The health and psychological consequences of cannabis use - chapter 5

THIS DOCUMENT HAS BEEN RESCINDED: Chapter 5.4 Psychomotor effects.

Page last updated: 1994

Please note the information in this publication may no longer be current but is retained on our website for historical or research purposes.

<< Previous Chapter | Contents | Next Chapter>>


5. The accute effects of cannabis intoxication


5.4 Psychomotor effects


A major societal concern about cannabis intoxication is its potential to impair psychomotor performance in ways which may directly affect the well-being of non-users of cannabis. The prototype outcome is an automobile accident caused by a cannabis user driving while intoxicated. It is well known that individuals who drive while intoxicated with alcohol are dangerous to others in proportion to their level of intoxication. Is there evidence that intoxication with cannabis produces impaired psychomotor performance of a nature and degree sufficient to warrant restrictions upon its use by automobile drivers? To what extent has cannabis intoxication contributed to road accidents?

Psychoactive substances typically have both acute and chronic effects on performance of a variety of tasks. Given the fact that most tasks of interest to researchers require effort and concentration, only those substances which enhance these very general abilities typically improve performance. Recreational drugs are usually valued for effects which remove the user from mundane concerns, produce relaxation, and enhance experiences which would normally interfere with concentration on a skilled task. Consequently, many societies enact restrictions on the use of such drugs, either during specific tasks such as motor vehicle driving, or at any time, as is the case with cannabis in most Western societies, and with alcohol in many Islamic societies.

The subjective effects of cannabis include feelings of well-being and relaxation, and sensory and temporal distortions which might be expected to decrease performance in situations where perceptual accuracy and attention are important. In deciding whether the recreational use of cannabis presents a danger to the user and others we need to consider two things: (1) the extent to which its use disrupts the performance of potentially dangerous tasks such as motor vehicle driving or the operation of machinery, and (2) the effect that the drug has on the user's compliance with restrictions upon its use. The second point refers to any disinhibitory effects of the drug which might predispose users to ignore prohibitions on driving, or may increase their willingness to take risks while intoxicated.

The risks of cannabis intoxication and driving will be assessed in the following way. First, laboratory evidence on the effects of cannabis on various psychomotor tasks will be reviewed. In the following review of this evidence, when a number of studies have produced similar results, only the most typical studies will be cited. (For a more complete review of such studies see Chait and Pierri, 1992). Second, the possible mechanisms of the psychomotor effects of cannabis will be briefly discussed. Third, the literature on the effects of cannabis on performance in driving and flying simulators will be briefly reviewed. Fourth, the experimental literature on the effects of cannabis intoxication on on-road driving will be reviewed. Finally, the limited epidemiological evidence on the contribution of cannabis to motor vehicle accidents will be considered.

Top of page

5.4.1 Effects of cannabis on psychomotor tasks


Muscle control. Standing steadiness (Kiplinger et al, 1971) and hand steadiness (Klonoff et al, 1973) are both adversely affected by cannabis. Finger or toe tapping speed does not appear to be reliably affected (Weckowicz et al, 1975; Evans et al, 1976; Milstein et al, 1975; Dalton et al, 1975), as only one study (Klonoff et al, 1973) found a decrement in finger tapping.

Reaction time. Simple reaction time does not appear to be reliably affected by cannabis. Some studies have reported decrements in mean reaction time (Borg et al, 1975; Dornbush et al, 1971), or the variability of reaction time (Braden et al, 1974), while others have found no difference (Evans et al, 1973). Choice reaction time tasks, in which the response is conditional not only upon the occurrence of a stimulus, but also the presence of some other discriminant (such as the pitch of a tone or the colour of a visual stimulus), have been administered to determine the effect of cannabis. In a number of these studies, reaction time was indeed slower after cannabis use (Borg et al, 1975; Block & Wittenborn, 1984; 1986), although there were some studies which found no change (Peeke et al, 1976; Block & Wittenborn, 1984). With only one exception (Low et al, 1973), errors in choice reaction time were not increased by cannabis.

Single tasks of manual dexterity. Pursuit rotor tasks, in which the subject attempts to follow a rotating target with a pointer, are generally performed worse after cannabis use (Manno et al, 1971; Manno et al, 1970), although studies employing regular users (Salvendy & McCabe, 1975; Carlin et al, 1972) have found no effect, suggesting that the regular users developed tolerance to the effects of cannabis. Other tracking tasks are generally not affected (Zacny & Chait, 1991; Heishman et al, 1989). Tests in which the subject must manipulate and accurately place small items (Dalton et al, 1975; 1976; Evans et al, 1973) are usually affected, while sorting tasks may (Chait et al, 1985) or may not (Kelly et al, 1990) be performed less well.

Concurrent tasks. Most concurrent task studies use one task which requires almost continuous attention, typically tracking, and one in which significant stimuli occur sporadically, often within a larger number of non-significant stimuli. The tasks are often referred to as the central and peripheral tasks respectively. The performance of concurrent tasks is almost always affected negatively by cannabis, although the effects on the component tasks are not consistent. Number or proportion of peripheral targets missed (MacAvoy & Marks, 1975; Marks & MacAvoy, 1989; Casswell & Marks, 1973; Moskowitz et al, 1972), proportion of hits (Moskowitz, Sharma & McGlothlin, 1972), false alarms (Chait et al, 1988, MacAvoy & Marks, 1975; Moskowitz & McGlothlin, 1974) or reaction time to peripheral targets (Perez-Reyes et al, 1988; Evans et al, 1976; Moskowitz et al, 1976) may suffer, but no interpretable pattern of decrements has emerged. It may be the case that while overall performance on concurrent tasks is decreased during cannabis intoxication, differences in the tasks used produce various patterns of effect. While there has been some speculation as to whether the effects of cannabis in concurrent tasks might be concentrated on the central or peripheral tasks (Moskowitz, 1985), no observed pattern has emerged to clearly support these conjectures.

Top of page

5.4.2 Possible mechanisms of psychomotor effects


Sensory disturbances. Reports of the subjective experience of cannabis intoxication include altered experience in all sensory modalities, as well as in the perception of space and time (Tart, 1970). Since almost all tasks of psychomotor performance include important visual and auditory components, sensory disturbances might well affect the ability to perform such tasks. Studies of the ability to discover embedded figures within complex designs have shown that this is impaired by cannabis (Carlin et al, 1972; Carlin et al, 1974; Pearl et al, 1973). Performance decrements due to cannabis in the Stroop colour naming test have been reported (Carlin et al, 1972; 1974), although it is not clear whether disturbed perception has any significant effect upon this task.

Central Nervous System depression. Both the toxic and behavioural effects of cannabis indicate that it acts as a CNS depressant, at least in moderate to high doses. It might seem reasonable to hypothesise that this general effect might contribute to slowed reaction times, inability to maintain concentration, and lapses in attention. This is certainly the case with alcohol and other CNS depressants. When compared to the relatively large and reliable depressant effects of moderate doses of alcohol, it is clear that this effect of cannabis is not the primary mediator of performance changes. It must be stressed, however, that high doses of cannabis would make this aspect of its action on psychomotor skills more important.

Motivational changes. A great deal has been written about the supposed effects of cannabis on human motivation. Hypotheses concerning the motivational effects of chronic cannabis use have been reviewed separately (see chapter 7.2). Cannabis users routinely report reduced desire for physical activity and increased difficulty of concentrating on intellectually demanding tasks such as reading for study (Tart, 1970). Studies designed to test the effect of cannabis on the willingness to perform laboratory "work" have found no striking decrements (Mendelson, 1983). This is consistent with comparisons of manual workers who used cannabis with those who did not (Rubin & Comitas, 1975; Stefanis et al, 1977). Indeed, the counter-argument that cannabis users can voluntarily compensate for some of the impairing effects of the drug has received experimental support (Cappell & Pliner, 1973; Robbe & O'Hanlon, 1993). As discussed below, motivational changes are surely important in decisions made outside the laboratory, but there appears to be no reliable evidence that motivational changes are responsible for any major proportion of the psychomotor effects of cannabis.

Top of page

5.4.3 Effects of cannabis on simulated driving and flying


Simulated driving tasks. As the previous sections have shown, there is considerable evidence that cannabis intoxication has some negative effects upon performance which become more pronounced with increasing task difficulty. Motor vehicle driving is a complex task, especially in conditions of heavy traffic or poor road or weather conditions, and as such, it might be expected to be adversely affected by cannabis. Simulated driving tasks require skills which are similar to those involved in driving, which can be performed under controlled laboratory conditions. When special efforts are made to simulate the performance characteristics of a car, simulations have two major advantages (Smiley, 1986). First, cannabis users an be tested after taking doses of cannabis which it would be unethical to use on the road. Second, they can be placed in simulated emergency situations which test their level of impairment in ways that would be impermissible on the road. The disadvantage of simulator studies derives from the difficulty of achieving sufficient fidelity to on-road driving tasks. One of the earliest studies by Crancer et al, (1969) found only that "speedometer errors" increased in simulated driving after cannabis use. In one of the more influential studies, Dott (1972) reported an apparent decrease in the willingness to take risks in simulated passing of another vehicle after cannabis use, while alcohol had the opposite effect. Alcohol also tended to hamper the subjects' response to stimuli signalling an emergency condition, while cannabis had little effect on this response. Both, however, increased reaction time to a more routine signal. A similar dissociation of the effects of alcohol and cannabis was reported by Ellingstad, et al, (1973) who found that cannabis did not appear to increase risk-taking, whereas alcohol did. Cannabis affected the ability to judge the time taken to pass another vehicle, while alcohol did not. Moskowitz et al (1976) found that alcohol altered the visual search patterns of subjects performing a simulated driving task, while cannabis did not. The alterations found with alcohol were, in theory, consistent with a reduced ability to scan for hazardous events, but no reliable difference in task performance was found with either drug.

Smiley (1986) critically reviewed the research on the effects of cannabis intoxication on simulated driving. She argued that the earlier studies which showed fewer effects on car control than later studies suffered because of their unrealistic car dynamics. Later studies which used driving simulators with more realistic car dynamics have shown impairments of lane control after cannabis use. Some of the studies have also shown reductions in risk-taking as manifested in slower speeds, and maintenance of a larger distance from the car in front in following tasks (Smiley, 1986).

Simulated flying. Janowsky et al (1976) found substantial increases in the number and magnitude of errors during a simulated flight after taking cannabis. These were principally in keeping the plane at the proper altitude and heading. Yesavage et al (1985) originally reported negative effects of cannabis on some components of a simulated flying task up to 24 hours after smoking, but this study did not include a control group. A later study (Leirer et al, 1989) which attempted to replicate this result with a control group found only an effect one to four hours after smoking. A third study which also included a control group (Leirer et al, 1991) again demonstrated decrements in the composite performance score up to 24 hours after smoking cannabis. Much has been made of these findings by critics of cannabis use, but the effects are very small and of uncertain significance for flying safety. Jones (1987) has argued that the use of cannabis by pilots in the 24 hours preceding flying may be more an indicator of poor judgment, rather than a cause for concern about the residual psychomotor effects of cannabis.

Top of page

5.4.4 Effects of cannabis on on-road driving


It is often remarked that the activity most often cited as dangerous when performed under the influence of recreational drugs - motor vehicle driving - is one of the least studied. Given the concern about the safety of the experimental subject in drug and driving experimentation, it is understandable that such studies have been relatively uncommon. A review by Nichols (1971) found that there were no well controlled observations of the effects of cannabis on driving performance. This situation changed with research commissioned by the Canadian Commission of Inquiry into the Non-Medical use of Drugs. A comprehensive report published by Hansteen et al (1976) showed that a moderate dose of alcohol (approximately 0.07 BAC) or THC (5.9mg) impaired driving on a traffic-free course (as measured by the number of times the lane-defining cones ("witch's hats") were struck). Driving speed was decreased after cannabis but not after alcohol use. Smiley et al (1975), using a different type of course, found that reaction time to signal stimuli was increased with the combination of cannabis and alcohol. Klonoff (1974) studied driving on a closed course, and in city traffic, after a placebo and two doses of smoked cannabis (4.9mg and 8.4mg THC). Closed course driving was scored by the number of cones hit on a precisely laid out path. Driving in traffic was scored by observation of eleven categories of driving skill, similar to those used in some driving tests. Driving on the closed course was impaired by both doses, as indicated by a higher proportion of subjects whose performance declined after cannabis use. Driving in traffic, however, while showing a trend toward poorer performance, was not significantly affected, and the effects of cannabis were much more variable. Sutton (1983) also found that cannabis had little effect on actual driving performance. Peck et al (1986) recorded performance on a range of driving tasks on a closed circuit on four occasions after the administration of placebo, up to 19mg of smoked THC, 0.84g/kg of alcohol, and the combination of both drugs. On most individual and derived composite measures, cannabis impaired performance. This study is important in that there was a high degree of concordance between objective performance measures (e.g. number of traffic markers hit during manoeuvres), subjective estimates of performance by the drivers, and ratings by police observers. However, the conclusion reached was that the effects of cannabis on driving performance were somewhat less than those of alcohol. Robbe and O'Hanlon (1993), have reported the methodology, but not the detailed results, of a study of driving in traffic. Their brief report suggests that their results also indicated little impairment of actual driving skills after cannabis. They speculated that since drivers were aware of their intoxication, they had successfully attempted to counter the impairment.

Overall, the effects of cannabis use on on-road driving have been smaller than the comparable effects of intoxicating doses of alcohol in the same settings (Smiley, 1986). The most consistent cannabis effect has been that drivers reduce their risk by slowing down; a finding that contrasts with the consistent finding that subjects typically increase their speed when intoxicated with alcohol. It is probably this compensatory behaviour by cannabis users that explains the comparatively small effects of cannabis intoxication in on road studies. For ethical reasons such studies have not been able to adequately test the response of cannabis intoxicated drivers to situations that require emergency decision, in which there is less opportunity to compensate for impairment. The few studies which have attempted to simulate this situation (e.g. by using subsidiary reaction tasks in addition to driving) have shown that cannabis intoxication impairs emergency decision-making (Smiley, 1986).

The small effects of cannabis on driving performance seem at odds with its effects on laboratory tasks requiring divided attention. Peck et al (1986) have pointed out, however, that the subtle performance effects of drugs in laboratory divided attention tasks may be poor predictors of driving performance. While the combination of performance abilities which is tapped by the typical divided attention task, such as concurrent pursuit tracking and visual discrimination, is plausibly related to driving, the tracking task is usually a much more difficult task than driving under normal conditions. Much more attention must be allocated to the central task in most divided attention tests, for example, leading to a substantial decrease in performance when drugs such as cannabis are taken. In addition, in the laboratory the subject is unable to vary a key task parameter, such as driving speed, in order to compensate for any perceived impairment. Hence, while laboratory divided attention tasks may be ideal for detecting small drug effects, they may over-estimate the effects of drugs on actual driving. It is not surprising then that many studies which have used both types of test have reported less effect on actual driving than on laboratory tasks or simulated driving.

Top of page

5.4.5 Studies of cannabis use and accident risk


While cannabis produces decrements in psychomotor performance in laboratory and controlled settings, it does not necessarily follow that these decrements will increase the risk of being involved in accidents. It may be, for example, that cannabis users are less likely to drive than drinkers because they are more aware of their intoxication. The survey evidence suggests that this is not the case. Several surveys (e.g. Dalton et al, 1975; Thompson, 1975; Klonoff, 1974; Robbe & O'Hanlon, 1993) have found that cannabis users are generally aware that their driving is impaired after using cannabis but the majority had driven, or would drive, after using cannabis, despite this recognition of impairment (Klonoff, 1974). This finding is consistent with observations on the recreational use of alcohol when driving (Smart, 1974).

Even if cannabis users drive when intoxicated it does not necessarily follow that they will be over-represented among drivers involved in accidents. It could be, for example, that cannabis users take special care and avoid risk-taking when driving while intoxicated. This possibility is difficult to investigate because there have been no controlled epidemiological studies conducted to establish whether cannabis users are at increased risk of being involved in motor vehicle or other accidents. This is in contrast to the instance of alcohol use and accidents, where case-control studies have shown that persons with blood alcohol levels indicative of intoxication are over-represented among accident victims (Holman et al, 1988).

In the case of cannabis, all that is available are studies of the prevalence of cannabinoids in the blood of motor vehicle and other accident victims (see McBay, 1986 for a review). Most often these have been retrospective studies of the prevalence of cannabinoids in blood tested post-mortem, which have found that between 4 per cent and 37 per cent of blood samples have contained cannabinoids, typically in association with blood alcohol levels indicative of intoxication (e.g. Cimbura et al, 1982; Mason and McBay, 1984; Williams et al, 1985). Zimmerman et al (1983) have reported similar revalence data on blood cannabinoid levels among Californian motorists tested because of suspicion of impairment by the Highway patrol. Soderstrom et al (1988) have conducted one of the few prospective studies among trauma patients rather than accident fatalities, which showed a high prevalence of bloods positive for cannabinoids (35 per cent).

These studies are difficult to evaluate for a number of reasons. First, in the absence of information on the prevalence of cannabinoids in the blood of non-accident victims, we do not know whether persons with cannabinoids are over-represented among accident victims (Terhune, 1986). Although a prevalence of 35 per cent may seem high, this is of the order of the prevalence of cannabis use among young males who are at highest risk of involvement in motor vehicle and other accidents (Soderstrom et al, 1988). Second, there are major problems in using cannabinoid blood levels to determine whether a driver or pedestrian was intoxicated with cannabis at the time of an accident (Consensus Development Panel, 1985). The simple presence of cannabinoids indicates only recent use, not necessarily intoxication at the time of the accident (see above pp35-36). Third, there are also serious problems of causal attribution, since more than 75 per cent of drivers with cannabinoids in their blood also have blood levels indicative of alcohol intoxication (McBay, 1986). On the basis of the available evidence, it is accordingly difficult to draw any conclusions about the contribution that cannabis intoxication may make to the occurrence of motor vehicle accidents (Terhune, 1986).

One approach that has been used in an attempt to get around the absence of data on the prevalence of cannabis use among drivers not involved in accidents has been to perform "culpability analyses" (Terhune, 1986). In such analyses, decisions are made as to which drivers killed in fatal accidents are culpable (i.e. responsible for the accident). Drivers with no alcohol or other drugs in their blood are then used as the control group in analyses of the relationship between the presence of drugs in blood and degree of culpability. These studies have their problems: the culpability of the drug-free drivers is usually high thereby reducing the ability to detect an increase in culpability among drivers with alcohol and cannabis; different studies use different criteria for deciding that when a driver was intoxicated with cannabis; and as a consequence, different studies have produced very different estimates of the relationship between cannabinoids in blood and driver culpability (although most have shown an increased culpability for drivers with intoxicating levels of alcohol in their blood). As Simpson (1986) concluded after reviewing the culpability literature: "the results are mixed and inconclusive" (p28).

Top of page
Gieringer (1988) used a different approach to circumvent the absence of data on the prevalence of cannabinoids in drivers not involved in accidents. He used data from a National Institute of Drug Abuse (NIDA) household survey of drug abuse in the United States to estimate the proportion of all drivers who might be expected to have blood and urine samples positive for cannabinoids. On the basis of these data, he estimated that cannabis users are two to four times more likely to be represented among accident victims than non-cannabis users, and that cannabis users who also used alcohol were even more likely to be over-represented among the victims of motor vehicle accidents.

Gieringer's inference about the risks of combining alcohol and annabis when driving receive some support from the studies of Mason and McBay (1984) and Williams et al (1985). Mason and McBay estimated that at most one driver in their series of 600 drivers killed in single-vehicle accidents was significantly impaired by cannabis use alone, compared with between nine and 28 drivers who were impaired by marijuana and alcohol, and 476 drivers who had blood alcohol contentrations (BACs) greater than 0.10. Williams et al (1985) investigated the relationship between alcohol and cannabis use and driver responsibility for fatal accidents (as judged from police investigations of each accident) involving young men in California. Using the small drug-free group as the comparison, they found that both alcohol (OR=4.7 [95 per cent CI: 2.1, 10.3]) and alcohol and marijuana in combination (OR=8.6 [95 per cent CI: 3.3, 22.2]) significantly increased the odds of the driver being adjudged to be responsible for the accident. Marijuana-only drivers, however, were less likely to be adjudged responsible for their accident (OR=0.5 [95 per cent CI: 0.2, 1.3]), although numbers were small (N=19).

There is also indirect evidence that cannabis use produces an increase in the risk of accidents, from surveys of self-reported accidents among adolescent drug users. Two such surveys have found a statistically significant relationship between marijuana use and self-reported involvement in accidents, with marijuana smokers having approximately twice the risk of being involved in accidents of non-marijuana smokers (Hingson et al, 1982; Smart and Fejer, 1976).

More direct evidence of an association between cannabis use and accidents is provided by two epidemiological studies, one of cannabis use and mortality (Andreasson and Allebeck, 1990), and the other of cannabis use and health service utilisation (Polen et al, 1993). Andreasson and Allebeck reported a prospective study of mortality over 15 years among 50,465 Swedish military conscripts. They found an increased risk of premature mortality among men who had smoked cannabis 50 or more times by age 18 (RR=4.6, 95 per cent CI: 2.4, 8.5). Violent deaths were the major cause of death contributing to this excess mortality, with 26 per cent of deaths being motor vehicle and 7 per cent other accidents (e.g. drownings and falls). The increased risk was no longer statistically significant (RR=1.2 [95 per cent CI: 0.7, 1.9]) after multivariate statistical adjustment for confounding variables such as anti-social behaviour, and alcohol and other drug use in adolescence (Andreasson and Allebeck, 1990), reinforcing Gieringer's suggestion that the combination of cannabis and alcohol may be the important risk factor for accidents.

Polen et al (1993) compared health service utilisation by non-smokers (N=450) and daily cannabis-only smokers (N=450) screened at Kaiser Permanente Medical centres between July, 1979 and December, 1985. They reported an increased rate of medical care utilisation by cannabis-only smokers for respiratory conditions and accidental injury over a one to two-year follow-up. There was also an interaction between cannabis and alcohol use, in which cannabis users who were the heaviest alcohol users showed the highest rates of utilisation. This result is suggestive but minimally informative about the risks of motor vehicle accidents, because all forms of accidental injury were aggregated.

Top of page

5.4.6 Conclusions on cannabis and driving


There is no doubt that cannabis adversely affects the performance of a number of psychomotor tasks, an effect which is related to dose, and which is larger, more consistent and persistent in difficult tasks involving sustained attention. The acute effects on performance of typical recreational doses of cannabis are similar to, if smaller than, those of intoxicating doses of alcohol. Alcohol and cannabis differ in their effects on the apparent willingness of intoxicated users to take risks when driving, with persons intoxicated by cannabis engaging in less risky behaviour than persons intoxicated by alcohol.

While cannabis produces decrements in performance under laboratory and controlled on-road conditions, it has been difficult, for technical and ethical reasons, to establish conclusively whether cannabis intoxication increases the risk of involvement in motor vehicle accidents. There is sufficient consistency and coherence in the evidence from studies of cannabinoid levels among accident victims, and a small number of epidemiological studies, to infer that there probably is an increased risk of motor vehicle accidents among persons who drive when intoxicated with cannabis. A crude estimate of the risk is of the order of two to four times for persons driving under the influence of cannabis. This increased risk may be largely explained by the combined use of cannabis with intoxicating doses of alcohol. Further research is required to elucidate this issue, although it will not be easily resolved because of the technical obstacles to such research. 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 alcohol and cannabis use.