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

CHAPTER 7
Driving Response

Smiley (1986) produced a review article of all simulator and on road studies of cannabis and driving that had been published up to 1986. This has since been updated for a forthcoming World Health Organisation (WHO) report on cannabis to include all simulator and on road studies published up to 1997 (Smiley, 1998). It is notable that no simulator or road studies of cannabis have been published since 1994. Indeed, the most recent simulator study was published over 15 years ago in 1983. The 1994 treatise by Robbe is the most recently published road study of cannabis. The limited number of simulation and road studies of cannabis may be attributed to the ethical, legal and practical encumbrances inherent in this type of research.

7.1 Simulation Studies

A modified presentation of the tabulated summary of simulator studies reviewed by Smiley (1998) is shown in Table 11. ²² In addition to the main effects on driving performance reported for each study, the table includes information about the methodology of each study. Most of the studies have used a within subject design including a placebo condition (control). A number of these studies have used doses comparable to that indicated by Robbe (1994) to be typical (i.e., average) for regular cannabis users (300 µg/kg). It is generally these higher doses that have effects on performance. Moreover, all the effects were observed when the task was initiated within 1/2 an hour of consumption. Given the length of the simulated tasks, this suggests that the effects of cannabis were observed between 15 minutes and approximately 1 hour after consumption (Smiley, 1998). Indeed, the earliest simulation study did not observe any effect of cannabis at periods of 2.5 and 4 hours after consumption (Crancer et al., 1974, cited by Smiley, 1998).

The effects of cannabis can be described by three general performance characteristics. First, cannabis increases variability of longitudinal (speed and headway) and lateral control (lane position). This may be more evident under high workload and unexpected conditions such as curve negotiation and compensation for lateral displacement (wind gusts). These conditions resemble pursuit and compensatory tracking tasks. The increased variability may be attributed to specific acute effects of cannabis consumption including unstable motor control and reduced capacity to divide attention between task elements (e.g., maintain lane position and monitor speed) and sustain attention (vigilance) to task feedback (e.g., lane position deviation).

Second, decision times may increase to evaluate a situation and determine an appropriate response. This has been observed with respect to overtaking. The time taken to decide to attempt to overtake may increase. To the extent that this increase is considered in the decision to respond, it may also explain the effect of increased distance accepted to overtake. Although speculative, the case of cannabis resulting in fewer attempts to overtake under "cued" emergency conditions may indicate a strategy to avoid a complex decision process in deference to the external signal. Such a strategy would reduce cognitive workload. This is rational given that some subsidiary tasks have indicated less spare mental capacity after cannabis consumption. Strategies that do not impose upon limited resources are 'economical'.

Third, the style of driving performance after consumption of cannabis can be interpreted as cautious. Evidence of increased caution includes fewer overtaking attempts, larger distances required for overtaking, slower speeds, and larger headways. This caution can describe either the behaviour or the strategy of the driver. For example, cautious behaviour may arise without deliberation as a result of alterations in perception and control (e.g., distorted perception of time and space). Alternatively, a driver may decide upon a deliberate strategy to act cautiously by adopting a reduced threshold of acceptable risk. This decision may be motivated by the recognition of performance impairment. Of course, neither basis is mutually exclusive; changes in behaviour may be a result of both (unconscious) psychomotor impairment and (conscious) cognitive strategy.

A number of studies have included alcohol as a verum condition. Alcohol is a standard impairment agent that can be used to indicate the relative impairment effect of cannabis. With few exceptions, alcohol generally produces a greater impairment of performance effectiveness and efficiency than cannabis (for the range of doses administered). Moreover, the form of impairment from alcohol consumption appears to be qualitatively different than for cannabis. Notably, alcohol seems to result in a 'riskier' driving style (e.g., faster speeds) rather than one that is more cautious. There is also some indication that alcohol may affect oculomotor activity which is an earlier processing stage than distortion of perception.

There are a few inconsistencies in the results summarised in Table 11. For example, the earliest study (Crancer et al., 1969 cited in Smiley, 1998) found few effects of cannabis despite using similar doses to later studies. This has been related to the possibility that the potency of the cannabis may have been over-estimated such that the actual dose was 14-36% of what was indicated (Rafaelsen et al., 1973 cited in Smiley, 1998). Moreover, Smiley et al. (1981 cited in Smiley, 1998) found more evidence of cannabis impairment than did Stein et al., (1983 cited in Smiley, 1998) despite similar methodologies and doses of cannabis. ²³ This may relate to Smiley et al.'s, (1981) use of performance incentives that may have increased motivation as well as dose levels that produced greater than the normal 'high' for the subject sample. Also, Stein et al. (1983) tested cannabis over a shorter period and included a non-random subsidiary task (Smiley, 1998). This shorter period and the ability of subjects to anticipate (secondary) task demands may have limited the effects observed.

The reported effects of cannabis on simulated driving performance must be interpreted within the limitations of the methodologies represented by the range of simulator studies. Some of the evidence has been observed for tasks and environments that have very limited realism or interaction. The relevance of this evidence to actual driving is equivocal. Indeed, in the absence of a standard test protocol, including a specified driving task and simulation environment, it is not possible to directly compare results between studies. There is also no standard format for reporting results such as the relation between dose and levels present at the time of testing. ²⁴ This is compounded by differences between studies in dose, time elapses since testing, and measures analysed. There is further variation in terms of the type of subject used whereby the demographic and psychosocial variables may interact with the effects of cannabis. As a result, it is not possible to directly or accurately ascertain the effects of cannabis dose on driving performance, or the relative effect compared to alcohol from the existing set of published simulation studies.