Background
Psychoactive substances are prevalent in emergency department samples of injured drivers, and several studies have found high prevalence rates not only of alcohol but also of medicinal and illicit drugs [
1‐
3]. Studies of drivers in normal moving traffic have shown lower prevalence rates [
4‐
6]. In line with this, case–control studies have demonstrated that presence of psychoactive substances in drivers is associated with an elevated risk of traffic injuries [
7‐
10]. The elevated risk can be explained by the impairing effects of psychoactive substances on driving ability (see for instance [
11]). In this study we apply a case–control design to address psychoactive substance use and non-fatal traffic injuries.
Several types of methodological weaknesses hamper many of the case–control studies that have assessed the relative risk of psychoactive substance use for traffic injuries. These comprise non-representative samples of controls (e.g. other patients) and non-equivalent biological samples for cases and controls ([
7]). In an ideal situation controls should be random drivers in normal traffic and blood samples should be collected from both cases and controls, as blood is the biological sample that best reflects the drug concentrations in the central nervous system (CNS). Earlier case–control studies of psychoactive substance use and non-fatal traffic accidents have not been performed in an ideal situation. Mura and co-workers used blood samples from both cases and controls, but the controls were not random drivers in normal traffic [
9], which implies a possible selection bias in the control group. In the study by Movig and co-workers controls were sampled from drivers in normal traffic, but blood samples were compared with urine samples [
8]. As psychoactive substances have longer detection times in urine than in blood [
12] their presence in urine does not necessarily reflect impaired driving. Comparisons of blood and urine samples can therefore bias the relative risk estimates downwards.
The ideal situation with bloods samples from controls in normal moving traffic is, however, very difficult to obtain. It is costly to obtain a sufficiently large sample of random drivers in normal traffic and it is difficult to collect blood samples in normal moving traffic. Drivers may be reluctant to participate in roadside studies and to give blood samples, even with monetary incentives [
13]. Encouraging high levels of participation is very important as the total prevalence may be as low as 4.5% or less for both alcohol and other psychoactive substances among drivers in moving traffic [
4‐
6].
One way of overcoming the problem with collection of blood samples in random drivers is to use oral fluid for drug screening in moving traffic. Collection of oral fluid has proved successful in random drivers, with low refusal rates [
5]. Oral fluid can be compared to blood samples as both specimen types are positive for drugs for the same average length of time after intake of a single dose. The distribution of drug concentrations in oral fluid reflects the distribution of drug concentrations in blood in a population [
14]. Comparisons between blood samples and oral fluid samples can be made if equivalent cut-off concentrations in blood and oral fluid are used [
14]. One previous case–control study has compared psychoactive substances in cases and random driver controls applying comparable biological samples (blood and oral fluid) to assess relative risk estimates for traffic injuries [
7]. While these estimates were for fatal traffic injuries we have in this study applied the same study design with random driver controls and comparable biological samples to assess relative risk estimates with respect to non-fatal injuries.
Combinations of drugs are prevalent among injured drivers [
2,
3,
15] while combinations in moving traffic are rare [
5]. Some studies suggest that combinations of various psychoactive substances constitute a particular increased risk of traffic injuries [
8,
11]. In line with this, a recent Norwegian case–control study of fatal road traffic accidents found few single substance cases in the accident group but a major increase in risk for drivers with combinations of psychoactive drugs [
7]. It therefore seems particularly important to assess the impact of multiple psychoactive substance use on traffic injuries.
The aim of our study was to compare the prevalence of alcohol and psychoactive drugs in drivers admitted to hospital for treatment of non-fatal injuries after road traffic accidents with prevalence in drivers in normal traffic, and to calculate odds ratios for single and multiple substance use.
Discussion
This study revealed high risk estimates for alcohol and multi-drug use when comparing drivers admitted to the emergency department for injuries in car accidents with drivers in normal traffic from the hospital catchment area. The study found no statistically significant risk associated with the use of one non-alcohol psychoactive substance alone.
Other studies have also found significantly increased risk for accidents associated with the use of alcohol (see Taylor et al. for a review [
27]). For instance, the Grand Rapids Study in 1964 found BAC levels above 0.04% to be associated with increased accident rates, and higher BAC levels gave higher accident rates [
28]. Similar findings were published in another American study by Blomberg and colleagues in 2009 [
29]. In the Netherlands, Movig and colleagues found that drivers who tested positive for alcohol (BAC above 0.8 g/l) had an increased likelihood of being injured in an accident; OR = 15.5 (95% CI:7.1- 33.9) [
8] compared with 36.1 (95% CI:13.2- 98.6) in our study. The higher OR reflects a lower estimated prevalence of alcohol in the control group in our study. A previous study has shown that the prevalence of alcohol above the legal BAC limit in normal traffic is relatively low in Norway compared to other European countries [
5], which is probably due to the restrictive BAC limits (0.2 g/kg) and related sanctions in Norway.
Single use of non-alcohol psychoactive substances was not statistically significantly associated with injury risk in our study, which is probably due to the low number of cases. With a larger case group, it might have been possible to identify psychoactive non- alcohol single substances associated with traffic accidents, as this was the most prevalent finding in the control group. In the present study, combinations of non- alcohol psychoactive drugs were more prevalent than single non-alcohol substances in the case group.
Prevalence of stimulant drugs was high in the case group. In other studies, prevalence of cocaine and amphetamines was low and no significant risk increase was found for single use of these substances [
8,
9]. Stimulant substances and amphetamines in particular are prevalent in Norwegian drivers apprehended by the police and the drugs are often found in combination with other psychoactive substances [
30]. No stimulant substances was found alone in both the case and control group in this study so it was not possible to calculate an OR for the association between accident risk and single stimulant drugs.
The highest relative risk estimate in this study was for alcohol combined with one or more other psychoactive substances. Similar findings have been reported in earlier studies [
7,
8]. Reviewing the literature on the effect of cannabis compared with alcohol on driving Sewell and co-workers [
31] concluded that combining alcohol with marijuana results in impairment even at doses which would be insignificant were they of either drug alone. Moreover, any combination of multiple psychoactive substances was associated with an increased risk of traffic injury. This was also reported by Gjerde and co-workers (2011).
Strengths and limitations
A major strength of this study was the large control group. This enabled positive findings of different drugs and combinations even though the prevalence of psychoactive drug use among drivers in normal traffic was low. Another major strength was that analyses of the blood and oral fluid samples were performed by a laboratory accredited for analyzing psychoactive drugs in forensic toxicology since 1996. Equivalent cut-off thresholds for oral fluid and blood were used to compare the data, strengthening the risk estimates. Biological samples from cases and controls were analyzed for a wide spectrum of psychoactive substances, thereby allowing for a better assessment of combinations of psychoactive substances.
The number of cases was relatively low and consequently the test power for analyses of various specific substances and combinations of substances was low. Six drivers were apprehended by the police for drunk driving and refused to participate in the study; the OR for alcohol would have been somewhat lower if these drivers had been included in the analysis. On the other hand, blood samples were obtained up to six hours after the injury, which implies that some cases may be false negatives, in which case the relative risk estimate may be biased downwards.
The use of blood samples from controls would be better than oral fluid if the participation rate is high. However, we expected that using oral fluid with high participation rate was better than using blood with low participation rate. A limitation when using oral fluid was the uncertainty in determining cutoff concentrations in oral fluid that were equivalent to those in blood. In our study, we used those determined by the DRUID project, which so far are the best documented equivalent cutoffs. By using prevalence determinations with equivalent cutoff concentrations, the effect caused by inter-individual variations in oral fluid to blood ratios for drugs are reduced compared to using average OF/B ratios for estimations of drug concentrations in blood based on concentrations in oral fluid.
Driver culpability was not considered and risk estimates might have been higher for culpable drivers. There were some statistical differences in the distribution of age and time in the two groups but these differences were considered minor limitations. A major limitation was the unknown number of non-responders in the case group, and the relatively small case group. The authors have no data suggesting the numbers of non-respondents in the sub-sample of drivers or the direction of any possible bias.
Implications
The high prevalence of alcohol in injured drivers and the significant increase in injury risk when driving under the influence of alcohol (whether alone or in combination with other substances) highlight the importance of curbing the extent of such driving. Various alcohol policy strategies are shown to be effective in reducing drunk driving and traffic crashes. These comprise not only low legal BAC limit and related sanctions [
32,
33] but also universal strategies as high taxes on alcoholic beverages [
34] and high minimum legal age for drinking or purchase [
35]. In Norway these effective strategies are employed to a larger extent than in most other OECD countries [
36], and therefore other prevention strategies should also be considered. Multi-component programs with community mobilization [
37] installation of ignition interlocks in drunk driving offenders’ vehicles [
38] and mass media campaigns combined with high visibility enforcement [
39] are strategies that are shown to be effective and therefore seem particularly relevant in this respect.
The latter type of strategy may also be relevant with respect to other substances than alcohol. An Australian study of ecstasy users found risk perception to be correlated with driving under the influence (DUI) of alcohol or other drugs [
40]. The users who reported DUI did not perceive their driving to be impaired and believed that there was a low accident risk and a low risk of being apprehended by the police. Therefore, more information about the actual increase in injury risk with DUI and more visible enforcement may encourage individuals not to drive under the influence of psychoactive substances. Driving under the influence of alcohol or might be associated with alcohol dependence [
41]. A review of studies on of interventions to reduce alcohol problems among ED- patients, found that interventions reduced alcohol related injuries [
42]. Such interventions might be effective if they were offered to alcohol positive injured drivers.
The few single non-alcohol substances for which it was possible to calculate risk estimates, were not statistically significantly associated with accident risk, but combinations of two or more such substances were. Some of these substances are prescribed medicines and when prescribing more than one psychoactive medicinal drug the patient should be adequately informed about possible risks associated with multiple psychoactive substance use. Yet, it should also be noted that combined use of several psychoactive substances, even prescription drugs, may sometimes stem from illegal or informal sources.
Comparing oral fluid with blood samples gives more accurate estimates of present impairment compared to urine when using equivalent cut-off limits, and the use of oral fluid is, therefore, recommended in future studies. As combinations of different psychoactive substances were prevalent, one can not screen for just alcohol or just THC, for example, and calculate the risk for these substances. A full drug screen analysis is needed to ensure that a significant risk increase in one substance is not due to combinations with other substances.
Conclusion
Prevalence of psychoactive substances was higher among injured drivers than drivers in normal moving traffic. Alcohol and stimulant drugs were particularly prevalent among drug positive injured drivers. The adjusted OR was high for alcohol combined with drugs; for alcohol alone, and for combinations of two or several non-alcohol substances. Various preventive measures should be considered to curb the prevalence of driving under the influence of psychoactive substances as these drivers constitute a significant risk for other road users as well as themselves.
Acknowledgements
The collection and analysis of oral fluid samples from controls was sponsored by the European Commission through the 6th Framework Programme funded integrated project DRUID (contract no. TREN-05-FP6TR-S07.61320-518404-DRUID) and the Norwegian Research Council. The collection and analysis of blood samples from cases was sponsored by the Norwegian Directorate of Health. This report reflects the authors’ views only. The funders are not liable for any use that may be made of the information contained herein.
Thanks to the staff at the emergency department Ullevål for all help with the data collection.
Thanks to the staff at Division of Forensic Medicine and Drug Abuse of NIPH for analysis of alcohol and drugs in samples of oral fluid and blood.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
STB contributed to the data collection of the case group, data analysis and drafted this manuscript. HG organized the data collection of the control group and analysis of biological samples. PTN organized the data collection of the case group and analysis of biological samples. All authors contributed to the design of the study and to the outline of the present paper, discussion of the results, the evaluation of the analytical results and data analysis, the writing of the manuscript. All authors approve the final version of the manuscript.