Sign the Resolution
Contents | Feedback | Search
DRCNet Home | Join DRCNet
DRCNet Library | Schaffer Library | Drugs and Driving
Department of Forensic Medicine, 364 Little Lonsdale Street, Melbourne VIC 3000 Australia
The Simulated Evaluation of Drug Impairment (SEDI) is a computer-driven mathematical-simulation of the driving task that evaluates the response to three channels of visual information using four buttons and a foot-pedal. It is a divided attention task and sensitive to the impairment caused by sedatives. The instrument automatically records the accuracy, speed and correctness of all responses. Typically 250 changes in information are presented in a two minute test cycle. Previous work with SEDI has involved looking at the degradation of performance caused by drugs or alcohol using subjects familiar with the SEDI task.
This experiment examined the effects of low dose alcohol on the learning curve using subjects unfamiliar with the task. Forty police recruit volunteer subjects were randomly assigned to two groups who received either fruit juice containing alcohol or plain fruit juice. Each subject was given a brief introduction to the test and then 10 repeated attempts at the test cycle.
The control group were indistinguishable from American university students and military personnel. The "alcohol" group had an average blood alcohol concentrations (BAC) of 0.075%. They learned more slowly and achieved lower overall scores than the controls. Learning an unfamiliar task is significantly impaired by alcohol at a low dosage. This provides a simple model for studying adaption to an unfamiliar divided-attention task.
Screen-based tests provide a method for measuring performance impairment of complex tasks. This has a safety advantage over on-the-road testing. The more complex the test challenges, the lower the blood alcohol level at which error rates are observed.
The Simulated Evaluation of Drug Impairment (SEDI) is a computer simulation of a driving task constructed around divided-attention and performance decision making. Divided-attention tasks are studied by having the subject perform more than one task at a time, such as manoeuvring through a complex route and responding to one or more stimuli. Performance decision making is the process by which the subject has to make a choice of what action to take, for example; to brake or swerve to avoid a collision.
The SEDI task requires the individual to respond to three channels of visual/numerical information through four response buttons and a foot pedal. The central visual channel represents a digital speedometer and the peripheral channels represent information critical to driving decisions.
Administration of the SEDI is fully automatic. At the end of each two-minute test, a score is presented which summarises the accuracy, latency and false responses to all of the stimuli and all five of the response channels. The data are summarised to screen, printer and disc at the end of each subject trial.
Most tasks such as balance, tracking, body sway, time-on-target, short term memory, motor pursuit, reaction time and abstract reasoning are sensitive to high doses of alcohol and drugs, but divided-attention, cognitive tasks have been the only type of task to emerge as consistently sensitive to low doses of alcohol and drugs.
This experiment set out to examine the ability of subjects affected by low dosages of alcohol to learn a new task. SEDI clearly shows psychomotor impairment at blood alcohol concentrations >0.12%: Benson & Giguiere (1992).
All the previous work with SEDI to assess drug effects on performance used subjects who had an established personal baseline and then documented any deterioration in performance after consuming drugs. Individuals thus acted as their own controls, and were aware that they had taken a potentially impairing substance when retested.
This experiment aimed to test the proposition that the characteristic asymptotic learning curve for SEDI would be flattened by the effects on human performance. A separate control group was expected to show the characteristic learning curve. Because of the complex nature of the SEDI task, it was expected that measurable deterioration would occur at low drugs levels. To avoid expectation effects, all subjects were told that they would receive some alcohol. The beverage was prepared by a separate person, so that neither the individual doing the test, nor their supervisor was aware of whether the individual belonged to the control or alcohol group. Individuals were not told their blood alcohol concentration until after completing the tests.
Forty police recruits in residence at the Victoria Police Academy or staff of the Traffic Alcohol Section volunteered to join the experiment. No process was involved in subject selection. The experiments took place during free time in the evenings.
A full explanation of the experiment was offered and valid consent obtained. Subjects were told that the experiment was assessing new breath testing devices and a device for assessing performance while drinking. They were warned that they may receive a small dose of alcohol that may take their blood alcohol concentration (BAC) to a maximum of 0.10 grams of alcohol per 100 millilitres of blood.
Subjects were randomly and blindly allocated to the control or alcohol group. They were allowed to learn their final blood alcohol concentration. No subjects were allowed to drive on the evening of the experiment.
The subjects averaged 24.2 years (s.d.4.5). The age distributions of those given alcohol and the control group were not significantly different (t=0.7793, p>0.10).
Five of the alcohol group and nine of the control group were female.
The experiment used SEDI Version 2.3 (1993) Serial Number B000104 operated by a Toshiba Laptop computer. The operator gathered basic demographic data and recorded the results of breath testing.
A preliminary breath test was conducted on all subjects with a Lion Alcometer (Model SD2). All subjects entered the experiment with no measurable breath alcohol. All subjects were given their choice of cola or orange juice containing a measured volume from one of two identical brown flasks. One contained distilled water and the other absolute alcohol. The volume added was 1.25 millilitres per kilogram body weight for women and 1.45 millilitres per kilogram for men. Subjects sat in a lounge watching television or reading for three quarters on an hour. Duplicate breath alcohol analyses were conducted using Draeger Alcotest Model 7110 before and after testing.
All subjects were read prepared instructions. They could have the instructions repeated as often as required and were allowed to experiment with SEDI in the learning mode until they understood what was required. The task was repeated 10 times with a one minute break between attempts. All parameters were saved to disk in ASCII format and printed out later.
The results of testing confirmed the hypothesis that adaption to the challenge of learning SEDI was adversely affected by alcohol.
All of the control subjects had breath alcohol readings of 0.00%, before and after testing. Experimental subjects had breath analysis results averaging 0.071% (sd .021). Tests were commenced half to one hour after drinking. The pretest alcohol concentration was lower (average 0.068% sd 0.019) than the post test reading (average 0.075 sd 0.021), but this difference was not statistically significant (t=0.12564, p >0.05).
The control group adapted to the new task with a learning curve similar to the US norms obtained from pilots and university students. The overall improvement in SEDI score shown by the controls consisted of three component factors - an increase in accuracy, an increase in speed and a reduction in errors (Figure 1). The subjects also improved both speed and accuracy (Figure 2), but made three times as many errors. There was no significant difference in the mean number of correct responses (controls 25.59, subjects 28.89, F=1.41, p=0.23). The subject group achieved a mean response time 150 milliseconds faster than the controls (subjects 768, controls 814, F=1.99 p=0.07). However, the subjects made three times as many errors (mean 18.9) as the control group (mean 6.3, f-17.37, p<.001).
Performance of Controls
Performance of Subjects
The resultant SEDI composite score increased more slowly for the subjects than for the controls. The control group improved their score an average of 1.6 points per attempt. The Pearson correlation coefficient between the number of trials and the final score was significant (rho=0.734, t=4.58, p=0.000). The linear regression equation that explained the control groups SEDI scores was
SEDIscore = 7.653 + 1.620 * Trial number
Subjects improved their scores at a slower rate with less correlation between the number of trials and the final score, although the relationship was still significant. The Pearson correlation coefficient was 0.2150 (t=2.835, p=0.005). The linear regression equation that described the scores was
SEDIscore = 5.877 + 0.548 * Trial number
The composite results are shown in Figure 3.
Composite SEDI Scores
The effects of alcohol are first manifest in the brain centres involved in highly integrated functions, such as skilled performance: Ritchie (1985). The analysis of sensory information, the control of intricate movement patterns and short term memory are specially sensitive to alcohol. The effects on human skills and performance commence at the lowest measurable blood alcohol concentrations: Moskowitz & Robinson (1988).
There is no evidence of a threshold effect for alcohol because some impairment of performance occurs at the lowest B.A.C. that can be measured; nor is there a level at which a sudden transition from unimpaired to impaired can be expected: whatever the blood alcohol concentration examined, at least some skills can be shown to be significantly impaired. The effects of alcohol are dependant both on the quantity consumed and the nature of the task required: Moskowitz (1984).
SEDI offers an unfamiliar task that requires considerable concentration, divided attention, rapid reaction, speed and accuracy. It was proposed as a research tool for comparing the impairment caused by psychoactive substances. Most of the previous work with SEDI has been in a situation where subjects could act as their own controls. They had enough time to learn the task while unaffected by the substance being investigated, before attempting the task while affected by a substance.
In this experiment subjects were adapting to an unfamiliar task while affected by an unknown amount of alcohol. Alcohol affected subjects learned to respond as often as the control group but at the expense of making more mistakes correctly.
Reaction time can be adversely affected by alcohol at low levels. Some studies have displayed deterioration at levels as low as 0.02%, but a level of 0.07% may be needed to produce a significant deficit with common tasks: Starmer (1989). The nature of the stimulus and the reaction required are complicating factors in interpreting the reported results. In a complex task such as SEDI, it was hypothesised that deterioration of reaction time would be shown at low levels of blood alcohol. Paradoxically the subject group acted faster than the control group. This may have been due to subjects pursuing speed rather than accuracy. It may have been due to disinhibition causing subjects to lose interest in accuracy.
SEDI is a research tool that shows promise in the investigation of the subtle effects of low doses of psychotropic substances. Low doses of alcohol caused subjects to adapt more slowly to the new task and they made three times as many errors as the controls. Learning an unfamiliar task is significantly impaired by alcohol at a low dosage. This provides a simple model for studying adaption to an unfamiliar divided-attention task.
Benson SG and Giguiere WE "Psychomotor Impairment by Ethanol in the Postabsorbtive Phase", in Alcohol, Drugs and Traffic Safety T92 Ed by Utzelman HD, Berghaus G and Kroj G, Verlag TÜV Rheinland, Cologne 1993 pp 643-648.
Moskowitz H & Robinson C. Effects of low doses of alcohol on driving-related skills. A review of the evidence. National Highway Traffic Safety Administration, Washington DC, 1988. (PB88-241443) (DOT HS 807280)
Moskowitz H "Attention tasks as skills performance measures of drug effects"(1984) British Journal of Clinical Pharmacology 18(1984):Suppl 1 51s-61s.
Ritchie JM "The Aliphatic Alcohols" in The Pharmacalogical Basis of Therapeutics 7th Edition, Gillman AG, Goodman LS, RallTW & Murad F (eds.) MacMillan, New York. 1985:p372.
Starmer GA. "Effects of low to moderate doses of ethanol on human driving performance" in Crow KE & Batt RD. Human Metabolism of Alcohol (Volume 1): Pharmacokinetics, Medicolegal aspects and General Interest CRC Press, Boca Raton, Florida, 1989. p 101 -133.