Schaffer Online Library of Drug Policy Sign the Resolution
Contents | Feedback | Search
DRCNet Home
| Join DRCNet
DRCNet Library | Schaffer Library | Drugs and Driving 


Behavioral Effects of Alcohol and Cannabis: Can Equipotencies be Established?

H.-P. Krüger and G. Berghaus

Center for Traffic Sciences, University of Würzburg, Röntgenring 11, D-97070 Würzburg, Germany


In an extended review of the literature dealing with low alcohol effects, Krüger et al. (1990, 1994) introduced a new classification system for the study variables. Main characteristics of the new system were the ability to distinguish between automatic and control processes in performance areas and the explicit introduction of social effects (social moods, social behaviour). For each of the categories, hazard functions were calculated that showed loss of efficiency (and diminished performance) as alcohol concentration increases. Because the same classification system was used by Berghaus (1995) in his review of marihuana effects, it is now possible to compare hazard functions for both alcohol and marihuana effects and thus determine equipotential concentrations of alcohol and marihuana for the different classes of variables.


Alcohol and cannabis are quite different drugs, and their pharmacological characteristics are not comparable. Many studies have been conducted to specify and quantify cannabis effects on different aspects of behavior. These were reviewed by Burns & Moskowitz (1981), Chesher et al. (1984, 1986), Moskowitz (1985), Robbe (1994), and most recently by Berghaus (1995). The reviews show equivocally that, compared to alcohol, cannabis leads to a different structure of behavioral effects. Therefore, a differential comparison between both drugs is necessary that compares effects within classes of behavior defined as homogeneously as possible. For each behavior class, functions of equipotency must be determined according to the criterion, "producing the same effect on a specified behavior". A global evaluation of substance effects, which is needed with regard to traffic safety, follows from integrating those different functions which must be weighted with respect to the criterion, for example, safe driving.


Krüger (1990, 1993) and Krüger et al. (1990) reviewed the literature about alcohol effects. Only those studies meeting the following criteria were included: supplied empirical data from experiments controlled by placebo; used observables with face validity for safe driving; supplied information about the quantity of alcohol consumed; gave the time-interval between drinking and testing; gave blood alcohol concentrations during the test (by combining the last two pieces of information). To summarize results from different studies, it is necessary to aggregate observables into broader classes of behavior, especially performances. Eight classes were chosen: encoding and decoding of information, tracking, psychomotor tasks, visual functions, reaction time, attention tests, divided attention, and simulated or real driving tasks. Each reviewed study included from 5 to 20 different observations which were assigned to one of the 8 broader classes. The effect of alcohol (actual BAC at the testing time as compared to placebo) was characterized for each observation as +1 (better than placebo), 0 (no difference) or -1 (worse than placebo). When blood alcohol concentrations were not available, they were estimated by applying the WIDMARK formula (using the information about the consumed quantity of alcohol and the time of testing and assuming a standard body weight of 75 kg).

In addition to this data-analysis procedure, Krüger (1993) introduced a new technique into meta-analysis. Each observation within a study was taken as a "voter". If, for example, a significant deterioration in performance was observed at a BAC level of 0.05% the voter would have voted "no" for all smaller BAC values and "yes" for all BAC values equal to or greater than 0.05%. Or, in terms of survival analysis, a performance has "survived" up to the BAC value at which a significant deterioration in this performance was observed. Starting from this critical BAC value, the performance is looked on as being "dead". If, at a given BAC level, no deterioration was found, the performance has "survived" up to this level. At higher measurements, the performance is treated as a "missing value" as it is not clear at which BAC value the deterioration would have become significant.

Following this procedure, each observation yields a survival function which now can be integrated for (arbitratry) very small BAC classes. Comparing this integrated function to the number of observed results, a combined survival function is calculated. It starts at a BAC of 0% with 100% performances surviving, indicating that at this level none of the studies found an effect. At increasing BAC levels, more and more effects occur, resulting in a decline of this function. The steepness of the decline is expressed in the so-called hazard function. The steeper the function at a given BAC, the more likely that an additional increase in BAC will have deterioration effects.

Exactly the same procedure of collecting, selecting, excerpting, and analysing studies was used by Berghaus (1995). The studies were selected using the same inclusion and exclusion criteria, the assignment of observables to broader performance classes was identical, and the same classification was used to determine whether or not an effect was found. To calculate THC blood concentration at the time of testing, a standardized absorption and elimination curve of THC in the blood after smoking a 1-mg dose of cannabis (Sticht & Käferstein, 1995, in this same volume) was used (again taking the information about consumed quantities and time between smoking and testing). As with alcohol, survival functions were calculated.


The meta-analysis of alcohol effects is based on 197 published studies with a combined total of 1,245 single observations. The cannabis review is based onto 60 studies with a combined total of 1,344 reported observations. Integrating the results for all performance classes yields the survival functions in Figure 1. Both survival functions show:

  1. The higher the blood concentration, the more often negative effects are found, and
  2. Even small concentrations of either alcohol or cannabis may have effects on performance.

Figure 1
Survival Functions for Alcohol (right side) and Cannabis (left)

At a given abscissa value, the function should be read as "percentage of scientific observations which did not find significant deterioration effects". The arrows give the median of the functions. At a BAC value of 0.073% and at a THC value of 11 ng/ml, half of the reported effects were significant.

Both reviews observed only a few instances where performance under the influence of the substance was better than placebo. In addition, for both substances the following statements are valid:

  1. The same blood concentration has more deterioration effects during the absorption rather than the elimination phase.
  2. Infrequent or light users experience greater negative effects than heavy users.

Taking the global performance, 50% of all observed effects were negative in cases when a BAC value of 0.073% was reached. A plasma concentration of 11 ng/mL THC results in an equivalent deterioration. This value will be reached approximately 1 hour after smoking a standard cigarette containing 10 mg of cannabis. In Table 1 the global performance is split into the 8 different performance classes. For each class, a survival function was calculated. The concentrations of alcohol and cannabis were determined with 50% of the observations showing a significant deterioration effect.The rank orders of the medians are different for both substances, showing that the effect structures of alcohol and cannabis are quite different.

Table 1
For alcohol and cannabis, the number of observations (n) in each performance class and the median of the survival functions are given, sorted by the respective medians. In addition, the rank of the medians of alcohol is given.

Alcohol Cannabis
n class median % rank of median n class median ng/mL rank of median alcohol
74 simulated / real driving .064 1 73 tracking 6 5
57 en-/decoding .068 2 29 psychomotor tasks 8 6
116 divided attention .068 3 44 attention 9 8
213 visual functions .069 4 59 divided attention 11 3
88 tracking .070 5 25 visual functions 12 4
145 psychomotor tasks .073 6 113 simulated /real driving 13 1
108 reaction time .077 7 63 en-/decoding 15 2
122 attention .078 8 14 reaction time 15 7
923 global performance .073   420 global performance 11  

This is true not only for the medians but for the whole function. Figure 2 shows the equivalence curves for the two substances. The solid line formed by the global performance has to be interpreted as the reference for the comparison of alcohol and cannabis. Functions below this solid line indicate that cannabis has a deteriorating effect on this performance at lower concentrations as would be expected from the global equivalence. Functions above the global curve mean that the respective behavior is (relatively) more sensitive to alcohol than to cannabis.

Figure 2
Equivalence Curves for Alcohol and Cannabis for Four Performance Classes and the Global Performance

For each performance class and each substance, the percentiles 10, 25, 50, 75, 90 of the respective survival function were determined in terms of either BAC% or ng/mL THC. The pairs of percentile values were plotted into the figure (points, asterix, other symbols). These points were approximated by a smoothed function.


Actual driving and simulated driving are most sensitive to alcohol, followed by En-/Decoding. Driving is a systemic behavior for which, at a low sampling rate, different aspects of the situation must be recognized and integrated. The same holds true in the case of divided attention. In addition to the necessity to detect independent stimuli simultaneously, an appropriate reaction must be chosen. En-/Decoding is a high level cognitive function that involves complex activation of a series of mental processes. The sedating effect of alcohol heavily disturbs these integrative performances, whereas simple attentional processes (as measured by usual attention tests) are not as affected. Psychomotor skills, especially tracking but also simple reaction tasks, are only affected if alcohol concentration is very high. Thus, the effect structure of alcohol can be described as first disturbing higher cognitive processes, especially those that require integrative performances. Compared to those effects, the losses in psychomotor tasks and simple attentional processes are much smaller.

In contrast, cannabis first affects all tasks requiring psychomotor skills and continuous attention. Thus, tracking as a fast feedback loop between continuous visual inspection and spontaneous motor reaction to changes is very sensitive to short-term distortions in attention. On the other hand, integration processes and higher cognitive functions are not as time critical as motor reactions. A short attention lapse can be compensated for by increased activity afterwards. Or, as in the case of the integrative task of driving, the negative effects of these short distortions can be reduced by lowering the difficulty - and thus the time critical aspects - of the task. This interpretation would explain the often reported fact that drivers under the influence of cannabis drive at markedly decreased speeds (for example Robbe, 1994).

To summarize, a comparison of our two reviews corroborates the results of previous reviews. In addition, quantitative equipotency functions are given between blood concentrations of both substances where equipotency is defined as "equiefficacy on behavior". These functions differ in level and structure for different classes of behavior. Therefore, with respect to traffic safety, it is very difficult to decide which substance is more dangerous. The different effect structures of the substances must cause performance failures in different traffic situations. There is evidence that the types of accidents differ for alcohol and cannabis (Terhune et al., 1992). Thus, determing "equivalent danger" would imply a model of accident-prone situations. In addition, those variabilities in the equipotency functions must lead to differential effects with different types of drivers. A type would be a differential structure of abilities and weaknesses. Thus, even within the same class of behavior, the general equipotency of alcohol and cannabis may be modified by the characteristics of the driver.


Berghaus, G. (1995). A review and metaanalysis of cannabis effects with special emphasis on driving (in preparation).

Burns, M. & Moskowitz, H. (1981). Alcohol, marijuana and skills performance. In L. Goldberg (Ed.), Alcohol, Drugs, and Traffic Safety. Vol. 3. Stockholm: Almqvist & Wiksell.

Chesher, G.B., Bird, K., Crawford, J., Dauncey, H., Nikias, N. & Stramarcos, A. (1984). Further studies in the psychopharmacology of cannabis and alcohol: Acute and chronic effects. Research Grant Report Series, New South Wales Drug and Alcohol Authority, Sydney, Australia.

Chesher, G.B., Dauncey, H., Crawford, J. & Horn, K. (1986). The interaction between alcohol and marijuana: A dose dependent study on the effects on human moods and performance skills. Report No. C40. Federal Office of Road Safety, Federal Department of Transport, Australia.

Sticht, G. & Käferstein, H. (1995). Pharmacokinetic evaluation of published studies on controlled smoking of marihuana. of cannabis. Proceedings of the 13th International Conference on Alcohol, Drugs and Traffic Safety (in this same volume).

Krüger, H.-P. (1990). Niedrige Alkoholkonzentrationen und Fahrverhalten. Bergisch Gladbach: Unfall- und Sicherheitsforschung Straßenverkehr, Heft 78.

Krüger, H.-P. (1993). Effects of low alcohol dosages: A review of the literature. In H.-D. Utzelmann, G. Berghaus & G. Kroj (Eds.). Alcohol, drugs, and traffic safety - T92. Köln: TÜV-Akademie Rheinland. S. 763-778

Krüger, H.-P., Kohnen, R., Diehl, M. & Hüppe, A. (1990). Auswirkungen geringer Alkoholmengen auf Fahrverhalten und Verkehrssicherheit (Problemstudie). Abschlußbericht zum FP 8707 für die Bundesanstalt für Straßenwesen. Bergisch Gladbach: Bundesanstalt für Straßenwesen. Forschungsberichte Band 213.

Moskowitz, H. (1985). Marijuana and driving. Accident Analysis and Prevention, 17, 323-346.

Robbe, H.W.J. (1994). Influence of Marijuana on driving. Maastricht: Institute for Human Psychopharmacology.

Terhune, K.W., Ippolito, C.A., Hendricks, D.L., Michalovic, J.G., Bogema, S.C., Santinga, P., Blomberg, R. & Preusser, D.F. (1992). The incidence and role of drugs in fatally injured drivers. National Highway Traffic Safety Administration, October 1992 (DOT HS 808 065).