In 2019, 10,196 people in the United States were killed in motor vehicle crashes involving an alcohol-impaired driver; this number increased to over 11,500 in 2020 and over 13,500 in 2021 and 2022 (National Center for Statistics and Analysis [NCSA], 2024). Data limitations, such as testing differences across states, prevent calculation of equivalent numbers for crashes involving a drug-impaired driver (Berning et al., 2022). However, among seriously injured or killed drivers seen in selected trauma centers and medical examiner officers in late 2019, 22.0% tested positive for alcohol, 19.1% for cannabinoids, and 6.8% for opioids; these percentages were 25.2%, 32.9%, and 14.9%, respectively, during early months of the COVID-19 pandemic (April–June 2020; Office of Behavioral Safety Research, 2021).

The United States made progress in reducing alcohol-impaired driving deaths for decades before plateauing in the 1990s/2000s (National Academies of Sciences, Engineering, Medicine [NASEM], 2018; Venkatraman et al., 2021). Despite known effective interventions (Kirley et al., 2023; NASEM, 2018), the number of alcohol-impaired driving deaths per year had remained consistent (around 10,000) from 2010 to 2019 (NCSA, 2021) and has increased to over 13,500 in 2022 (NCSA, 2024). It is important to understand how impaired driving, also referred to as driving under the influence (DUI), might be changing, especially considering the changing landscape as many states continue to legalize adult medical and nonmedical marijuana use (Insurance Institute for Highway Safety [IIHS], 2024a).

Multiple nationally representative surveillance systems collect data on self-reported drinking and driving, such as the Centers for Disease Control and Prevention’s (CDC) Behavioral Risk Factor Surveillance System (BRFSS)¹, CDC’s Youth Risk Behavior Surveillance System (YRBSS)², and the National Survey on Drug Use and Health (NSDUH)³. Analyzing BRFSS data, Barry et al. (2021) found that an estimated 4 million adults engaged in 147 million alcohol-impaired driving episodes in 2018. Analyzing 2019 national YRBSS data, Yellman et al. (2020) found that among high school students who had driven in the 30 days before the survey, 5.4% drove after drinking alcohol. Both studies found variation across sociodemographic groups, including higher prevalence among males than females (Barry et al., 2021; Yellman et al., 2020).

NSDUH also collects information on DUI. Analyzing NSDUH data, Lipari et al. (2016) calculated the 2014 prevalence of DUI of alcohol (11.1%), illicit drugs including marijuana (4.1%), and alcohol and illicit drugs simultaneously (2.4%). DUI of alcohol and of illicit drugs were highest among respondents aged 21–29 years and 20–23 years, respectively; both DUI types were more common among males than females. Analyzing 2002–2014 trends, Lipari et al. (2016) found decreases over time in DUI of alcohol and of illicit drugs. Azofeifa et al. (2015) analyzed 2002–2014 NSDUH data among respondents aged 16–25 years and excluded respondents who used nonmarijuana illicit drugs. They found that DUI of alcohol decreased whereas DUI of marijuana did not.

Among drivers aged ≥16 years in the United States in 2019, DUIA was the most common type of impaired driving, with an estimated 19,795,246 drivers (8.3%) engaging in this behavior at least once during the past 12 months. Additionally, an estimated 12,657,284 drivers (5.3%) engaged in DUIM, and an estimated 2,196,518 drivers (0.9%) engaged in DUID (i.e., cocaine or crack, heroin, hallucinogens, inhalants, or methamphetamine). All three DUI types were highest among drivers who were male, had never been married, identified as not heterosexual, and did not always wear a seatbelt. DUIA and DUIM were highest among drivers aged 21–25 years; DUID was highest among drivers aged 21–25 or 26–34 years. DUIA was highest among drivers with a college degree, higher family income, and full-time employment. DUIM and DUID were higher among drivers with lower family income and who were unemployed and lower among drivers with a college degree. Only DUIM differed by metropolitan status and was lowest among respondents living in nonmetro areas.

DUIA was highest among drivers identifying as non-Hispanic White or non-Hispanic more than one race; DUIM was highest among people identifying as non-Hispanic more than one race and lowest among people included in the non-Hispanic other category. Group heterogeneity makes it difficult to draw conclusions. There were no significant DUIM differences between non-Hispanic White and non-Hispanic Black or Hispanic drivers.

When comparing DUIA and DUIM by age group, drivers aged 16–20 years exhibited the largest discrepancy, with DUIM (11.0%) almost three times as high as DUIA (4.0%); this age group cannot legally buy or use alcohol or nonmedical marijuana under most circumstances in any state. Drivers aged 21–25 years were the only other age group for which DUIM (13.6%) exceeded DUIA (12.9%); however, the absolute difference was less than one percentage point. Significance testing was not performed across DUI types.

Prevalence difference results are similar to those of previous studies using national survey data (e.g., Azofeifa et al., 2019; Barry et al., 2021; Rudisill & Smith, 2020; Yellman et al., 2020). Differences exist but may be due to study sample or variable recoding differences. Higher prevalence of DUI among respondents identifying as gay, lesbian, or bisexual (LGB) was also found by Rudisill and Smith (2020). LGB people experience social and structural inequities, including stigma and discrimination, and are disproportionately affected by multiple health threats (CDC, 2014), including a higher prevalence of substance use and use disorders, compared with people identifying as heterosexual (McCabe et al., 2009). Prevention services developed specifically for this population that consider unique experience and stressors are needed (CDC, 2014). The association between inconsistent seatbelt use and DUI has been identified in survey data (Barry et al., 2021; Yellman et al., 2020) and crash data (Shyhalla, 2014; NCSA, 2024). Adoption of primary enforcement seatbelt laws (Shults et al., 2004; Kirley et al., 2023) covering all seating positions is a proven strategy to increase seatbelt use and can help protect all car occupants if a crash occurs.

There were significant 2016–2019 trend changes in DUIA and DUIM prevalence among drivers aged ≥16 years, with DUIA decreasing slightly and DUIM increasing. These changes were not found to differ across multiple sociodemographic groups, including family income poverty level and race/ethnicity. There were a few groups for which changes in behavior patterns were found, which could guide prevention efforts. For drivers who did not always wear a seatbelt, DUIA decreased while DUIM increased, bringing the 2019 prevalence of DUIM (11.5%) closer to DUIA (12.1%). A similar pattern was noted for drivers aged 26–34 years; additionally, this group experienced a significant increase in DUID from 2016 (1.5%) to 2019 (2.0%).

The overall decrease in DUIA among drivers aged ≥16 years is encouraging; however, DUIA did not decline for all groups, and prevalence remains high. Millions of drivers still drive impaired by alcohol every year, with more than 13,500 people dying in alcohol-impaired driving crashes in 2022 (NCSA, 2024), despite the availability of many proven prevention strategies. Effective countermeasures include ignition interlocks for all people convicted of alcohol-impaired driving, lowering blood alcohol concentration (BAC) limits, alcohol policies to decrease drinking to impairment, and supporting affordable alternative transportation options, among others (Kirley et al., 2023; NASEM, 2018).

Most high-income countries have BAC limits of ≤ 0.05 g per deciliter (g/dL; World Health Organization, 2018; Yellman & Sauber-Schatz, 2022), but all U.S. states, except for Utah, have a BAC limit of 0.08 g/dL. In 2018, Utah became the first U.S. state to implement a 0.05 g/dL BAC law (Thomas et al., 2022). Lowering BAC limits to 0.05 g/dL deters alcohol-impaired driving across a range of high BACs—not just 0.05–0.08 g/dL (NASEM, 2018). These lower limits have saved lives globally (NASEM, 2018) and in Utah (Thomas et al., 2022). An estimated 1,790 lives could be saved each year if all states adopted a 0.05 g/dL BAC limit (Fell & Scherer, 2017).

Intersections between alcohol-impaired drivers and legal or medical systems provide opportunities to apply effective countermeasures through screening or assessment for excessive drinking or alcohol use disorder (AUD). People who drink excessively can benefit from brief interventions assessing readiness, motivators, and barriers to behavior change. People with AUD can benefit from long-term, specialized AUD treatment programs (Kirley et al., 2023; NASEM, 2018). Both interventions address what can be an underlying cause of DUI behavior to prevent reoccurrence. Voas et al. (2016) found that incorporating AUD treatment into an ignition interlock program reduced DUI recidivism even after interlock removal.

The increased DUIM and steady DUID (with the increase for drivers aged 26–34 years) are also concerning. Marijuana and other illicit drugs can negatively impact skills needed for safe driving (Busardo et al., 2018; McCartney et al., 2021). Before 2016, nonmedical marijuana use for adults aged ≥21 years was legal in four states and the District of Columbia (DC). By 2019, this number rose to 10 states and DC, and by September 2024, 24 states and DC allowed for adult nonmedical marijuana use (IIHS, 2024a). A study estimating the effects of marijuana legalization and retail sales on crash rates from 2009 to 2019 in 11 states found increases of 5.8% for injuries and 4.1% for fatalities; however, these effects varied by state and were based on early data that may not tell the full story (Farmer et al., 2022).

Few effective countermeasures have been identified for drug-impaired driving. Enforcement of drug-impaired driving laws is promising, and trained drug recognition experts (DREs) are often used to identify potential impairment by conducting a standardized 12-step protocol (Kirley et al., 2023). Like for alcohol-impaired drivers, intersections between drug-impaired drivers and legal or medical systems may provide opportunities to assess and identify individuals with substance use disorder and link them to specialized treatment.

Further research on effective countermeasures will be assisted by continued advancement of roadside detection methods for drugs and standardized implementation of toxicology testing protocols (D’Orazio et al., 2021). Currently, drug data in the Fatality Analysis Reporting System (FARS) have limitations that preclude comparisons across geography or time (Berning et al., 2022). Additionally, interpretation is complicated. A positive test does not always mean the driver was impaired, and drugs such as opioids may have been administered during treatment (Berning et al., 2022; Berning & Smither, 2014). In addition to containing a provision on advanced impaired driving prevention technology for motor vehicles, the recent Bipartisan Infrastructure Law (i.e., the Infrastructure Investment and Jobs Act) contains a provision directing the National Highway Traffic Safety Administration (NHTSA) to investigate barriers states face in submitting toxicology information into FARS and make recommendations to address them (NHTSA, n.d.). Addressing these barriers and improving drug-impaired driving data in FARS will ultimately improve countermeasure evaluation.