DRUG USE AMONG INDIVIDUALS INJURED IN NON-FATAL MOTOR VEHICLE CRASHES AND RELATED POLICIES IN 2023

Abstract

Objective:

Although driving-related policies aimed at mitigating drug-related motor vehicle crashes (MVCs) have been implemented in diverse communities, data regarding their effectiveness are largely absent. A comprehensive evaluation of these policies is necessary. Furthermore, as US states legalize the recreational use of cannabis, the impact of these policies on drug-related crashes also needs to be evaluated. The objective was to assess the association between drug use prevalence among individuals (age 16+) injured in non-fatal MVCs in 2023 and various related policies, including drug-impaired driving policies, sobriety checkpoints, enforcement programs, and state cannabis legalization status.

Methods:

We analyzed 2023 emergency medical services (EMS) records of individuals (age 16+) injured in non-fatal MVCs across 19 US states where EMS personnel indicated drug use, excluding duplicate, incomplete, and alcohol-only records. Using these counts, we calculated the prevalence of drug use among individuals (age 16+) injured in non-fatal MVCs in each state. The association between drug use prevalence and state-level policies, including drug-impaired driving laws (i.e., per se or zero tolerance), sobriety checkpoints, enforcement programs, and cannabis legalization laws, was evaluated using adjusted Poisson regression with random effects for state differences. Policies were assessed individually and in a full model to evaluate their individual and additive effects.

Results:

In 2023, 11,538 individuals (68.2% male) were injured in drug-related non-fatal crashes. Neither drug-impaired driving policies, the use of sobriety checkpoints, nor the implementation of state judicial outreach liaisons influenced the prevalence of drug use among individuals injured in non-fatal crashes. In contrast, relative to states with no policy or cannabidiol/low tetrahydrocannabinol, those permitting recreational cannabis had significantly higher prevalence (adjusted prevalence ratio [aPR]: 1.57, 95% confidence interval [CI]: 1.22, 2.02). The implementation of sobriety checkpoints was associated with higher drug use prevalence (aPR: 1.59, 95% CI: 1.22, 2.09) when drug-impaired driving policies were absent, particularly in states permitting recreational cannabis.

Conclusion:

Our findings show differences in drug use prevalence among individuals (age 16+) injured in non-fatal MVCs based on state-level policies, highlighting the need for holistic enforcement strategies to address drug-related crashes, especially amid the increasing risks associated with the legalization of recreational cannabis use.

INTRODUCTION

Drug use is a known risk factor for motor vehicle crashes (MVCs) and has been linked to fatal MVCs. While MVC fatalities are significant, they represent only a portion of the overall burden of drug-related crashes. It is likely that drug use also increases the risk of non-fatal MVCs, though evidence on this is limited. In this study, we examine the prevalence of drug use among individuals injured in nonfatal MVCs. We also consider state-level factors that may influence crash risk, such as cannabis legalization, drug-impaired driving laws, and enforcement strategies (e.g., sobriety checkpoints).

Cannabis Policy Landscape

As of 2024, 24 United States (US) states and the District of Columbia have legalized recreational cannabis use, 14 states have legalized only medical cannabis use, and 9 states allow only cannabidiol (CBD) or low- Tetrahydrocannabinol (THC) products (National Conference of State Legislatures 2023; CDC 2024). All of the states with recreational cannabis laws also have medical cannabis laws. States with only medical cannabis laws permit cannabis for medical purposes as defined by each state (CDC 2024). In states allowing only CBD or low-THC products, CBD and low concentrations of THC, typically between 0.3% and 0.9%, are permitted. Some states with CBD/low-THC-only policies entirely restrict THC, allowing only CBD products (National Conference of State Legislatures 2023). As noted, cannabis legalization policies vary significantly across these categories, with differences in definitions, allowable products, and THC concentrations. While multiple studies have linked cannabis legalization to fatal MVCs (Calvert and Erickson 2020; Marinello and Powell 2023; Park et al. 2024), its effect on non-fatal MVCs remains underexplored.

Drug-Impaired Driving Laws and Enforcement Strategies

Despite the evolving landscape of cannabis legalization, policies for mitigating drug-related MVCs have largely remained unchanged. Driving under the influence of drugs (DUID) is illegal in all 50 US states. This allows law enforcement officers to issue citations based on their observations and impairment assessments, such as sobriety tests. However, DUID prosecution often requires proof that drug use impaired the individual’s ability to drive, which presents significant challenges in many jurisdictions (Kirley et al. 2023). In response, some states have enacted laws that establish thresholds for drug presence where impairment may be presumed for prosecution. These drug-impaired driving policies include zero tolerance and per se laws (Kirley et al. 2023). Zero tolerance laws prohibit any detectable drug amount, whereas per se laws set specific concentration limits for substances such as THC. However, there is no universally accepted standard for defining drug-related driving impairment, and the substances covered under per se laws, along with dosage limits, vary across states (NASID 2023).

Key enforcement strategies to mitigate drug-related MVCs include sobriety checkpoints, Drug Recognition Experts (DRE), the National Law Enforcement Liaison Program (LEL), the Traffic Safety Resource Prosecutor (TSRP), and the State Judicial Outreach Liaison (SJOL) (National Highway Traffic Safety Administration 2021). DRE, LEL, and TSRP programs exist in nearly all states, but the use of sobriety checkpoints and SJOL programs varies. Overall, there is limited understanding of how these mitigation policies are associated with drug-related non-fatal MVCs.

Motor Vehicle Drivers and Occupants

Drug-related MVC mitigation policies primarily focus on drivers, likely because of crash risks associated with their drug use and its impact on their ability to drive safely (Rudisill and Smith 2020), drug-related MVCs extend beyond just drivers. However, occupants riding with intoxicated drivers (RWID) are often intoxicated themselves (Soderstrom et al. 1996; Romano et al. 2012). Intoxicated occupants face the same rates of hospitalization and risk of MVC injury as intoxicated drivers (C. R. Schermer et al. 2001; Carol R. Schermer et al. 2001; Elliott et al. 2009). Intoxicated occupants also create an unsafe environment for drivers by causing distractions (Elliott et al. 2009), and distracted driving is a well-known risk factor for MVCs. Additionally, accurately identifying occupant status in drug-related MVCs, particularly non-fatal ones, can be challenging due to legal consequences that may lead individuals to misreport their occupant status (Elliott et al. 2009). Therefore, it is important to evaluate drug-related non-fatal MVCs more comprehensively and assess how these mitigation strategies are associated with drug-related non-fatal MVCs overall.

The Current Study

The purpose of this study is to estimate drug use prevalence among individuals injured in non-fatal MVCs. We investigate several factors that may be linked to drug-related non-fatal MVCs, including state-level policies on cannabis legalization, drug-impaired driving laws (i.e., per se or zero tolerance laws), sobriety checkpoints, and SJOL. This exploratory analysis offers broader insights into various factors that may increase the risk of non-fatal MVCs. Identifying an association allows us to determine whether further research is needed to explore the nature of these relationships in depth and evaluate the details of the findings.

MATERIALS AND METHODS

Data Source on Individuals Injured in Non-fatal MVC

We utilized the biospatial.io dataset, Emergency Medical Services (EMS) Electronic Patient Care Reports (ePCR) from all 50 US states, with comprehensive data available from 27 states. State data coverage percentages were determined using historical submission data and information from comparable geographical regions, known as Underlying Event Coverage (UEC). The UEC calculation also accounted for expected event changes due to the COVID-19 pandemic. The biospatial.io dataset strictly adhered to the National EMS Information System (NEMSIS) standard (2022) and validated records for compliance to the NEMSIS standard using the Schematron validation system (2022). The NEMSIS standard was developed by the National Highway Traffic Safety Administration (NHTSA) Office of EMS and the National Association of State EMS Officials to standardize how patient care information is documented in 911 dispatch call records across US EMS agencies, enabling comparability at the state and county levels. Its primary objective was to create uniformity in EMS reporting, addressing variations in reporting requirements and training across states. For this reason, the biospatial ePCR data was suitable for comparing state-level drug-related crash data concerning different state policies. For our analyses, we included 19 US states that met the criterion of a UEC of 75% or higher in 2023: Alabama, Arkansas, Colorado, Florida, Georgia, Illinois, Kansas, Kentucky, Maine, Michigan, Montana, New Mexico, North Carolina, Rhode Island, South Carolina, Utah, Virginia, Wisconsin, and Wyoming.

The EMS ePCR data provided the number of individuals injured in non-fatal MVCs in 2023 for each included state. Records of individuals with non-fatal MVC injuries, excluding fatalities, were identified as having non-severe (≥ 11) or severe (≥ 5 and < 11) injury levels based on the Revised Trauma Score (Won et al. 2024). Records were classified as MVC-related if they indicated two or more of the following: vehicular, pedestrian, or other risk factors related to MVCs; area of vehicle impact; patient’s location in the vehicle; use of occupant safety equipment (excluding ‘none’); or airbag deployment. A second evaluation was conducted to ensure records were correctly identified as MVC-related. This included checking for one or more of the following: ‘traffic or transportation incident’ or ‘automated crash notification’ in the dispatch complaint per NEMSIS Version 3, or ‘traffic accident’ in the dispatch complaint or cause of injury per NEMSIS Version 2. Specific ICD-10-CM codes indicating MVC injury (e.g., V02-V04, V09.0, V12-V79, V81-V88, etc.) or mentions of MVC-related terms in the narrative or patient complaints (e.g., ‘traffic accident,’ ‘airbag deployment,’ ‘motorcycle crash,’ ‘MVC,’ ‘pedestrian struck’) were also considered (biospatial.io 2022).

Drug use identification followed NEMSIS guidelines, excluding records involving solely alcohol. The criteria for identifying drug use encompassed patient admissions of drug use, positive drug levels in law enforcement or hospital records, the presence of drug paraphernalia at the scene, or identification of drug-related odors by EMS personnel (NEMSIS 2022). Information on drug testing results across all states, including whether tests were conducted and how many yielded positive results, was unavailable and could not be included in our analysis. Additionally, concerns in the literature (Slater et al. 2016; Berning et al. 2022) about the consistency and completeness of drug testing in MVCs, especially non-fatal ones, led us to exclude specific drug test results. Although drug testing data is ideal for identifying drug presence, the NEMSIS EMS dataset is widely used in the literature for identifying drug presence and drug-related cases.

We excluded records of individuals under 16 years old due to age-related variations in crash and substance use risks (Azagba et al. 2019; Alcover and Thompson 2020). Records lacking explicit indications of ‘treated and transported by this EMS’ were excluded to prevent duplicate entries from multiple EMS responses to a single crash. Included records indicated the following under patient disposition: ‘patient contact made,’ ‘patient evaluated and care provided,’ ‘initiated and continued primary care,’ and ‘transported by this EMS unit.’ To preserve patient confidentiality, state-level datasets with fewer than nine records were suppressed in the biospatial.io database. Due to the limited availability of crash and individual-level data, it was not possible to distinguish between drivers, passengers, motorcyclists, and pedestrians. Given that not all fatal MVCs are reported to EMS, especially those resulting in immediate fatalities, our analyses focused on non-fatal MVCs. Using these record counts extracted from the biospatial.io dataset, we calculated the 2023 prevalence of drug use among individuals 16 years of age and older injured in non-fatal MVCs in each state. This was done by dividing the number of individuals injured in non-fatal MVCs with reported drug use by EMS by the total number of individuals injured in non-fatal MVCs in each state.

State Policies and Program Data

Data on Drug-impaired driving policies were gathered from National Highway Traffic Safety Administration reports (Walsh 2009; Lacey et al. 2010; National Highway Traffic Safety Administration 2021) and state legislation, with verification from recent data (National Conference of State Legislatures 2022; NASID 2023; Governor Highway Safety Association 2023). We cross-referenced official state legislation websites for states with gaps or discrepancies, such as Kentucky, North Carolina, and Virginia. Colorado’s law, reliant on jury determinations rather than biospecimen tests, excluded it from being categorized as having a drug-impaired driving policy (National Conference of State Legislatures 2022). Professional programs like DRE, LEL, TSRP, and SJOL (American Bar Association 2023) were identified from NHTSA reports (National Highway Traffic Safety Administration 2021). DRE, TSRP, and LEL were excluded because all states included in our analyses enacted these programs in 2023. Cannabis legalization data in 2023 were sourced from the National Conference of State Legislatures report on state cannabis laws (National Conference of State Legislatures 2023).

Data Analyses

To assess differences in drug use prevalence among individuals (age 16+) injured in non-fatal MVCs based on state-level policies, we first conducted a descriptive analysis. Independent t-tests and Kruskal-Wallis tests were used to compare the mean annual prevalence of drug use among individuals injured in non-fatal MVCs in 2023 across each policy group: drug-impaired driving policy (yes or no), allowance of sobriety checkpoints (yes or no), implementation of SJOL (yes or no), and state cannabis legalization status (CBD/low THC or no policy, only medical cannabis, recreational cannabis).

Adjusted Poisson regression models were then performed for each policy group in 2023. Poisson regression was the most suitable model for our data, as it effectively handles low prevalence proportions, which matched the characteristics of our dataset. Additionally, our data met the assumptions of Poisson regression, followed a Poisson distribution, and showed no evidence of overdispersion. Alternative models, such as negative binomial and logistic regression, were not appropriate for our analysis. To account for state-specific differences and the potential influence of the overall number of individuals injured in non-fatal MVCs, we included random effects for state differences and adjustments for the number of individuals injured in the Poisson model.

Poisson regression analyses were also used to assess the additive effects of various policies on the prevalence of drug use among individuals (age 16+) injured in non-fatal MVCs, including drug-impaired driving policies, sobriety checkpoints, and SJOL. Wyoming was excluded from the additive models due to the absence of relevant policies. The first additive model specifically examined the association between the number of policies ranging from 1 to 3 and the prevalence of drug use among individuals injured in non-fatal MVCs. The second and third additive models evaluated combinations of policies categorized into Groups A, B, and C, considering different state cannabis legalization statuses.

Group A included a drug-impaired driving policy and permitted a sobriety checkpoint. Group B permitted a sobriety checkpoint but lacked a drug-impaired driving policy. Group C had a drug-impaired driving policy but did not permit a sobriety checkpoint. The reference category for the second additive model was Group A with CBD/low THC only/no cannabis legalization policy. The reference group for the third additive model was Group A (having a drug-impaired driving policy and permitted sobriety checkpoints). All statistical analyses were performed using SAS, Version 9.4 (SAS Institute 2022). This study obtained an exemption from the University of Florida Institutional Review Board as it did not involve human subjects.

RESULTS

In 2023, a total of 11,538 individuals 16 years of age and older in 19 states were injured in non-fatal MVCs with indications of drug use. The majority (68.2%, N=7,869) were male, and 50.4% (N=5,815) were non-Hispanic white. Regarding age distribution, 29.8% (N=3,434) fell within the 20–29 age bracket, closely followed by 27.6% (N=3,183) within the 30–39 age bracket. The specific states and their respective drug-impaired driving policy and program status are detailed in Table 1, with Table 2 providing information on the number of states in each policy group.

Table 1.

List of Drug-Impaired Driving Policies and Program Status by State

State	Drug-Impaired Driving Policy	Years Since Drug-Impaired Driving Policy	Sobriety Checkpoint Permitted	State Judicial Outreach Liaisons Program	States’ Cannabis Legalization Status
Alabama	No	0	Yes	No	Medical-only
Arkansas	No	0	Yes	No	Medical-only
Colorado	No	0	Yes	No	Recreational
Florida	No	0	Yes	No	Medical-only
Georgia	Yes	22	Yes	Yes	CBD/Low THC
Illinois	Yes	26	Yes	Yes	Recreational
Kansas	No	0	Yes	No	No Policy
Kentucky	Yes	19	Yes	Yes	Medical-only
Maine	No	0	Yes	No	Recreational
Michigan	Yes	20	No	No	Recreational
Montana	Yes	10	No	No	Recreational
New Mexico	No	0	Yes	Yes	Recreational
North Carolina	Yes	17	Yes	Yes	CBD/Low THC
Rhode Island	Yes	17	No	No	Recreational
South Carolina	No	0	Yes	Yes	CBD/Low THC
Utah	Yes	29	Yes	No	Medical-only
Virginia	Yes	18	Yes	Yes	Recreational
Wisconsin	Yes	20	No	Yes	CBD/Low THC
Wyoming	No	0	No	No	CBD/Low THC

Note: The status of drug-impaired driving policy includes the presence of zero tolerance or per se laws.

Source: (American Bar Association, 2023; Governor Highway Safety Association, 2023; International Association of Chiefs of Police, 2022; Lacey et al., 2010; NASID, 2023; National Conference of State Legislatures, 2022, 2023; National District Attorneys Association, 2023; National Highway Traffic Safety Administration, 2023a; US Department of Transportation, National & Highway Traffic Safety Administration, 2021; Walsh, 2009)

Table 2.

Summary of 2023 Mean Prevalence of Drug Use Among Individuals (Age 16+) Injured in Non-fatal Motor Vehicle Crashes (MVCs) by State Policies

Variable	N (States)	Number of Individuals with Drug Use Injured in Non-fatal MVCs (N=11,538)	Mean Prevalence (SD)	Pa
Drug-Impaired Driving Policy				0.29
No	9	4,954	0.028 (0.014)
Yes	10	6,584	0.023 (0.005)
Sobriety Checkpoint				0.60
No	5	1,399	0.023 (0.008)
Yes	14	10,139	0.026 (0.012)
State Judicial Outreach Liaisons				0.52
No	11	5,076	0.026 (0.012)
Yes	8	6,462	0.023 (0.008)
State Cannabis Legalization				0.02
CBD/Low THC or No Policy	6	3,539	0.020 (0.003)
Medical-only	5	2,974	0.019 (0.005)
Recreational	8	5,025	0.032 (0.012)

Note: The status of drug-impaired driving policy includes the presence of zero tolerance or per se laws.

^a.p-value from Independent T-test (Sobriety Checkpoint and State Judicial Outreach Liaisons) and Kruskal-Wallis Test (Drug-Impaired Driving Policy and State Cannabis Legalization Status)

The binary analysis, employing an independent t-test for sobriety checkpoint and SJOL and a Kruskal-Wallis test for drug-impaired driving policy and state cannabis legalization status, revealed no statistically significant differences for all policies and programs except state cannabis legalization status, where significance was observed at p=0.02. States with recreational cannabis policies demonstrated a mean prevalence of 0.032 (SD 0.012) for drug use among individuals (age 16+) injured in non-fatal MVCs, compared to a prevalence of 0.019 (SD 0.005) for states permitting only medical cannabis and 0.020 (SD 0.003) for states with CBD/low THC or no specific cannabis policy.

The adjusted Poisson regression model reaffirmed consistent observations. Individual models, adjusted for state differences and the number of individuals injured in non-fatal MVCs, indicated no significant differences based on drug-impaired driving policy, sobriety checkpoint permission, or the presence of SJOL. However, the model evaluating the drug use prevalence among individuals (age 16+) injured in non-fatal MVCs concerning state cannabis legalization status revealed significant findings. States permitting recreational cannabis exhibited a 57% higher prevalence (95% confidence interval [CI]: 1.22, 2.02) of drug use among individuals (age 16+) injured in non-fatal MVCs compared to states with CBD or low THC-only policies or no specific policy (as depicted in Table 3).

Table 3.

Poisson Regression: Mean Prevalence of Drug Use Among Individuals (Age 16+) Injured in Non-fatal Motor Vehicle Crashes (MVCs) and Related Policies in 19 States

	Individual Models	Full Model
Variable	Adjusted Prevalence Ratio (95% Confidence Interval)	Adjusted Prevalence Ratio (95% Confidence Interval)
Drug-Impaired Driving Policy
No	REF	REF
Yes	1.14 (0.68, 1.99)	1.33 (0.80, 2.21)
Number of Years Since Drug-Impaired Driving Policy	0.98 (0.96, 1.00)	0.98 (0.96, 1.00)
Sobriety Checkpoint
No	REF	REF
Yes	1.35 (0.94, 1.93)	1.59 (1.22, 2.09)***
State Judicial Outreach Liaisons
No	REF	REF
Yes	0.99 (0.70, 1.39)	0.91 (0.76, 1.09)
State Cannabis Legalization
CBD or Low THC/ No Policy	REF	REF
Medical-only	1.00 (0.81, 1.25)	0.86 (0.63, 1.16)
Recreational	1.57 (1.22, 2.02)***	1.52 (1.25, 1.84)***

Note: The status of drug-impaired driving policy includes the presence of zero tolerance or per se laws.

^***p<0.001.

All separate models with a random effect for the state, controlling for the number of individuals injured in any non-fatal MVCs

In the full Poisson regression model encompassing all policies and programs, states permitting sobriety checkpoints exhibited a 59% higher prevalence (95% CI: 1.22, 2.09) of drug use among individuals (age 16+) involved in non-fatal MVCs compared to states without these checkpoints when accounting for the status of all other policies and programs. The significant association with cannabis legalization status persisted, revealing a 52% higher prevalence (95% CI: 1.25, 1.84) of drug use among individuals (age 16+) injured in non-fatal MVCs in states permitting recreational cannabis compared to those with CBD or low THC or no policy (refer to Table 3).

To assess the observed higher drug use prevalence among individuals (age 16+) injured in non-fatal MVCs within states permitting sobriety checkpoints, we referred to the results of the additive models (N= 18 US states). In the additive model examining various combinations of drug-impaired driving policy and permitted sobriety checkpoints alongside state cannabis legalization status, Group A indicates states with both a drug-impaired driving policy and sobriety checkpoint, Group B represents states with only a sobriety checkpoint, and Group C denotes states with only a drug-impaired driving policy (see Table 4). Among states with CBD/low THC only or no cannabis legalization policy, Group C exhibited a lower prevalence of drug use among individuals (age 16+) injured in non-fatal MVCs (adjusted prevalence ratio [aPR]: 0.78, 95% CI: 0.67, 0.91) compared to Group A. Among states permitting only CBD/low THC cannabis use or none, the sobriety checkpoints alone did not appear effective in preventing drug use among individuals (age 16+) injured in non-fatal MVCs. In comparison to states in Group A with CBD/low THC/no policy, Group B states with recreational cannabis policies had a higher prevalence (aPR: 2.20; 95% CI: 1.86, 2.61) of drug use among individuals (age 16+) injured in non-fatal MVCs. This result suggests that the presence of sobriety checkpoints may not effectively mitigate drug use when recreational cannabis is legalized without corresponding drug-impaired driving policies.

Table 4.

Definition of Policy Groups (A-C) Included in Additive Statistical Models: Drug-Impaired Driving Policies and Programs Status

Policy Group	Enacted Drug-Impaired Driving Policies	Permitted Sobriety Checkpoints
A	X	X
B	-	X
C	X	-

Note: Drug-impaired driving policy is considered to have a zero tolerance or per se law for drug use. The symbol X represents the presence of such a law.

When examining differences by policy groups among states with permitted recreational cannabis, Policy Group B, with a sobriety checkpoint and no drug-impaired driving policy, had a higher prevalence (aPR: 2.06, 95% CI: 1.89, 2.25) of drug use among individuals (age 16+) injured in non-fatal MVCs compared to Group A, which had both a sobriety checkpoint and drug-impaired driving policy. This finding indicates that even among states with similar cannabis legalization status, the lack of comprehensive drug-impaired driving policy may render sobriety checkpoints less effective in reducing drug use involved in MVCs. All additive models can be found in Table 5.

Table 5.

Additive Statistical Models: Combinations of Policies and Mean Prevalence of Drug Use Among Individuals (Age 16+) Injured in Non-fatal Motor Vehicle Crashes (MVCs) in 18 States

Variables	# of States	Mean Prevalence (SD)	Adjusted Prevalence Ratio (95% Confidence Interval)	p-value
MODEL 1:
Number of Policies (Range: 1–3)	18	0.02 (0.01)	0.89 (0.75, 1.06)	0.19
MODEL 2:
Combination of Policies
CBD/ Low THC/None, A	2	0.021 (0.003)	REF	-
CBD/ Low THC/None, B	2	0.020 (0.004)	0.97 (0.75, 1.24)	0.80
CBD/ Low THC/None, C	1	0.016 (0.00)	0.78 (0.67, 0.91)***	0.001
Medical Cannabis Only, A	2	0.023 (0.001)	1.08 (0.93, 1.26)	0.30
Medical Cannabis Only, B	3	0.017 (0.006)	0.82 (0.56, 1.20)	0.31
Recreational Cannabis, A	2	0.022 (0.001)	1.07 (0.91, 1.24)	0.41
Recreational Cannabis, B	3	0.046 (0.04)	2.20 (1.86, 2.61)***	<0.001
Recreational Cannabis, C	3	0.026 (0.009)	1.23 (0.86, 1.75)	0.25
MODEL 3:
Policy Combinations for States with Recreational Cannabis
A: Yes Checkpoint, Yes Policy	2	0.022 (0.001)	REF	-
B: Yes Checkpoint, No Policy	3	0.046 (0.04)	2.06 (1.89, 2.25)***	<0.001
C: No Checkpoint, Yes Policy	3	0.026 (0.009)	1.15 (0.83, 1.60)	0.39

Note:

^***p<0.001.

The number of policies pertains to the presence of drug-impaired driving policies (i.e., zero tolerance or per se laws), sobriety checkpoints, and/or State Judicial Outreach Liaisons. Policies combination A permits sobriety checkpoints with a drug-impaired driving policy in place. Combination B permits sobriety checkpoints but does not have a drug-impaired driving policy. Combination C does not permit sobriety checkpoints but has a drug-impaired driving policy in place. All included states had either sobriety checkpoints or a drug-impaired driving policy. All models incorporated a random effect for state differences.

DISCUSSION

We examined the association between drug use prevalence among individuals (age 16+) injured in non-fatal MVCs and state-level policies, including having a drug-impaired driving policy, permitting sobriety checkpoint, an SJOL, and state cannabis legalization. As discussed in the Introduction, this study was not designed to establish causal inferences. Rather, its purpose was to identify associations that could be more rigorously investigated in future research. No statistically significant associations were found between drug-impaired driving policies or SJOLs and drug use prevalence among individuals (age 16+) injured in non-fatal MVCs across the 19 states. Sobriety checkpoints appeared to be linked to lower drug use among injured individuals in non-fatal MVCs only when implemented alongside other programs. Our findings suggest a higher prevalence of drug use among individuals injured in non-fatal MVCs in states with sobriety checkpoints but without drug-impaired driving policies, compared to states with both measures, especially in states with recreational cannabis use. States permitting recreational cannabis use also had higher drug use prevalence among individuals (age 16+) injured in non-fatal MVCs compared to those with CBD, low THC only, or no cannabis policy.

The absence of a statistical association in our findings regarding drug-impaired driving policies (i.e., zero tolerance or per se) may be attributed to our dataset, which included all individuals (age 16+) injured in non-fatal MVCs, not just drivers. This broader scope could have obscured the specific impacts of policies designed to target drivers. However, given that drug-related MVCs are a public health concern affecting more than just drivers (C. R. Schermer et al. 2001; Elliott et al. 2009; Romano et al. 2012; US Department of Transportation National Highway Traffic Safety Administration 2019) and the challenges in correctly identifying drivers at the time of an MVC (Elliott et al. 2009), it may be useful to consider additional interventions that extend beyond drivers in addressing drug-related crashes.

Variations in drugs types and state-specific drug thresholds may also contribute to the lack of statistical significance regarding drug-impaired driving policies (NASID 2023). The efficacy of per se laws for drugs can be complicated by variations in individuals in drug tolerance, dosage, use timing, chronic versus occasional use, and poly-drug use, which affect the detectable drug concentrations (Reisfield et al. 2012). Furthermore, challenges in collecting samples to enforce per se or zero tolerance laws remain. Implicit consent laws for chemical testing vary by state, with some requiring evidence of impairment for DUI cases during court summons (Reeder 1972). While all 50 states have implied consent laws for alcohol testing, not all do for drugs. Some states impose restrictions on officers obtaining biospecimens for drug testing, often requiring a sample collection warrant (Walsh 2009; Chow et al. 2019). Unlike alcohol, drug roadside testing is complicated by varying detection times, depending on the toxicology testing method and drug type. For instance, THC can be detected for up to eight days in oral fluid tests (Andås et al. 2014), while amphetamine are detectable for up to five days in urine tests. Persistent drug users present additional difficulties in screening (Chow et al. 2019). In most states, law enforcement primarily uses blood analysis for drug testing, which poses challenges due to its invasiveness, the need for a warrant, and requiring a healthcare professional or facility for blood draws (Kirley et al. 2023). The NHTSA suggests training law enforcement to conduct blood draws (Kirley et al. 2023), but this remains a challenge during traffic stops and MVCs.

Our findings suggest that sobriety checkpoints alone are insufficient without corresponding drug-impaired driving policies and enforcement measures. While these checkpoints are effective for alcohol, significant barriers remain concerning drug threshold cutoffs and enforcement. Without comprehensive and consistent implementation, sobriety checkpoints are unlikely to reduce drug-related MVCs (Morrison et al. 2021). Additionally, our analysis focused solely on the presence of sobriety checkpoints as a drug-related MVC mitigation policy, without examining their enforcement or testing practices due to data limitations. This highlights the need for additional data and future research on enforcement practices in states that permit sobriety checkpoints. Future research should investigate enforcement rates, checkpoint frequency, the number of drivers tested, drug test results, and outcomes following these tests, as well as community awareness and perceptions of sobriety checkpoints and state laws on drug-related MVCs and overall MVC injury rates. Our findings provide a starting point for these policy investigations.

We found a significant association between recreational cannabis legalization and drug use prevalence among individuals (age 16+) injured in non-fatal MVCs, with a 57% higher prevalence compared to states with CBD, low THC, or no cannabis policy. This finding is particularly relevant as more states legalize recreational cannabis use. Studies have similarly linked cannabis use to crash risk (Preuss et al. 2021), crash injuries (Brubacher et al. 2018), and higher-risk driving behaviors (Choo et al. 2022), along with cognitive and psychomotor effects (Chow et al. 2019). Research also indicates that cannabis legalization increases MVCs (Aydelotte et al. 2019; Farmer et al. 2022; González-Sala et al. 2023) and cannabis-related driving under the influence (Johnson et al. 2012), although some studies have reported insignificant associations (Aydelotte et al. 2017; González-Sala et al. 2023). Factors such as recreational cannabis store density may contribute to these mixed results (Aydelotte et al. 2019; Santaella-Tenorio et al. 2020).

Another factor contributing to our finding of higher drug use prevalence among individuals (age 16+) injured in non-fatal MVCs in states with recreational cannabis legalization may be the increased rates of cannabis use in these states (Hall and Lynskey 2016; Grigsby et al. 2020; Gunadi et al. 2022). It remains unclear whether cannabis legalization caused this increase in drug use or if these states already had elevated rates before legalization. Nonetheless, states that legalized recreational cannabis use, compared to states with only CBD, low THC, or no cannabis policy, exhibited a higher prevalence of drug use among individuals (age 16+) injured in non-fatal MVCs. While the causation of legalization alone for the higher prevalence is unknown, it reflects a state-level policy characteristic associated with higher drug use among individuals injured in non-fatal MVCs. Our findings, in conjunction with existing literature, emphasize the need for implementing safety measures to address drug-related MVCs, particularly in states with recreational cannabis use.

Lastly, a major concern raised by our findings is the disparity in state-level drug testing rates, with low rates of drug testing for drivers who died at the scene and even lower rates for surviving drivers. As mentioned earlier, incomplete data on drug testing results severely limits research on drug-related MVCs. In 2020, NHTSA reported that only 54% of drivers who died in MVCs were tested for drugs, while a mere 17% of surviving drivers involved in fatal crashes were tested (National Transportation Safety Board 2022). Testing rates for non-fatal MVCs are even lower compared to fatal MVCs (Slater et al. 2016; National Transportation Safety Board 2022). A comprehensive approach to collecting population-level data is crucial for evaluating the effectiveness of different policies and programs. Increasing and standardizing drug testing rates, along with uniform recording practices, would facilitate more detailed research. Mandatory testing for all individuals involved in MVCs, including data on drug type, dose, and polydrug use, is essential for accurately assessing risks and protective factors associated with drug-realted MVCs and advancing research in this field.

Limitations and Strengths

Our findings rely on EMS-recorded data quality standardized through NEMSIS protocols. Despite efforts to reduce misclassification, drug use determination depends on EMS perception, potentially leading to overestimation or underestimation of drug-related MVC incidence. Indications of drug use do not definitively establish crash causes, hindering causal inferences. Non-reporting to EMS may underestimate drug-related MVCs, and state-level aggregation may vary in local-level data representation.

Our procedure to enhance reliability may contribute to underreporting, such as excluding states with incomplete data and using exclusive ‘treated and transported’ cases. Additionally, reliance on secondary sources for legalization and state policy data and limitations in available crash-level and enforcement data need to be revised for our analyses. While recreational cannabis legalization shows a higher prevalence of drug use among individuals injured in non-fatal MVCs, causation regarding cannabis-related crashes remains unclear. This analysis specifically examined the association between state-level policies and drug use prevalence among individuals (age 16+) injured in non-fatal MVCs. It is essential to note that drug use prevalence does not equate to impairment, and correlation does not imply causation. Due to limited crash and person-level details in the data, we were unable to distinguish between drivers, passengers, motorcyclists, and pedestrians, nor identify the type of drug involved in the crash, which poses a limitation for more in-depth analyses. Our analyses also focused on the presence of these state-level policies, rather than enforcement activity. These limitations should be considered when interpreting our results.

Nonetheless, our study is, to our knowledge, the first to explore insight on current drug-related MVC policies and programs on drug use among individuals (age 16+) injured in non-fatal MVCs. This contributes valuable insights and highlights gaps in current research. Furthermore, it underscores the association between cannabis legalization policies and the prevalence of drug use among individuals injured in non-fatal MVCs, which is an important finding given increasing availability and use of cannabis. Considering the constraints imposed by the COVID-19 pandemic on traditional law enforcement data collection regarding DUI and drug-related MVCs obtained from DUI arrests, our use of population-level data, such as EMS data, provides a more comprehensive representation of the issue.

CONCLUSION

In conclusion, our analyses of the association between drug-impaired driving policies, SJOLs, and permitted sobriety checkpoints with the prevalence of drug use among individuals 16 years of age and older injured in non-fatal MVCs in 2023 found no statistically significant individual associations. However, a significant association was observed between state cannabis legalization status and drug use prevalence, with states permitting recreational cannabis use exhibiting higher prevalence of drug use. This finding is not surprising, given existing reports indicating higher rates of cannabis use in these states. The directionality remains unclear; it is uncertain whether cannabis legalization led to increased drug use among individuals injured in non-fatal MVCs or whether these states already had higher rates prior to legalization. Nonetheless, states with recreational cannabis legalization showed a higher prevalence of drug use among individuals injured in non-fatal MVCs. The full model indicated a higher prevalence of drug use among individuals (age 16+) injured in non-fatal MVCs when sobriety checkpoints were assessed alongside other programs. We found that sobriety checkpoints were linked to a higher crash-related drug use prevalence when the drug-impaired driving policy was absent, compared to states with both sobriety checkpoints and a drug-impaired driving policy. These findings show differences exist in drug use prevalence among individuals (age 16+) injured in non-fatal MVCs depending on state-level policy landscapes. They underscore the complex interplay of policies, highlight the need for comprehensive road safety approaches, and emphasize challenges in drug-related MVC research. Further research is necessary to investigate specific directions, causes, and contributing factors driving these differences.