Roerecke, M., Tallon, J. M., . . . Rehm, J. (2014). Cannabis and traffic collision risk: Findings from a case-crossover study of injured drivers presenting to emergency departments. International Journal of Public Health, 59(2), 395–404. doi:10.1007/s00038-013-0512-z

Bondallaz, P., Favrat, B., Chtioui, H., Fornari, E., Maeder, P., & Giroud, C. (2016). Cannabis and its effects on driving skills. Forensic Science International, 268, 92–102. doi:10.1016/j.forsciint.2016.09.007

Cerdá, M., Mauro, C., Hamilton, A., Levy, N. S., Santaella-Tenorio, J., Hasin, D., . . . Martins, S. S. (2020). Association between recreational marijuana legalization in the United States and changes in marijuana use and cannabis use disorder from 2008 to 2016. JAMA Psychiatry, 77(2), 165–171. doi:10.1001/jamapsychiatry.2019.3254

Cherpitel, C. J. (2007). Alcohol and injuries: A review of international emergency room studies since 1995. Drug and Alcohol Review, 26(2), 201–214. doi:10.1080/09595230601146686

Cherpitel, C. J., Witbrodt, J., Korcha, R. A., Ye, Y., Monteiro, M. G., & Chou, P. (2019). Dose- response relationship of alcohol and injury cause: Effects of country-level drinking pattern and alcohol policy. Alcoholism: Clinical and Experimental Research, 43(5):850– 856. doi:10.1111/acer.13986

Cherpitel, C. J., Ye, Y., Andreuccetti, G., Stockwell, T., Vallance, K., Chow, C., & Brubacherc, J. R. (2017). Risk of injury from alcohol, marijuana and other drug use among emergency department patients. Drug and Alcohol Dependence, 174, 121–127. doi:10.1016/j.drugalcdep.2017.01.019

Chihuri, S., Li, G., & Chen, Q. (2017) Interaction of marijuana and alcohol on fatal motor vehicle crash risk: A case–control study. Injury Epidemiology, 4(1), 8. doi:10.1186/s40621-017-0105-z

Copeland, J., Gilmour, S., Gates, P., & Swift, W. (2005). The Cannabis Problems Questionnaire: Factor structure, reliability, and validity. Drug and Alcohol Dependence, 80(3), 313–319. doi:10.1016/j.drugalcdep.2005.04.009

Delling, F. N., Vittinghoff, E., Dewland, T. A., Pletcher, M. J., Olgin, J. E., Nah, G.,…Marcus, G. M. (2019). Does cannabis legalisation change healthcare utilisation? A population- based study using the healthcare cost and utilisation project in Colorado, USA. BMJ Open, 9(5), 27432. doi:10.1136/bmjopen-2018-027432

18 References

Dubois, S., Mullen, N., Weaver, B., & Bédard, M. (2015). The combined effects of alcohol and cannabis on driving: Impact on crash risk. Forensic Science International, 248, 94–100. doi:10.1016/j.forsciint.2014.12.018

Eichelberger, A. H., & Kelley-Baker, T. (2020). Measuring drug use among drivers: How accurate is self-reported use? Journal of Studies on Alcohol and Drugs. 81(1), 104–114. doi:10.15288/jsad.2020.81.104

El-Guebaly, N., Armstrong, S. J., & Hodgins, D. C. (1998). Substance abuse and the emergency room: Programmatic Implications. Journal of Addictive Diseases, 17(2), 21–40. doi:10.1300/J069v17n02_03

Governors Highway Safety Association. (2018). Drug-impaired driving: Marijuana and opioids raise critical issues for states.

Hartman, R. L, Brown T. L, Milavetz G., Spurgin, A., Pierce, R. S., Gorelick, D. A. . . . Huestis, M. A. (2016). Cannabis effects on driving longitudinal control with and without alcohol. Journal of Applied Toxicology, 36(11), 1418–1429. doi:10.1002/jat.3295

Hartman, R. L., Brown, T. L., Milavetz, G., Spurgin, A., Pierce, R., Smither, D., . . . Huestis, M. (2015). Cannabis effects on driving lateral control with and without alcohol. Drug and Alcohol Dependence, 154, 25–37. doi:10.1016/j.drugalcdep.2015.06.015

Hartman, R. L., & Huestis, M. A. (2013). Cannabis effects on driving skills. Clinical Chemistry. 59(3), 478–492. doi:10.1373/clinchem.2012.194381

Heishman, S. J., Singleton E. G., & Liguori, A. (2001). Marijuana Craving Questionnaire: Development and initial validation of a self-report instrument. Addiction, 96(7), 1023– 1034. doi:10.1046/j.1360-0443.2001.967102312.x

Insurance Institute for Highway Safety. (2018). Fatality facts 2018: Yearly snapshot.

Insurance Institute for Highway Safety. (2020). Marijuana laws by state [table]. Retrieved from https://www.iihs.org/topics/alcohol-and-drugs/marijuana-laws-table

Insurance Institute for Highway Safety. (2020). State marijuana laws [map]. Retrieved from https://www.iihs.org/topics/alcohol-and-drugs#marijuana-laws

Kristensen, T. S., Hannerz, H., Høgh, A., & Borg, V. (2005). The Copenhagen Psychosocial Questionnaire—A tool for the assessment and improvement of the psychosocial work environment. Scandinavian Journal of Work, Environment & Health, 31(6), 438–449. doi:10.5271/sjweh.948

Kuendig, H., Hasselberg, M., Laflamme, L., Daeppen, J. B., & Gmel, G. (2008). Alcohol and nonlethal injuries: A Swiss emergency department study on the risk relationship between acute alcohol consumption and type of injury. The Journal of Trauma Injury Infection and Critical Care, 65(1), 203–211. doi:10.1097/TA.0b013e318068fc64

19 References

Legleye, S., Karila, L., Beck, F., & Reynaud, M. (2007). Validation of the CAST, a general population Cannabis Abuse Screening Test. Journal of Substance Use, 12(4), 233–242. doi:10.1080/14659890701476532

Mitchell, O., & Mackenzie, D. L. (2004). The relationship between race, ethnicity, and sentencing outcomes: A meta-analysis of sentencing research.

National Highway Safety Traffic Administration. (2016). 2013–14 National roadside study of alcohol and drug use by drivers. U.S. Department of Transportation.

Palamar, J., Le, A., & Han, B. (2021). Quarterly trends in past-month cannabis use in the United States, 2015–2019. Drug and Alcohol Dependence, 219, 108494. doi:10.1016/j.drugalcdep.2020.108494

Roberts, B. A. (2019). Legalized emergency departments: A cautionary review of negative health and safety effects. Western Journal of Emergency Medicine, 20(4), 557–572. doi:10.5811/westjem.2019.4.39935

Sewell, R. A., Poling, J., & Sofuoglu, M. (2009). The effect of cannabis compared with alcohol on driving. The American Journal on Addictions, 18(3), 185–193. doi:10.1080/10550490902786934

Smith, M. J., Alden, E. C., Herrold, A. A., Roberts, A., Stern, D., Jones, J., . . . Breiter, H. C. (2018). Recent self-reported cannabis use is associated with the biometrics of delta-9- tetrahydrocannabinol. Journal of Studies on Alcohol and Drugs,79(3):441–446. doi:10.15288/jsad.2018.79.441

Taylor, B., Irving, H. M., Kanteres, F., Room, R., Borges, G., Cherpitel, C., . . . & Rehm, C. J. (2010). The more you drink, the harder you fall: A systematic review and meta-analysis of how acute alcohol consumption and injury or collision risk increase together. Drug and Alcohol Dependence, 110(1-2), 108–116. doi:10.1016/j.drugalcdep.2010.02.011

Thies, C. F., & Register, C. A. (1993). Decriminalization of marijuana and the demand for alcohol, marijuana and cocaine. The Social Science Journal, 30(4), 385–399. doi:10.1016/0362-3319(93)90016-O

Walsh, J. M., Flegel, R., Atkins, R., Cangianellia, L. A., Cooper, C., Welsh, C., & Kerns, T. J. (2005). Drug and alcohol use among drivers admitted to a Level-1 trauma center. Accident Analysis & Prevention, 37(5), 894–901. doi:10.1016/j.aap.2005.04.013

20

TABLES

Table 1: Characteristics of the study population Total By site Portland Sacramento Denver N 7111 373 137 201 Men (%) 59.8 68.9 55.5 45.8* Age (mean) 43.1 46.1 40.3 39.6* Race/ethnicity (%) White 66.8 83.3 46.0 50.5* Black 7.8 2.7 13.9 13.0 Hispanics 19.9 9.2 31.4 32.0 Other 5.5 4.8 8.8 4.5 Education level (%) High school graduate or less 32.2 31.2 33.8 33.0 Some college 42.1 45.0 43.4 36.0 College graduate or more 25.6 23.8 22.8 31.0 Income (monthly, after taxes) (%) < $1,500 25.3 24.8 19.8 28.9 $1,500–3,999 44.4 44.8 43.6 44.2 ≥ $4,000 30.3 30.4 36.6 26.9 Insurance Private 48.8 48.9 55.9 43.9* Medicare/Medicaid/other 40.5 41.7 38.2 39.9 None 10.7 9.4 5.9 16.2 Trauma level: MVC (%) Full or 911 13.6 14.4 8.5 33.3* Other trauma level 86.4 85.6 91.5 66.7 Disposition (%) Discharged 47.3 17.1 73.3 82.6* Admitted 51.2 81.8 23.0 16.9 Other 1.5 1.1 3.7 0.5 Note: MVC = motor vehicle crash. 1 Total valid cases are defined by MVC-injured patients who consented to the survey and did not drop from the survey shortly after the interview started. * Significant difference across site (p < 0.05): F-test for continuous measures and chi-square test for categorical measures.

21

Table 2. Self-report of acute cannabis and alcohol use within 8 hours of injury (%, N=6931) Cannabis Alcohol Cannabis & only only alcohol Neither Chi2 test2 (%) (%) (%) (%) (p value) Total 4.5 10.7 3.2 81.7 N.A. Gender Women 3.6 6.8 2.5 87.1 0.023 Men 5.1 13.2 3.6 78.1 Site Portland 3.1 13.7 4.2 79.0 0.002 Sacramento 3.7 5.2 0.0 91.1 Denver 7.5 9.0 3.5 80.0 Age 18–29 6.6 12.7 6.1 74.6 0.002 30–49 5.2 10.8 3.6 80.3 ≥ 50 2.0 9.0 0.4 88.6 Race/ethnicity White 4.6 11.1 3.1 81.2 0.996 Black 3.6 10.9 3.6 81.8 Hispanics 4.3 9.4 2.9 83.4 Other 2.6 7.9 5.3 84.2 Education level High school graduate or less 6.8 11.7 4.9 76.6 0.038 Some college 3.8 12.2 2.4 81.5 College graduate or more 2.2 6.7 2.2 88.8 Income (monthly, after taxes) < $1,500 9.7 11.0 3.9 75.3 0.015 $1,500–3,999 2.2 12.1 2.9 82.8 ≥ $4,000 3.2 9.1 3.7 84.0 Insurance Private 2.4 9.5 3.3 84.8 0.002 Medicare/Medicaid/other 5.1 9.9 2.6 82.4 None 11.0 19.2 5.5 64.4 Trauma level (MVC) Full or 911 1.6 14.5 3.2 80.7 0.763 Other 3.9 11.6 3.1 81.5 Disposition Discharged 5.7 6.6 3.1 84.6 0.026 Admitted 3.8 13.4 3.2 79.6 Injury severity Admitted into ICU 4.2 12.0 1.4 82.4 0.500 Not ICU-admitted 4.7 9.6 3.6 82.0 Motor vehicle type Car/truck/van 3.9 10.2 3.0 82.8 0.862 Motorcycle/motor scooter 6.5 11.6 2.9 79.0 Other (dirt bike, snowmobile) 4.4 11.1 4.4 80.0 Note: ICU = intensive care unit; MVC = motor vehicle crash; N.A. = not applicable. 1 Among those who reported valid answers to the alcohol and cannabis use questions within 8 hours of the MVC. 2 Chi-square test for independence comparing acute substance use and sociodemographic and injury characteristics. 22

Table 3. Crash characteristics (in %) by self-reported cannabis and alcohol use within 8 hours of injury Cannabis Alcohol Cannabis & only only alcohol Neither Chi2 test1 (%) (%) (%) (%) (p value) Miles per hour ≤ 15 mph 13.8 13.0 26.7 29.9 0.023 16–40 mph (n=260) 55.2 59.3 26.7 40.7 > 40 mph 31.0 27.8 46.7 29.4 Wearing seat belt 85.0 85.4 81.3 92.3 0.136 Airbag deployed 77.8 59.5 46.2 59.3 0.326 Vehicle rollover 40.7 36.2 27.8 18.5 0.001 Ejected from vehicle 32.3 31.3 29.4 17.4 0.010 Intrusion into compartment 61.1 37.5 50.0 42.2 0.378 Primary impact site Front 76.0 78.7 86.7 61.1 0.027 Side 8.0 13.1 0.0 21.2 Rear 16.0 8.2 13.3 17.7 Other vehicle involved 46.7 40.8 33.3 70.7 < 0.001 Time of crash 6:00 a.m. to 1:59 p.m. 38.7 4.1 9.1 42.3 < 0.001 2:00 p.m. to 9:59 p.m. 51.6 63.0 45.5 45.5 10:00 p.m. to 5:59 a.m. 9.7 32.9 45.5 12.2 Note. Statistically significant p values are bolded. 1 Chi-square test for independence comparing crash characteristics and acute substance use.

23

Table 4. Types of acute cannabis use within 8 hours of injury and positive biosample N % Any self-reported cannabis use 8 hours before injury (total valid n=697)1 53 7.6 Type of acute cannabis (among cannabis users n=53) Marijuana only 40 75.5 Hash only 10 18.9 Both marijuana and hash 3 5.7 Type of marijuana use (among marijuana users n=43)2 Joint, spliff, or blunt 10 23.8 Stationary vaporizer 1 2.4 Vaping pen 10 23.8 Waterpipe/bong 10 23.8 Glass pipe/bowl 11 26.2 Edible food product 4 9.5 Absorbed in month 1 2.4 Missing 1 Type of hash use (among hash users n=13)2 Dry sift concentrate (kief, , hash) 1 7.7 Water-processed concentrate 1 7.7 Butane solvent extraction 3 23.1 Oil 1 7.7 Wax 7 53.9 Supercritical CO2 extraction 1 7.7 Context of cannabis use; with whom (among cannabis users n=53) Alone 28 53.8 With others 24 46.2 Missing 1 Context of cannabis use; location (among cannabis users n=53) Own home 27 51.9 Other’s home 10 19.2 Private vehicle 0 0.0 Other place 15 28.8 Missing 1 Source to obtain the cannabis (among cannabis users n=53) Grew it 7 13.5 Got it free 8 15.4 Paid a friend or acquaintance 7 13.5 Got it from a dispensary 23 44.2 Obtained from dealer 3 5.8 Other 1 1.9 More than one source 3 5.8 Missing 1 Co-use drugs (among cannabis users n=53) Other illicit drugs 6 11.3 Drug that reduced anxiety or caused sleepiness 1 1.9 Drug that increased alertness or energy 1 1.9 Heroin 0 0.0 Drug used to relieve pain 4 7.5 Drug that caused visual hallucinations 0 0.0 Methadone 0 0.0 No other drugs 47 88.7 Biosamples (among cannabis users n=53) THC positive 35 76.1 THC negative 11 23.9 Missing 7 1 Percentage calculated among those providing a valid answer on self-reported acute cannabis use (N=697). 2 Summation of Ns larger than total N and summation of percentages (%) larger than 100% are due to multiple selections. 24

Table 5. Description of past-year use of cannabis, alcohol, and other drugs Total By site Portland Sacramento Denver N % N % N % N % Any cannabis use PY (total valid n=678) 296 43.7 132 38.0 50 37.6 114 57.6* PY cannabis frequency (among PY cannabis users n=296) Less than monthly 83 28.0 38 28.8 18 36.0 27 23.7 Monthly 40 13.5 18 13.6 4 8.0 18 15.8 Weekly 65 22.0 29 22.0 15 30.0 21 18.4 Daily or nearly daily 108 36.5 47 35.6 13 26.0 48 42.1 Type of PY typical cannabis use (among PY cannabis users n=296) Marijuana only 253 85.8 119 90.2 41 82.0 93 82.3* Hash only 17 5.8 9 6.8 2 4.0 6 5.3 Both marijuana and hash 25 8.5 4 3.0 7 14.0 14 12.4 Missing 1 0 0 1 Type of PY marijuana use (among PY marijuana users n=278)1 Joint, spliff, or blunt 133 48.4 41 34.2 30 62.5 62 57.9* Stationary vaporizer 8 2.9 4 3.3 0 0.0 4 3.7 Vaping pen 57 20.7 21 17.5 10 20.8 26 24.3 Waterpipe/bong 50 18.2 31 25.8 4 8.3 15 14.0* Glass pipe/bowl 101 36.7 56 46.7 10 20.8 35 32.7* Edible food product 49 17.8 13 10.8 5 10.4 31 29.0* Drink 3 1.1 1 0.8 1 2.1 1 0.9 Absorbed in the mouth 12 4.4 8 6.7 1 2.1 3 2.8 Medication or capsule 6 2.2 0 0.0 1 2.1 5 4.7 Other 17 6.2 8 6.7 1 2.1 8 7.5 Missing 3 3 0 0 Type of hash use (among PY hash users n=42)1 Dry sift concentrate (kief, hashish, etc.) 3 7.1 1 7.7 2 22.2 0 0.0 Water-processed concentrate 5 11.9 2 15.4 0 0.0 3 15.0 Butane solvent extraction 11 26.2 2 15.4 2 22.2 7 35.0 Oil 14 33.3 9 69.2 2 22.2 3 15.0* Wax 18 42.9 3 23.1 4 44.4 11 55.0 Budder 5 11.9 1 7.7 1 11.1 3 15.0 Honeycomb 4 9.5 1 7.7 1 11.1 2 10.0 Shatter 12 28.6 1 7.7 1 11.1 10 50.0* Risk perception cannabis use (among PY cannabis users and non-missing on perception, n=287-289) Monthly—moderate/high 31 10.7 18 14.0 4 8.2 9 8.1 Weekly—moderate/high 48 16.7 20 15.5 10 20.8 18 16.4 Daily—moderate/high 99 34.4 45 34.9 21 42.9 33 30.0 CUDIT score (among PY cannabis users and non-missing on CUDIT score, n=275) Mean score 7.6 7.2 7.7 8.0 Alcohol use PY (total valid n=682) None in PY 176 25.8 104 29.7 35 26.1 37 18.7* Less than monthly 132 19.4 63 18.0 32 23.9 37 18.7 Monthly 154 22.6 64 18.3 32 23.9 58 29.3 Weekly 168 24.6 89 25.4 29 21.6 50 25.3 Daily or nearly daily 52 7.6 30 8.6 6 4.5 16 8.1 Any illicit drug PY (total valid n=681)1 120 17.6 66 19.0 20 14.8 34 17.2 Drug that reduced anxiety or caused sleepiness 66 9.7 37 10.6 11 8.1 18 9.1 Drug that increased alertness or energy 25 3.7 11 3.2 5 3.7 9 4.5 Heroin 7 1.0 4 1.1 1 0.7 2 1.0 Drug used to relieve pain 49 7.2 33 9.5 7 5.2 9 4.5 Drug that caused visual hallucinations 7 1.0 2 0.6 2 1.5 3 1.5 Methadone 6 0.9 3 0.9 2 1.5 1 0.5 Note. CUDIT = Cannabis Use Disorder Identification Test; PY = past year. 1 Summation of Ns larger than total N and summation of percentages (%) larger than 100% are due to multiple selections. * Significant difference across site (p < 0.05) from chi-square test. Note that for substance use categories that are not mutually-exclusive (i.e., when multiple selections are possible), the chi-square test was performed for each category separately (versus all others). 25 APPENDIX

Table A1. Lab test for acute cannabis and alcohol (valid N=701)1 Cannabis Alcohol Cannabis & only only alcohol Neither Chi2 test (%) (%) (%) (%) (p value) Total 13.0 6.6 4.8 75.6 N.A. Gender Women 12.7 4.2 3.2 79.9 0.051 Men 13.2 8.1 6.0 72.7 Site Portland 12.4 9.6 6.6 71.4 < 0.001 Sacramento 10.9 2.2 0.0 86.9 Denver 15.4 4.0 5.0 75.6 Age 18–29 16.6 5.0 8.0 70.4 0.015 30–49 14.3 7.2 4.8 73.7 ≥ 50 8.9 7.3 2.4 81.4 Race/ethnicity White 12.4 7.1 5.0 75.2 0.197 Black 25.5 1.8 3.6 69.1 Hispanics 10.6 7.1 4.3 78.0 Other 7.7 2.6 5.1 84.6 Education level High school graduate or less 18.0 4.9 6.8 70.3 0.013 Some college 11.7 7.9 5.1 75.3 College graduate or more 9.1 5.7 1.7 83.4 Income (monthly, after taxes) < $1,500 19.9 5.1 8.3 66.7 0.002 $1,500–$3,999 11.0 7.4 5.5 76.1 ≥ $4,000 8.0 8.0 2.1 81.8 Insurance Private 8.7 5.4 3.3 82.5 < 0.001 Medicare/Medicaid/other 16.2 7.9 4.7 71.2 None 20.3 6.8 12.2 60.8 Trauma level (MVC) Full or 911 16.9 10.8 9.2 63.1 0.167 Other 12.0 7.2 5.0 75.8 Disposition Discharged 12.3 3.7 2.2 81.8 < 0.001 Admitted 13.6 8.7 7.3 70.4 Injury severity Admitted into ICU 10.4 12.5 5.6 71.5 0.004 Not ICU-admitted 13.6 4.5 4.5 77.4 Motor vehicle type Car/truck/van 12.1 6.0 5.2 76.7 0.766 Motorcycle/motor scooter 16.1 5.1 5.1 73.7 Other 13.5 9.0 3.4 74.2 Note. Statistically significant p values are bolded. 1 Valid either in alcohol or cannabis lab test among the total 711 cases (see footnote of Table 1). Note if a case is missing either a blood alcohol concentration (BAC) or cannabis test, it is coded as negative. A case is coded as missing only if both BAC and cannabis tests are missing.

26 Appendix

Table A2. Crash characteristics (in %) by lab test on acute cannabis and alcohol use Cannabis Alcohol Cannabis & only only alcohol Neither Chi2 test1 (%) (%) (%) (%) (p value) Miles per hour ≤ 15 mph 24.7 11.5 17.4 29.8 0.038 16–40 mph (n=260) 50.6 69.2 43.5 39.6 > 40 mph 24.7 19.2 39.1 30.6 Wearing seat belt 88.1 85.7 73.9 93.2 0.006 Airbag deployed 55.9 64.0 63.2 60.3 0.881 Vehicle rollover 21.8 29.7 34.6 19.8 0.177 Ejected from vehicle 23.5 26.3 41.4 17.0 0.005 Intrusion into compartment 49.0 55.0 42.9 41.2 0.500 Primary impact site Front 62.7 85.3 91.7 61.3 0.006 Side 24.0 5.9 8.3 20.4 Rear 13.3 8.8 0.0 18.3 Other vehicle involved 62.8 39.0 35.3 70.1 < 0.001 Time of crash 6:00 a.m. to 1:59 p.m. 32.1 0.0 11.8 44.1 < 0.001 2:00 p.m. to 9:59 p.m. 58.3 61.8 50.0 43.0 10:00 p.m. to 5:59 a.m. 9.5 38.2 38.5 12.8 Note. Statistically significant p values are bolded. 1 Chi-square test for independence comparing crash characteristics and acute substance use.

Table A3. Crossover table comparing self-reported and lab test data on acute cannabis and alcohol use* Lab test1 Cannabis Alcohol Cannabis & Self-report only only alcohol Neither Total Cannabis only 17 1 2 11 31 Alcohol only 8 31 16 18 73 Cannabis & alcohol 7 3 9 3 22 Neither 57 6 6 488 557 Total 89 41 33 520 683 1 Note if a case is missing either a blood alcohol concentration (BAC) or cannabis test, it is coded as negative. A case is coded as missing only if both the BAC and cannabis tests are missing. *p < 0.001 for both marginal homogeneity (Stuart-Maxwell) test (testing the marginal distributions of two measures are equal) and Kappa test (which tests the agreement between two measures).

27