An assessment of the relationship between sedatives and driving accidents requires the survey of literature dealing with: (1) the effects of sedatives on actual driving behaviors, (2) the epidemiological studies of sedatives and traffic accidents, and (3) the physiological, psychological, and behavioral effects of sedatives on factors related to driving.

Only a few studies have tested the effects of sedatives either in a simulator or in the field. Loomis and West () tested eight subjects in a driving simulator from 1 to 6 hours after they were given various drugs. The simulator consisted of an automobile steering wheel and brake accelerator pedals arranged as in a standard automobile. The steering wheel operated a model car placed on a moving belt 150 ft. long and 30 in. wide with an opaque l-in. strip running down it lengthwise, which simulated the road bed. The strip was shifted randomly, moving smoothly from side to side as the belt advanced. Accelerator and brake pedals actuated and controlled the rate of belt movement, and the steering wheel controlled the position of the model car. A light source placed 14 in. above the car was capable of producing an amber, red, or green light. The subject was required to keep the car centered on the road bed and to respond to the lights by depressing the accelerator pedal when the green light appeared, releasing the pedal when an amber light appeared, and depressing the brake pedal to stop the belt when a red light appeared.

Response measures included braking time following the appearance of a red light, time required to release foot pressure on the accelerator pedal at the appearance of an amber light, and a steering score, which measured the cumulative time during which the car was not “centered on the road.”

The experimenters describe the procedure as follows. ‘Tests were performed at 1 (trial #1) and 2 (trial #2) hours after the administration of the drug. At 24 to 3 hours after the initial medication, standard lunch … was consumed. One hour later the driving test was conducted following which the second dose of medication was taken by mouth … The five drugs were: placebo (com-starch) 200mgm; secobarbital sodium 100mgm; chlorpromazine hydrochloride (Thorazine) 50mgm; meprobamate (Equanil) 400mgm; and phenaglycodol (Ultran) 300mgm.” It is not clear from this description whether the drug amounts cited were the sum of the first and second doses or whether this amount was given at each of the two drug administration times.

Secobarbital, chlorpromazine, and meprobamate produced impairment on the tests. The decrease under secobarbital was the largest, with a 115.6 percent decrement apparent at the fifth hour after initial drug administration. However, it is not clear whether this occurred after 200 mg (a high dose) or after 100 mg of secobarbital.

The Loomis-West simulator represents a part-task simulator in that it measures only some of the tasks related to driving. The simulator does represent some of the demands of driving in that it requires the subject to perform multiple acts simultaneously.

Another study which was carried out on a closed course was conducted by Betts et al. (). They examined the effects of 150 mg sodium amytal on three vehicle-handling tasks. These were a weaving test, a parking test, and a gap estimation test requiring subjects to estimate the gap between two traffic cones through which they could drive. Fifty men and 50 women served as subjects. The results indicated that the male subjects increased their failures in gap estimations under the barbiturate, and the women decreased their distances from the curb in the parking test, but increased their successes in gap estimations. The qualitative difference in performance between men and women on the gap estimation may indicate a heightened risk-taking attitude in men and a lowered risk-taking attitude in women under the drug condition.

This is a typical closed-course study designed to determine the effects of drugs on driving behavior. The results are suggestive of impairment in certain areas, but they cannot be extended to real world driving situations. The closed course lacks all of the environmental stimuli which place demands on driving. Closed courses are necessary for conducting safe tests, but they reduce the validity of the results.

Flying simulators have also been used to determine the effects of sedatives on complex behavioral tasks. One such study is by Harper and Kidera (). The authors tested 30 pilots in a twin turbojet flight simulator. The measures included the following parameters: (1) airspeed, (2) altitude, and (3) ILS glide slope and localizer indicator. In addition, an observer scored the subject on “procedures and techniques.” The subjects also filled out a questionnaire which was designed to measure feelings of alertness, fatigue, and quality and soundness of sleep the previous night.

The 30 subjects were divided into three groups, with each group receiving one of the three treatments: (1) placebo, (2) glutethimide (500 mg), or (3) flurazepam (30 mg). The subjects were trained for 4 hours, and then 1% hours of performance data were recorded as a baseline. These data were compared to those obtained after the subjects were given the dosages for two consecutive nights.

The tests were given approximately 12 hours after the second dose. The simulator observations were presented only as improvement, decrement, or no change, as compared to the baseline data. No quantitative scores were provided. Six of the ten subjects under glutethimide and four of the subjects under flurazepam showed decrements. No subjects showed no change, and five showed improvement. The results from the subjective questionnaire for glutethimide indicated seven subjects to be feeling worse than they did when their baselines were recorded. Three under placebo reported their feelings to be worse than their baseline days. The flight data were scored in the same way as the simulator observer data. Four subjects under glutethimide, three under flurazepam, and four under placebo showed ‘decrement” performance in their final approach.

The authors’ summary statement is, “Using flight recorder data as the objective measurement, flurazepam (30 mg) and glutethimide (500 mg), after two nights’ dosage, had no apparent effect on flight performance twelve hours after use.” This is based largely on the lack of differences between the drug and placebo groups. However, this conclusion may be unwarranted since the study does suffer from some methodological problems. The apparent lack of quantification does not indicate what the various degrees of impairment were in the different groups; and the use of the simulator observer and the subjective rating scale does not allow a clear-cut determination of the drugs’ performance effects per se.

Two other studies with flying simulators which did not contain these shortcomings were conducted by McKenzie and Elliott () and Hartman and McKenzie (). In the first study, subjects were tested under 200 mg secobarbital on a flying simulator. The simulator required division of attention and placed multiple monitoring demands on the subject. This is analogous to the driving situation, in which it is necessary to keep the car centered in a lane and simultaneously monitor other environmental signals that may occur. This simulator, therefore, did require use of some skills necessary for driving. It consisted of a cockpit-like console with stick, rudder, and throttle controls. A panel containing four instruments for airspeed, turn, bank, and engine rpm was mounted in front of the subject. The dial needles were programmed to run independently, but simultaneously. The subjects were required to monitor the dials and use the controls to keep the needles in the center position.

The testing occurred 10 hours after ingestion of the drug and lasted for 12 additional hours, with 15-second rest periods following each minute of operation. The results clearly indicated that multiple tracking was degraded 10 hours after ingestion and performance impairment continued to show for the remaining 12 hours.

It is interesting to note that the impairment lasted well beyond the peak pharmacological effect of the drug. From this study it is apparent that the behavioral impairment does not linearly follow the blood levels of secobarbital. In assessing the effects of drugs on driving behavior, it is valuable to keep this in mind.

The second study, by Hartman and McKenzie, used a similar flying simulator and tested four subjects 10 hours after drug administration. The test lasted for 4 hours and was performed under 200 mg, 100 mg, or placebo doses. Performance was degraded under the high dose, but not the low dose. One interesting aspect of the study was the enhancement of performance under placebo. This was accompanied by the subjects’ subjective reports to the effect that they slept as well after receiving placebos as after receiving 200 mg secobarbital. This indicates a strong placebo effect and raises questions regarding the reliability of subjective reports.

It is important to note that the impairment was sustained for 14 hours after drug administration. These residual or “hangover” effects are important for driving. If impairment is exerted for a considerable period of time, then the likelihood of involvement in accidents is increased, even though the blood levels of the drug may not be high. The hangover effect is also evident in other studies.

One example of this hangover effect is a study by Walters and Lader (). They administered 100 and 200 mg of sodium butabarbitone and 5 and 10 mg of nitrazepam to 10 subjects in a balanced double-blind design. The tests included EEG recordings at rest and also during an auditory reaction time task, a key-tapping rate test, a digit symbol substitution task, and a subjective ratings test for alertness. Testing took place 12 hours after drug administration. The results showed that “many tests were affected significantly by the drugs.” This indicates that drug effects are demonstrable for extended periods of time. As the authors stated,” … we must remain aware that, although these hypnotics lessen the distress of insomniac patients, psychological impairment and electrophysiological changes are inevitably left the next morning. Indeed, it is unrealistic to expect any adequate hypnotic drug to be devoid of pharmacological effects on wakening.” These residual effects are factors which must be faced in research related to drugs and driving.

While the above simulator studies are important in yielding information regarding the effects of drugs, the extension of the results to the real world has to be done with caution. A simulator only partially taps the demands of operating a real vehicle in real situations. The differences between drug effects on simulator performance and on flying an airplane are shown in the study by Billings et al. ().

In this experiment, five highly experienced pilots were tested under 0, 100, and 200 mg secobarbital, both in a Cessna Model 172 airplane and in a Link-Singer GAT-1 simulator. Before testing, each subject was permitted “to familiarize himself with the vehicle and to practice instrument approaches until he was satisfied with his performance.” On test days, the pilots conducted two instrument flights after ingesting the secobarbital. After the flights, duplicate studies were performed on the same pilots under similar conditions using the simulator. The responses recorded were tracking performance in two axes and airspeed control. The average decrement at the 100- and 200-mg doses, as compared to the placebo, were 17 percent and 26 percent in the simulator and 2 percent and 14 percent in the airplane. The manual control decrement at the 100-mg dose in the airplane was not different from placebo. At the higher dose level, little difference was observed for tracking in the first flight, but significant differences were observed in the second flight. The differences were significant for both flights at both doses in the simulator. The paper does not state clearly at what time interval after the administration of the drug the testing took place. Also, the total time of testing is not specified. It appears that testing may have taken place immediately after administration, in which case the peak pharmacological effects may not have been reached during testing.

However, it is clear from the results that there are differences in performance between the simulator and the airplane. ‘This study serves as a reminder that a simulator–and this may be true of most Simulators–is not an airplane. ‘Flying’ it demands a different strategy than that utilized in flying the airplane for which it is a surrogate, and proficiency in the one vehicle does not imply equal proficiency in the other.” A similar statement can be extended to driving and driving simulators.

The literature on epidemiological evidence which implicates sedatives with driving accident involvement is scant. It deals largely with coroners’ reports, and published materials generally only indicate the number of driving fatalities which have occurred under the drug’s influence or report the number of persons with drugs present while they were driving.

A typical study dealing with sedatives is one by Gupta and Kofoed (), in which they assayed urine and blood samples from persons cited for driving while under the influence of drugs. They report the number of persons showing the presence of barbiturates and no alcohol. (For 1964, 18 such cases were cited in Ontario, Canada.)

There are a number of problems in using such data to implicate sedatives in causing driving accidents. The numbers of drivers apprehended under drugs do not indicate the nature of the relationship between the drugs and accident involvement. To determine this, it is essential to know (1) the number of drivers involved in accidents under the influence of sedatives and (2) the number of individuals who drive under sedatives in the location where the accident took place at the same time of day, but who are not involved in accidents. Only a comparison of the accident-involved drivers and the at-risk population will yield an index that implicates sedatives in causing driving accidents. Such studies are available for alcohol, but are sorely lacking for sedatives. Further, the Gupta and Kofoed study did not indicate what the drug blood levels were. It is important to know what the relationship is between drug levels and apprehension or accident involvement. It has been mentioned earlier that impairment under barbiturates is sustained for long periods of time. Thus, it is important to relate the residual effects of the drugs to driving accident causation. The existing epidemiological studies on sedatives suffer from many methodological problems, ranging from the populations sampled to the lack of quantitative data on drugs in body fluids.

The data from studies of the effects of sedatives on driving-related skills also provide information regarding drug involvement in traffic accidents. Such behavioral studies can focus on various aspects of driving skills. Driving itself is a complex task requiring acquisition of information, processing of information, and the execution of responses based on that information. These stages require the performance of such tasks as monitoring the environment for signals; tracking, or keeping the car on the road; and responding to the stimuli by braking, accelerating, or steering. There are a number of papers which deal with aspects of these driving-related skills.

Drugs could affect sensory mechanisms, which in turn could produce deleterious effects on driving behavior. There is a lack of literature on the effects of sedatives on visual functions. However, there are a number of papers dealing with the effect of sedatives on oculomotor functions, such as the one by Holzman et al. (). They examined the effects of single doses of chlorpromazine (0.667 and 1.33 mg/kg body weight), Valium (0.071, 0.142, and 0.284 mg/kg body weight), and secobarbital (100 mg) on smooth-pursuit eye tracking. Five male subjects (four for the barbiturate) were required to follow a pendulum that moved with a frequency of 0.4 Hz. The horizontal eye movements were measured by silver-silver chloride skin electrodes applied to the outer canthus of each eye and recorded on a dynograph. The scoring procedure required two scorers to independently classify the tracking as qualitatively “normal” or “deviant.” One quantification of the scores was the number of times the eyes stopped their pursuit of the target.

The results showed that only secobarbital disrupted eye-tracking performance. Two of the four subjects under the barbiturate replaced their ocular pursuit with saccades. One of the subjects who showed no qualitative disruption of his tracking pattern was given the higher dose of 130 mg of secobarbital. This dose was found to produce eye-tracking disruption which lasted 24 hours.

These results are suggestive of disruption of higher order processes. However, the study does contain some methodological problems. The number of subjects is too small to allow generalizations about the results. Also, the qualitative measures relied on the judgment of scorers, and even the quantitative measures used in the study are inadequate for describing the deviation of eye movements from smooth pursuit. Manual inspection of the data is, of course, time-consuming and limits the size of the study.

There are several studies dealing with the effects of sedatives on vigilance behavior. Vigilance is important in the operation of machinery or automobiles. Any unchanging environment can produce impairment in skills performance. Kopriva et al. () tested 90 professional drivers under 150 mg/70 kg body weight pentobarbital. Auditory signals were presented to the subjects at irregular intervals via earphones, such that the signals differed in spatial location (left, right, and midline). The subjects were asked to ignore the midline stimuli (50 percent of the presentations) and to depress a button when the other two kinds of stimuli occurred. Each subject “was also informed that the course of physiological functions after drug application permits to determine… whether the drug has any effect or not and that this information would be transmitted to him by a light signal in time, so that he would be prepared to counteract any disturbing effect of the drug.” Pentobarbital increased the number of misses, and the cue did not affect performance, but false alarms increased with the cue condition under this drug. The authors noted that this may be related to a shift in discrimination criteria. However, due to the lack of signal detection analyses, it is not clear whether the poorer performance under the drug is related to set or criterion shifts. It is nevertheless clear that pentobarbital affects signal detectability.

Psychomotor performance under sedatives has been studied by various investigators. One such study is by Goodnow et al. (). They tested 30 male subjects on placebo and 100 mg pentobarbital, using a crossover design. The test battery contained a number of tests including: (1) tapping speed, using a telegraph key; (2) auditory reaction time, in which the subject was required to depress a key at the presentation of an auditory stimulus; (3) naming of opposites, in which the time required to name the opposite of a common word was recorded; and (4) memory for digits, which was simply measurement of backward digit span. the pentobarbital degraded performance on all four of the tests 4 hours after administration. There was a trend toward poorer performance (not statistically significant) 14 hours after the drug dose.

This study represents a sound experiment, in that the design controlled for many sources of variance. Learning effects were controlled by performing the statistical analyses on the differences between the before and after tests for the placebo and drug treatments.

As has been mentioned earlier, driving is essentially a dual-task function, requiring the detection of environmental signals and tracking to keep the car on the road. Two examples of papers on tracking under sedatives are those by Borland and Nicholson () and Shroeder et al. ().

Borland and Nicholson tested seven subjects under placebo and 200, 300, and 400 mg of heptabarbitone on a tracking task. This required subjects to position a spot inside a randomly moving circle displayed on an oscilloscope. An error signal proportional to the distance between the spot and the center of the circle controlled the difficulty of the task by modulating the mean amplitude of the movement of the circle. Each trial lasted 10 minutes with only the last 500 seconds used as the performance measure. Each subject was trained until a plateau level was reached. Predrug trials were run, and post-drug trials were run at various intervals after administration. The various drug doses were administered in random order.

The results showed that heptabarbitone produced decrements in performance at 10 hours after the 200-mg dose, at 10- and 13-hour intervals after the 300-mg dose, and at 10-, 13-, 16- and 19-hour intervals after the 400-mg dose. The performance impairment was dose related. It is interesting to note that blood levels of heptabarbitone did not correlate with individual performance.

The study by Schroeder et al. tested the effect of placebo, secobarbital (100 mg) and d-amphetamine (10 mg) on tracking performance. Ten subjects were tested in each of the drug groups. The tracking performance was tested under two conditions: (1) while the subject sat stationary and (2) while the subject underwent angular motion. The tracking test consisted of an aircraft localizer/glide-slope indicator with the vertical needle being deflected to the right or left of center by a sinusoidal forcing function. The subject was required to keep the needle in the center by compensatory movements of a joy stick. After the subjects were trained on the task, there was a predrug test, and three post-drug sessions were conducted at 1, 2, and 4 hours after administration of the drugs. The response measures included tracking errors under both static and dynamic conditions and the number of nystagmic eye movements during the dynamic condition. During the static condition, secobarbital did not affect tracking scores. Amphetamine was found to improve scores over control for the 2- and 4-hour post-drug sessions. During angular acceleration, secobarbital produced more tracking errors and increased vestibular nystagmus as compared to both the control and amphetamine groups for all post-drug sessions.

Both of the above studies give examples of deleterious effects of sedatives on tracking performance. The second study did not demonstrate any effects under the stationary condition, but it would be worthwhile to examine the effects of the drug at more delayed intervals after drug administration.

Since alcohol is to a large extent consumed in conjunction with other drugs, it is important to evaluate the combined effects of sedatives and alcohol on driving skills performance.

Examples of this kind of investigation are presented in papers by Sellers et al. () and by Mould et al. (). In the first study, Sellers et al. tested the effects of placebo, alcohol, and chloral hydrate, singly and in combination, on changes in heart rate, arterial pressure, skin temperature, simple and complex reaction times, a tracking task, and a vigilance task. Five male subjects were tested under the following drug conditions: (1) placebo; (2) alcohol (0.5 g/kg body weight); (3) alcohol (0.5 mg/kg body weight) given ½ hour after chloral hydrate (15 mg/kg body weight); (4) chloral hydrate (15 mg/kg body weight); and (5) alcohol (0.5 g/kg body weight) after prior treatment with chloral hydrate (15 mg/kg body weight) for 7 days, with the last sedative dose being given 12 hours before alcohol.

Compared to the control conditions, alcohol was found to increase heart rate between 4 and 1 hour after administration. Chloral hydrate combined with alcohol increased the cardiac rate at ½ hour after administration more than did alcohol given alone or chloral hydrate given alone. Chloral hydrate followed by alcohol caused greater increases in skin temperatures than the other treatments. There was no drug effect on the reaction time tasks. Alcohol was found to degrade the tracking and vigilance tasks, but chloral hydrate given alone had no effect on these tasks. The decrement was pronounced (and larger than for alcohol alone) when chloral hydrate was given in combination with alcohol on both the tracking and vigilance tasks.

There are a number of methodological problems with this study. For example, the size of the sample is too small and the specifics of the experimental design are not clear. However, it is clear that alcohol and chloral hydrate in combination do cause greater performance deficits than does either substance taken alone.

Mould et al. tested six subjects on the effects of glutethimide (250 mg) alone and in combination with alcohol on reaction time, tracking, and a finger-tapping task. Glutethimide alone was found to affect all the behavioral tests. Glutethimide consumed simultaneously with 100 ml of vodka was found to increase the reaction time responses, but did not affect the other tests (nor did alcohol alone). This is in contrast to the previous study, where chloral hydrate in combination with alcohol had no effect on reaction time tests, but did affect the tracking task.

One interesting finding was that, when chloral hydrate and alcohol were given simultaneously, blood alcohol concentration was significantly higher (11 percent) at 105, 135, and 165 minutes after drug administration than when the alcohol was ingested alone. In contrast, a second experiment which was part of the same study tested phenobarbital (60 mg) given in combination with alcohol (50 ml vodka). A decrease in the blood alcohol concentration was shown at 30 and 90 minutes after administration. This study also suffers from the use of a small sample size and an ambiguous experimental design.

An example of a study which investigated the effects of drugs in combination (other than alcohol) on behavioral performance is one conducted by Dalton et al. (). They tested 12 subjects under marihuana (25 mg THC/kg body weight) and secobarbital (150 mg/70 kg body weight), alone and in combination, on a complex tracking task, a stability test, a tapping test, and mental performance tests. Marihuana alone was found to impair stability of stance and performance on the tracking and mental performance tests, but did not affect the tapping task. Secobarbital degraded performance on all the tests, except the stability of stance test. The combination of secobarbital and marihuana evidenced performance deficits on the pursuit, stability, and mental performance tests beyond those experienced under either drug alone, suggesting additivity of the component effects.

This study is an example of degraded performance when drugs are used in combination; it suggests a need to explore this area further.

Finally, the history of drug-taking can have differential effects under sedatives. Metabolic rates are known to be accelerated for certain drugs by other drugs (e.g., tolbutamide by alcohol) and other types of physiological changes may be induced by the use of some drugs which could modify the effects of yet other drugs. A study by Raft et al. () provides an example of the differences in responses to sedatives between light and heavy alcohol drinkers. They tested 12 heavy drinkers and 12 light drinkers, either under placebo or 200 mg of pentobarbital. One hour after drug administration, the experimenter subjectively evaluated the subjects’ impairment of memory, concentration, and “physiological signs,” including nystagmus. The heavy alcohol drinkers were found to be “resistant” to the effects of pentobarbital. The “low alcohol users exhibited significantly more signs of intoxication then did high alcohol users.”

The authors did not report the blood pentobarbital levels at the time of evaluation. The results, of course, have to be interpreted cautiously, since they are subjective impressions and not objective data. However, they are suggestive of behavioral tolerance buildup caused by the use of other drugs. Whether such tolerance lowers impairment under sedatives is left to further study. There is clearly a need for more research in this area.

It is clear from the above examples of research on sedatives that there are several areas dealing with sedatives and driving that require further investigation. These include: (1) the epidemiological evidence of relationship between sedative use and driving accident involvement; (2) the extended duration of action of sedatives; (3) vigilance performance under sedatives; (4) the interaction of sedatives and other drugs, including alcohol; and (5) physiological and behavioral tolerance buildup to sedatives as well as cross-tolerance between sedatives and other drugs.

There are many methodological problems inherent in epidemiological data collection for sedatives and driving. In addition, simulator and closed-course studies yield little data which provide information about either accident causation or mechanisms of sedative actions leading to driving accidents. The one area which clearly will provide much information about sedative use and driving is research dealing with the behavioral mechanisms affected by sedatives. It is suggested that further studies be performed which would systematically investigate the behavioral impairment experienced under sedatives.


Selections from the book: “Drugs and Driving”. Robert Willette, Ph.D., editor. State-of-the art review of current research on the effects of different drugs on performance impairment, particularly on driving. National Institute on Drug Abuse Research Monograph 11. March 1977.