Cannabis and driving: a review of the literature and commentary (No.12)

Back to contents


This report has summarised available research on cannabis and driving. The report has included the main review documents for research published before 1994 (e.g., Hall et al., 1994; Robbe, 1994) as well as primary sources for research published from 1994 using keyword searches of relevant databases. This synthesis of research was directed to identify key research objectives to develop a rational transport policy for cannabis and driving.

It is apparent that cannabis is the most common 'illicit' drug. Indeed, there is some evidence to indicate an increasing trend in its availability and use in the general population. Thus, in terms of drug use and traffic safety, this would suggest that cannabis represents the major drug type to be addressed by transport safety policy.

However, there is not sufficient evidence indicating the percentage of drivers that operate a vehicle after consuming cannabis, particularly during the time period of any intoxicating effect. As a result, there is no precise estimate of the percentage of drivers exposed to cannabis as an accident risk factor. Indeed, it is problematic to estimate the extent of exposure independent of other risk factors associated with cannabis use such as alcohol. Moreover, the demographic group most frequently using cannabis already has the greatest a priori accident risk due to driving inexperience and factors associated with youth relating to risk taking, delinquency and motivation. These demographic and psychosocial variables may relate to both drug use and accident risk, thereby presenting an artificial relationship between use of drugs and accident involvement.

The recent developments and discoveries in pharmacology such as cannabinoid receptors and endogenous ligands are important and exciting. These and the increased understanding of the mechanism of action of cannabis will mean that new or improved methods of detection are likely. Most promising to date in terms of reliability, detection of recent consumption and practical application are methods for determining presence of metabolites in saliva and sweat. These developments and ongoing research also have implications for therapeutic drug development. New compounds based on cannabinoids will need careful evaluation to confirm their lack of psychoactive and psychomotor effects.

Evidence of impairment from the consumption of cannabis has been reported by studies using laboratory tests, driving simulators and on-road observation. The laboratory tests generally indicate acute impairment of memory, attention and psychomotor control. Both simulation and road trials generally find that driving behaviour shortly after consumption of larger doses of cannabis results in (i) a more cautious driving style; (ii) increased variability in lane position (and headway); and (iii) longer decision times. Whereas these results indicate a 'change' from normal conditions, they do not necessarily reflect 'impairment' in terms of performance effectiveness since few studies report increased accident risk. However, the results do suggest 'impairment' in terms of performance efficiency given that the increased compensatory effort resulting from cannabis use limits the available resources to cope with any additional, unexpected or high demand, events.

In conclusion, cannabis impairs driving behaviour. However, this impairment is mediated in that subjects under cannabis treatment appear to perceive that they are indeed impaired. Where they can compensate, they do, for example, by not overtaking, by slowing down and by focusing their attention when they know a response will be required. However, such compensation is not possible where events are unexpected or where continuous attention is required. Effects of driving behaviour are present up to an hour after smoking but do not continue for extended periods.

"With respect to comparisons between alcohol and marijuana effects, these substances tend to differ in their effects. In contrast to the compensatory behaviour exhibited by subjects under cannabis treatment, subjects who have received alcohol tend to drive in a more risky manner. Both substances impair performance, however, the more cautious behaviour of subjects who have received cannabis decreases the impact of the drug on performance, where the opposite holds true for alcohol." (Smiley, 1998, p. 19)

It is notable that the studies based on laboratory tests tend to indicate more effects of cannabis consumption than those using simulation and road observation methods. The higher incidence of effects under laboratory test conditions relative to the 'natural' conditions of simulation and road studies has been attributed to (i) reduced error variance from greater control of test conditions; (ii) higher task demand under novel test conditions; (iii) irrelevance or non-equivalence of laboratory test to component of driving; (iv) greater latitude for compensatory effort under 'natural' conditions; and (v) self-selection under 'natural' conditions not to be exposed to risk (e.g., not drive).

"It is exceedingly difficult to explain the disparity in results obtained by laboratory tests and in driving situations. Rather than try, it seems better for the moment to assume that both sets of results are valid for the particular circumstances under which they were obtained. It demonstrates, however, that performance decrements obtained under the artificial and non-life threatening conditions in the laboratory do not automatically predict similar decrements in driving situations that are closer to real-world driving." (emphasis added, Robbe, 1994, p. 66).

The greater propensity for cannabis effects under laboratory test conditions is somewhat paradoxical given that the laboratory tests have typically used smaller doses of cannabis than the simulation and road studies. It is also controversial since the limited number of studies and absence of demonstrable effects under natural driving conditions has impeded the development of transport policy regarding cannabis use. Whereas evidence of drug impairment under laboratory test conditions is not sufficient to provide an increase in accident risk, it does demonstrate cause for concern. This concern should then guide subsequent research under simulation and road conditions to investigate more valid evidence of impairment. Such efforts should be guided by relevant laboratory tests that relate to a model of driving, and use of a standard test methodology and reporting format for both simulation and road based research. This will provide a logical sequence of inquiry that can include both the replication of key findings, and the comparison of effects between a range of study designs.

Attempts to estimate the accident risk associated with cannabis use have relied on epidemiological evidence from accident involved drivers. Whereas this evidence has identified the presence of cannabis amongst accident involved drivers, accident risk can not be calculated given the absence of valid baseline data for cannabis detected in the non-involved population. Moreover, the presence of cannabis is often confounded by alcohol, as well as demographic and psychosocial risk factors associated with both drug and alcohol use. Current methodologies can only determine the presence of cannabinoids, but not evidence of impairment.

Thus, not only is it problematic to estimate the percentage of accident involvements associated with cannabis use alone, there is no evidence that impairment resulting from cannabis use causes accidents. Attempts to alleviate these problems by calculating risk of culpability for an accident (rather than the risk of having an accident) suggest that cannabis may actually reduce responsibility for accidents. It is evident that further epidemiological research is necessary. Such research must adopt a 'Grand Rapids' methodology of obtaining valid baseline data matched to positive cases, as well as including sufficient sample sizes and a valid operational definition of 'responsibility'. Such research may benefit from differentiating between accident types and accounting for relevant covariates including driver age and sex.

Much of the interest in cannabis as a potential accident risk factor is related to the concern about alcohol. Both alcohol and cannabis have an intoxicating effect that alters the psychological state of the individual. However, the mechanism of action and form of intoxication of these drugs are distinct. Alcohol may provide a useful metric to evaluate the effect of cannabis. Moreover, given the existence of a set legal limit for alcohol, research of the dose equivalence between alcohol and cannabis for performance relevant to accident risk may provide a method of determining a safety critical limit for cannabis. German research based on meta-analyses has concluded that 50% of performance is impaired at 11ng/ml THC, making this an equivalent level of intoxication to 0.08% BAC, although more recent and driving specific studies need to be compared with respect to effect size to confirm these suggested dose equivalences.

However, it is important not to use parallel reasoning between alcohol and cannabis to dictate the research agenda and transport policy for cannabis alone. Such reasoning is particularly inappropriate for medicinal applications of cannabis derivatives.

"There has developed an understandable but regrettable tendency to separate alcohol from other impairing agents and at the same time to enact tough drugs-driving legislation which remains firmly based on experience with alcohol. This is illogical, inappropriate and usually quite unenforceable. There is often pressure to define, for legal purposes, critical body fluid concentrations above which all would be impaired and below which no impairment would be demonstrable. At present, this is not possible. In addition to the considerably more complex pharmacokinetic and pharmacodynamic effects of most drugs compared with those of ethanol, there is also the proposition that therapeutic drugs, used for legitimate purposes, may improve the driving ability of certain patients despite their ability to impair performance normal individuals." (Starmer et al., 1988, p. 35-36)

One approach to deriving a legal limit for cannabis during driving has been to set the threshold to the level at which 50% of results show impairment. For alcohol, Berghaus showed a BAC of 0.073% corresponded to impairment on 50% of 923 performance measures examined. The corresponding threshold for THC was 11ng/ml. This is the closest estimate of dose equivalence to date, although there are recent, well-controlled studies which have not been included in such meta-analyses. A necessary research undertaking would be a thorough meta-analysis of results to date, using statistical measures of effect size related to dose.

An alternative is to specify a zero limit threshold, where any level of detected drug is prohibited. However, such an approach is premised on the philosophy that any drug which alters the state of the driver is inconsistent with the responsibility of the driver to operate the vehicle only when in an optimal state. This approach is associated with its own impracticalities of defining an 'optimal' state and deciding if reasons for impairment (i.e. deviation from the optimal state) other than drug use can be prosecuted (e.g., fatigue, poor driving skills, age related decline in capacity to drive etc.).

Ultimately, the direction of transport policy will be decided by an assignment of relative priorities. On one hand, any drug that affects the alertness and capability of a driver to safely operate a vehicle must be precluded. On the other hand, there are other factors such as alcohol which have a stronger association with accidents.

"Of the many psychotropic drugs, licit and illicit, that are available and used by people who subsequently drive, cannabis may well be amongst the least harmful. Campaigns to discourage the use of cannabis by drivers are certainly warranted. But concentrating a campaign on cannabis alone may not be in proportion to the safety problem it causes" (Robbe, 1994, p. 177).

The main conclusion from this report is that there is insufficient evidence of the accident risk associated with cannabis. Future research directed to the formulation of transport policy is required to resolve many key issues that remain unresolved in relation to cannabis and driving. However, it must be recognised that these issues may not be readily resolved given the ethical, legal, and technical impediments of the research domain (Hall et al., 1994).

This call for additional research was set forth by Robbe (1994) at the conclusion of his oft cited treatise on cannabis and driving:

"This dissertation should not be considered as the final word. It should, however, remain for a while as a point of departure for subsequent studies that will ultimately complete the picture of cannabis effects on driving performance" (p. 177).

It is now imperative that funding is made available to facilitate further research. However, such research must also be supported by mechanisms to accommodate legal and ethical requirements in this area. "In the meantime, cannabis users should be urged not to drive while intoxicated by cannabis, and they should be particularly warned of the dangers of driving after combining both alcohol and cannabis" (Hall et al., 1994, p. 50).

Back to top