Motor vehicle collision remains one of the leading causes of injury mortality in the U.S. (Rockett et al. 2012). Research suggests that driving under the influence of drugs other than alcohol is a growing public health (Walsh et al. 2004) and global traffic safety concern (Morland 2000, Movig et al. 2004). In the U.S., the prevalence of drug-involved driving was estimated to be 11%-14% in 2007 (Lacey et al. 2009).

Driving under the influence of either illegal drugs or prescription medications may lead to driver impairment and/or an increased risk of motor vehicle collision. In regards to illicit substances, there are inconsistencies in the research concerning stimulants, such as cocaine or amphetamines, and driver cognizance (Kelly et al. 2004). As for prescription medications, benzodiazepines have been linked to an increased risk of motor vehicle collision (Walsh et al. 2004). There is evidence that those taking prescription opioids, such as oxycodone or hydrocodone, may be at an increased risk of traffic crash (Orriols et al. 2009). Contrarily, for the opioid methadone, there is no increased risk of motor vehicle collision associated with long term usage (Ogden and Moskowitz 2004).

The specific drugs consumed by fatally injured drivers and changes in their use over time throughout the U.S., including the role of prescription medications, have been largely under studied. Findings from the 2007 U.S. National Roadside Survey indicated that the occurrence of prescription narcotics, particularly the opioids oxycodone, hydrocodone, and methadone, and depressants, such as benzodiazepines, were common among those that drive under the influence with a prevalence of 1.2%-3.3% and 2.4%-3.4%, respectively (Lacey et al. 2009). Findings from the 2007 National Roadside Survey also suggested that cannabinoids and cocaine were common among those that drive under the influence (Lacey et al. 2009). Even though cannabinoids were more prevalent than cocaine amongst drivers under the influence (Lacey et al. 2009), cannabis has been legalized for medicinal use in several states (Koepsell et al. 1994), making it not entirely illegal. Therefore, cocaine is likely more representative of illicit drug use amongst those who drive under the influence as it a controlled substance typically not available outside of a healthcare institution.

Because of the potential for impairment and the prevalence of substance use amongst drivers, there is a need to discern how drug use is trending for public health intervention. Therefore, the purpose of this study is a trend analysis to examine the changes in drug use among fatally injured drivers in motor vehicle crashes from 1999 to 2010 in the U.S. Particular interest is given to changes in general drug usage, broad categories of drugs, and representative prescription medications and illegal substances including depressants, specifically benzodiazepines and barbiturates, opioids, explicitly methadone, hydrocodone, and oxycodone, and cocaine.

The descriptive characteristics of the fatally injured drivers with complete toxicological profiles are presented by test result status in Table 1. Of the 95,654 drivers included in the analysis, 23,500 (24.6%) tested positive for at least one drug. With the exception of the day of the week and survival time post-collision, all demographic characteristics were statistically significant. Of the total drivers tested, 87.5% were between the ages of 21-64 years. Males were tested more often than females and tended to be of white, non-Hispanic race/ethnicity. Approximately 8% of the total drivers tested held commercial drivers’ licenses. The vast majority of participants were administered a blood test (85.5%) to assess drug presence. Of the total drivers tested, approximately 62% had blood alcohol concentrations equal to 0 g/dl, compared to 33% of the total drivers tested had blood alcohol concentrations ≥ 0.08 g/dl. Over 95% of the drivers tested had not been previously convicted of DWI, while over 84% had not been involved in a recent collision. Of the total drivers tested, the day of the crash was not significant with almost equal percentages occurring during the week as opposed to the weekend. Of the total drivers tested, single-vehicle collisions were more prevalent than multiple-vehicle collisions (51.6% vs. 48.4%, respectively). Approximately 67.2% of the total drivers tested met their demise within 1 hour of crash, though this was not statistically significant (p=0.1519). Over the 12 year span, the total number of drivers tested steadily increased from 6,686 in 1999 to 7,032 in 2010 and this was significant (p= <0.0001). The results of the sensitivity analysis showed similar results (data not shown).

Table 2 presents the percentage of the population testing positive for drugs and the prevalence ratios for each demographic characteristic previously described. The prevalence ratio was shown to quantify whether the demographic characteristic/attribute was associated with either an increased or decreased occurrence of a drug positive result compared to the specified referent sub-group. Individuals aged <16, 55-64, and ≥65 years had a lower prevalence of drug positive results whereas individuals aged 21-24, 25-34, 35-44, and 45-54 years all had higher prevalence of drug positive results. Females tended to test positive for drugs slightly lower than males (PR=0.90; 95% CI 0.87, 0.92). Asians had the lowest prevalence of positive drug tests (PR=0.59; 95% CI 0.54, 0.65) compared to all other races. Both Hispanic ethnicity (PR=0.84 95% CI 0.81, 0.88) and commercial drivers (PR=0.87; 95% CI 0.83, 0.90) had a lower prevalence of drug usage. Individuals with a BAC greater than 0.00 g/dl tended to test positive for drugs. Those with BAC concentrations between 0.01-0.07 g/dl had the highest prevalence of drug positive test results (PR=1.52; 95% CI 1.46, 1.59) followed by those with a BAC ≥0.08 g/dl (PR=1.30; 95% CI 1.27, 1.33). Individuals that had a previous DWI or crash within the past 3 years both tended to test positive for drugs. The presence of drugs was detected 28% higher in single vehicle collisions compared to multiple vehicle crashes (PR=1.28; 95% CI 1.24, 1.30). Interestingly, the prevalence of drug positive tests steadily increased from 1999-2010. While the number of drivers testing positive increased over this time span, the percentage testing positive for drugs increased from 18.1% in 1999 to 29.1% in 2010. The results of the sensitivity analysis showed similar results (data not shown).

The prevalence rates and ratios of drug involvement amongst these fatally injured drivers are presented in Table 3. From 1999-2000 to 2009-2010, the rates of drug involvement in fatally injured drivers in the U.S. dramatically shifted. Between these two time points, the number of fatally injured drivers whom tested positive for any drug use increased 49% (rate ratio, PR=1.49; 95% CI 1.42, 1.55). With the exception of hallucinogens, increases of drug prevalence rates were observed in virtually all broad categories of drugs from 1999-2000 to 2009-2010. The largest increases were seen in narcotics (PR=2.73; 95% CI 2.41, 3.08) depressants (PR=2.01; 95% CI 1.80, 2.25), and cannabinoids (PR=1.99; 95% CI 1.84, 2.16). As for specific drugs, prescription drug use radically increased amongst fatally injured drivers between 1999-2000 and 2009-2010. The greatest increases in rate were seen for prescription opioids. The rate of hydrocodone and oxycodone present in fatally injured drivers was more than 6 times higher in 2009-2010 compared to 1999-2000 (PR=6.37; 95% CI 5.07,8.02), while methadone detection was over 4 times higher in 2009-2010 (PR=4.29; 95% CI 2.88, 6.37). The rate of prescription benzodiazepines found in fatally injured drivers more than doubled in 2009-2010 compared to 1999-2000 (PR=2.27; 95% CI 2.00, 2.58). Barbiturates were the only class of prescription drugs to experience significant declines in rate (PR=0.53, 95% CI 0.37, 0.75). The rate of cocaine use in fatally injured drivers did not increase between 1999-2000 and 2009-2010 (PR=0.94, 95% CI 0.84, 1.06).

The results of the sensitivity analysis showed comparable results (not shown). The number of fatally injured drivers from these states whom tested positive for any drug use increased 54% (PR=1.54; 95% CI 1.43, 1.66). The largest increases were seen in cannabinoids (PR=2.51; 95% CI 2.18, 2.89), depressants (PR=2.27; 95% CI 1.86, 2.78), and narcotics (PR=2.14; 95% CI 1.76, 2.60). The rate of hydrocodone and oxycodone present in fatally injured drivers from these states was over 5 times higher in 2009-2010 compared to 1999-2000 (PR=5.34; 95% CI 3.63, 7.85), while methadone detection was over 3 times higher (PR=3.51; 95% CI 1.88, 6.55). Rates of prescription benzodiazepines more than doubled in drivers from these states (PR=2.46; 95% CI 1.94, 3.10). The rate of cocaine use in fatally injured drivers from these states did not increase between 1999-2000 and 2009-2010 (PR=0.89, 95% CI 0.72, 1.11).

Over the 12-year period, the patterns of drug involvement in fatally injured drivers also changed. From 1999-2010, the overall drug testing percent (Fig. 1) and number testing positive for any drug use steadily increased (Fig. 2). In general, narcotics, depressants, and cannabinoids steadily increased over this time span, while stimulants were on the rise until 2005, but progressively declined (Fig. 3). As for specific drugs, rates of prescription oxycodone/hydrocodone, methadone, and benzodiazepines have also been steadily increasing (Fig. 4). Prescription barbiturates declined over the 12-year period. Contrarily, the occurrence of cocaine among fatally injured drivers steadily increased from 1999 to 2005 and rapidly declined from 2006 to 2010. The results of the sensitivity analysis using states with over 80% testing percentage showed comparable results (data not shown).

The principal finding of this study suggests that substance use is not only intensifying, but changing among those that drive under the influence of drugs in the U.S. With the exception of barbiturates, the occurrence of prescription drugs found in those involved in fatal traffic crashes has greatly increased from 1999 to 2010. The prevalence rates for narcotics, particularly hydrocodone, oxycodone, methadone, and depressants, such as benzodiazepines, have not only significantly risen, but are trending upwards. Contrarily, there appears to be a shift in illicit drug use. While drug prevalence rates for cannabinoids steadily increased from 1999 to 2010, this substance is becoming legalized for medicinal use in several states. Illegal drugs appear to be declining in fatal traffic crashes. The pervasiveness of cocaine in fatal vehicle crashes initially increased, but has been rapidly declining since 2006. Considering these observed trends, there is evidence that a shift from illegal to prescription drugs may be occurring in those involved in fatal traffic crashes.

These findings are consistent with the current literature. In the U.S., prescription opioid analgesic use has dramatically increased since the 1990's (Joranson et al. 2000, Paulozzi et al. 2006a). From 1997 to 2005, the retail sales of prescription methadone, oxycodone, and hydrocodone have risen 933%, 588%, and 198%, respectively (Manchikanti 2007). Similar trends have been seen in other countries. From 2000-2005, both prescription opioid and benzodiazepine consumption increased 1.5-10% (Ravera et al. 2009), whereas methadone (Ravera et al. 2009) and barbiturate use remained relatively stable in Europe (Nicholas et al. 2012).

The potential shift from illegal to prescription medication among those that drive under the influence may be evident in trends seen in emergency room departments and hospital admissions. Even though unintentional poisonings in the U.S. have dramatically increased, most of these incidences have been attributed to prescription drugs (Coben et al. 2010). From 1994-2002, emergency room cases involving narcotic analgesics increased 198% in comparison to cocaine, which only increased 39% (Paulozzi et al. 2006a). From 2004-2008, the number of emergency department visits due to the use of opioids and benzodiazepines increased 111% and 89%, respectively (Centers for Disease Control and Prevention 2010), which may be attributed to the fact that these two drugs are often used concomitantly (Jones et al. 2012). Similar trends have been observed in hospital admissions. Hospitalizations for poisonings due to prescription opioids, sedatives, and tranquilizers increased more than 65% from 1999-2006 (Coben et al. 2010). Of these prescription medications, hospitalizations due to methadone and benzodiazepines increased 400% and 39%, respectively, compared to barbiturates, which decreased 41% (Coben et al. 2010). Death rates attributed to opioid analgesics was four times higher in 2008 compared to the rate in 1999 (Centers for Disease Control and Prevention 2011). Of these increases, deaths attributed to oxycodone and hydrocodone increased 57%, and methadone increased over 200% (Paulozzi et al. 2006b). Deaths from opioid analgesics surpassed overdose deaths attributed to cocaine (Centers for Disease Control and Prevention 2011).

Even though trends in prescription drug consumption and healthcare can be observed, a scarcity of information exists on the occurrence of driving under the influence of drugs in the U.S. According to the NHTSA, from 2005-2009, the presence of both prescription and illegal drugs in fatally injured drivers increased from 28% to 33% (National Highway Traffic Safety Administration 2010). A study conducted in Washington State showed that cocaine use increased 2% and benzodiazepine use increased 275% in fatally injured drivers (Schwilke et al. 2006). Even though illicit drug use prevailed over prescription substances in this analysis, the findings indicate that prescription medications are becoming more ubiquitous in drugged driving incidences (Schwilke et al. 2006). Another U.S. study indicated that polydrug use (i.e. the concurrent usage of two or more drugs) may also be problematic (Brady and Li 2013). The primary concern with multiple medication use is the interactive effects of the drugs which may impair the driver even more then if the substances were used separately. While epidemiologic data regarding polydrug use are scant, approximately 20% of drivers found under the influence of drugs tested positive for two or more substances (Brady and Li 2013). Combinations of alcohol with cannabis and/or stimulants, including cocaine and methamphetamine, were found most prevalent (Brady and Li 2013).

Despite the lack of U.S. data, studies from other countries may provide evidence of shifting trends in drugged driving. The frequency of drugged driving has steadily increased in countries such as Sweden, Finland, Norway, Switzerland, France, Scotland, and Australia (Drummer et al. 2003, Mura et al. 2006, Bernard et al. 2009, Jones et al. 2009, Officer 2009, Ojaniemi et al. 2009, Senna et al. 2010, Blencowe et al. 2011, Bogstrand et al. 2011). Even though increases in illicit drug use, particularly cocaine (Mura et al. 2006, Senna et al. 2010), have been noted, greater increases in prescription drugs have been seen (Drummer et al. 2003, Bernard et al. 2009, Officer 2009, Ojaniemi et al. 2009, Blencowe et al. 2011, Bogstrand et al. 2011). Although increases in opioid analgesics and methadone have been observed, the largest increases have been seen in benzodiazepine use while driving (Drummer et al. 2003, Bernard et al. 2009, Officer 2009, Ojaniemi et al. 2009, Blencowe et al. 2011, Bogstrand et al. 2011).

The shifting trends in drug use among those that drive under the influence may be partly explainable. Cocaine is still a widely-used recreational drug in the U.S., but its popularity appears to be waning compared to alcohol, tobacco, and cannabis (Degenhardt et al. 2008). Recent declines have been attributed to the discontinuation of use and/or demise of frequent users (Reuter 2006). Cocaine is not as popular with young adults as teens appear to be abusing prescription drugs (Friedman 2006). Cocaine's recent decline may be attributed to changes in the price of the drug and more effective anti-drug policies (Costa Storti and De Grauwe 2009). Several societal and political changes may have contributed to the rise of prescription drugs among those that drive under the influence. Internet pharmacies have made obtaining prescription medications easier (Manchikanti 2007). In addition, numerous benzodiazepines and opioids came off patent protection from 2004-2008 enabling generic versions to be manufactured (Adis International 2002, Bailey et al. 2006, Srihari et al. 2009). Generic drugs are typically less expensive making them more accessible to patients (Srihari et al. 2009). Healthcare policy initiatives may have exacerbated this problem. National initiatives to address the under-treatment of pain have been linked to the large increases in opioid prescriptions (Manchikanti 2007, Pletcher et al. 2008). Methadone use among drugged drivers may not have risen as quickly as other drugs partly because it is heavily regulated (Mark et al. 2009). Declines in barbiturate use among those that drive under the influence may be attributed to the development of newer, less toxic medications (Doghramji 2006).

The findings and implications of this study show that driving under the influence of drugs is a major public health concern in the U.S. As indicated by the observed trends, this problem will likely amplify. Awareness campaigns are needed to show both the public and healthcare providers that driving under the influence of drugs, including common prescription medications, may be hazardous. The need for more enforcement is likely necessary.

The strengths of this study are the use of multi-state data over a 12-year time span. Data obtained from fatally injured drivers is more likely to be comprehensive with greater chances of reportable drug test results compared to those simply injured or cited in a traffic collision. The weaknesses of this study are the limitations of the FARS data. As previously mentioned, not all drivers killed in fatal crashes were tested for drugs. This issue was addressed in the sensitivity analysis. In addition, only a maximum of three drugs can be documented per person in FARS, which could potentially be a bias in cases of multiple-drug involvement. These findings do not attempt to prove that the presence of drugs found in the fatally injured drivers caused the traffic fatalities or that the drugs identified were misused.

Among drivers killed in U.S. traffic crashes, the prevalence of drug use is rising and a shift from illegal to prescription medications may be occurring. While illegal substances, such as cocaine, have been declining, increases in narcotics and depressants were observed. The changes in trend may be a result of recent societal and political movements. These findings indicate that driving under the influence of drugs, particularly prescription medications, is a growing public health and traffic safety concern in the U.S.