I – Background

1Obstructive sleep apnea is a common disorder characterised by episodes during sleep where breathing ceases and arousal occurs. By definition this occurs at least 10 times per hour and may occur as many as 50 to 100 times per hour. This continuing fragmentation of sleep during the night leads to chronic and often irresistible daytime sleepiness.

2It is estimated that approximately 10-15 % of persons in industrial society suffer problems related to sleep. It bas been further been estimated that approximately 5- 10 % have sleep disordered breathing. This problem typically affects males : only 5- 10 % of sleep apnea patients are female. The typical age range is from 30 to 60 years and there is some association with obesity. It is frequently the case that sleep apnea patients are unaware of their problem. It tends to come to light because of snoring, which irritates a spouse, and also can come to light because of an automobile accident.

3Studies suggest that sleep apnea patients are approximately 2 1/2 times as likely to be involved in traffic accidents as compared to other drivers (Gonzalez-Rothi et al, 1985 ; George et al, 1987 ; Findley et al, 1988). In Manitoba, George et al. (1987) compared 27 mates with obstructive sleep apnea to 270 random selected age-matched mates control. Of the OSA cases, 93 % had a history of accidents for which they were legally at fault compared to 54 % of Controls. Similarly, Gonzalez-Rothi et al. (1985) found 32 % of 78 OSA patients had near-miss or auto crashes in comparison to 7 % for control subjects. Although these studies suggest that sleep apnea patients are over-involved in traffic accidents, the findings must be considered over-estimates to some degree. Because one of the mechanisms for identifying sleep apnea patients is that they have accidents, there will be a built-in bias towards finding of higher accident rates among those sleep apnea sufferers who have sought treatment. Nevertheless, the finding of higher accident risk is likely to be correct based on our knowledge of the effects of sleepiness, whether from drugs, long hours, or circadian low points, on driving. Monotonous tasks like highway driving are likely to be significantly affected.

4There is a need to be able to predict driving impairment in sleep apnea sufferers. The current assessment of licensing rests on clinical tests and expert judgement by a physician. No test representative of driving is given and often there are not even tests representative of daytime sleepiness. Our study aimed to explore the relationship between clinical measures of sleep apnea, e.g. the apnea hypopnoea index, subjective measures of daytime sleepiness (the Stanford and Epworth sleepiness scales), the maintenance of wakefulness test, the multiple sleep latency test, and performance on a test which measures basic driving skills. This driving skill test had been previously demonstrated to be sensitive to alcohol and drug effects and to sleepiness (Moskowitz et al, 1985 ; Stein et al., 1992).

5The multiple sleep latency test is the standard test to assess daytime sleepiness and so was used in our study. However, the relationship between performance on this test and the likelihood of a person falling asleep while driving is questionable. It measures the tendency to sleep rather than the ability to stay awake.

II – Experimental Hypothèses

6We had several hypothèses. First of all, we hypothesised that this version of the driving skills test created for this study would be sensitive to alcohol effects. It has been conclusively established that alcohol impairs driving performance (Moskowitz and Robinson, 1988). Therefore, it is important to demonstrate that a test purporting to measure driving skills is sensitive to the effects of alcohol. The second hypothesis was that the performance of sleep apnea patients would differ from that of healthy subjects. Our third hypothesis was sleep apnea treatment would improve driving skills performance. Finally, we anticipated that there would be some relationship between the severity of sleep apnea and driving skills performance.

III – Experimental Methods and Procedures

7The test that we used was a divided attention test that combined tracking with visual search. This test was developed by Dr. Herbert Moskowitz and has been used extensively to look at the sedating effects of drugs (e.g. Moskowitz étal., 1985). It captures both fundamental aspects of driving, namely, tracking and monitoring. Subjects use a steering wheel to keep a cursor centred between 2 bars, representing the lane edges. The tracking task is a sub-critical tracking task based on the Critical Tracking Task (CTT), developed in the 1960s to test pilot and astronaut visual motor performance. The CTT requires the operator to control a display indicator with a hand control such as a steering wheel. The indicator is inherently unstable, and the task is not unlike balancing a broomstick in the hand. The subject balances the broomstick by correcting for instability. The CTT increases test difficulty by « shortening the broomstick » at a predetermined rate – the subject is forced to correct more and more quickly until his visual-motor reactions are unable to keep up and he fails.

8In the more recently devised sub-critical tracking task (SCTT) the operator performs in a similar manner but the degree of difficulty is held at a constant level. The tracking task is combined with a visual search task. Numbers appear randomly at the four corners of the CRT screen on which a tracking task also appears. The target number is 2. When the subject sees a « 2 », he presses one of 4 buttons mounted on the steering wheel, corresponding to the corner in which the « 2 » appeared.

9Subjects were given three 5-minute training runs. They were told that the steering task was the primary task as is the case in driving.

1, Test 1 : 0 % vs. 0.08 % Alcohol

10Twenty-five healthy male subjects with ages ranging between 30 and 57 years were tested. Half of the subjects received a sober condition first, the other half received 0.08 % blood alcohol concentration condition first. Our aim with this first test was to validate the sensitivity of this version of the divided attention test to alcohol effects and also to establish a range of acceptable and unacceptable performance on the test. Alcohol at 0.08 % is the legal limit in Canada. Therefore, performance worse than this due to sleepiness could be considered unacceptable.

11The resuit of our first study showed a clear difference in mean response time due to alcohol, significant at the 0.0005 level (see Table 1). Tracking error was not affected.

2. Test 2 : Sleep Apnea vs. Healthy Subjects

12Having established the sensitivity of our test to a well-known impairing substance, alcohol, we went on to use the test to examine the performance of obstructive sleep apnea patients. The performance of these patients was measured at several points during the day and then compared to values determined for the drunk and sober subjects. This experiment is currently still in progress but data for 19 sleep apnea patients is available. Subjects for this study are males currently not under treatment with a minimum of 10 disturbance events per hour as a result of their sleep apnea.

13Ail subjects were tested at 9 : 00 a.m., 11 : 00 a.m. and 1 : 00 p.m. The first 9 subjects, in addition to the first three test times, were also tested at 3 : 00 p.m. At each test point, they drove for 15 minutes. They then carried out the multiple sleep latency test for 20 minutes. The next 10 subjects, instead of carrying out a driving test at 3 : 00 p.m., drove at 2 : 00 p.m., had a 45-minute nap, then at 3 : 00 p.m. drove immediately after waking up from their nap, and at 4 : 00 p.m. drove again 1 hour after the end of their sleep. Data from both sets of subjects are being used to look at performance throughout the day, at 9 : 00 a.m., 11 : 00 a.m., and 1 : 00 p.m. Data from the second set of subjects are being used to look at the impact of the 45-minute nap on performance.

14All subjects received the Epworth sleepiness scale the night before they came into the laboratory. This sleepiness scale asks questions about general daytime sleepiness – how likely were they to doze off or fall asleep in situations such as sitting and reading, watching 1 V, sitting inactive in a public place, and so on. The Stanford sleepiness scale was given just prior to sleep and just prior to each of the driving tests. This scale measures how sleepy people are at that particular moment – whether they’re feeling highly alert or just about to go to sleep, or somewhere in the middle.

15All subjects spent overnight in the laboratory where an EEG recording was taken throughout the night to establish severity of sleep apnea.

16Figures 1 and 2 show the performance of the patients as compared to the sober and intoxicated drivers. For the tracking task, it can be seen that the sleep apnea patients were worse than the sober drivers and the intoxicated drivers, at significant levels of 0.01 and 0.02 respectively. Looking at mean reaction time, again the sleep apnea patients were worse than the sober drivers, here at a level of 0.02 significance. There were no significant differences between intoxicated drivers and sleep apnea patients. In other words, the performance of sleep apnea patients was worse than sober drivers on both measures of performance. With respect to the intoxicated drivers, the performance of sleep apnea patients was worse on one measure, the tracking, and was equivalent on the other measure, mean response time.

17We do not permit drivers at 0.08 % blood alcohol level to drive. Since the performance of sleep apnea patients is worse than these, this raises questions about their ability to drive safely. Although this task is not an on-the-road driving task, there have been numerous studies of various substances to establish that the finding of effects on performance on a task such as this on tracking and on attention to searching for targets win translate into effects on performance in a driving simulator and further translate again into effects on performance on the road. For example, see Smiley (1987) for a review of laboratory, driving simulator and on-road tests of diazepam effects.

3. Test 3 : Before Treatment vs. After Treatment

18The next part of our study looked at changes in driving skills performance before and after treatment. The most effective treatment for sleep apnea is positive airway pressure by a tight-fitting nasal mask during night-time hours when the person is trying to sleep. This is not a pleasant treatment and some patients cannot tolerate it. Approximately 2/3 of subjects comply with this treatment. Prior to carrying out the driving skills task, subjects slept overnight in the laboratory where their treatment could be monitored.

19To date four subjects have been treated between six weeks and three months after their initial assessment. As can be seen from Table 2, in every case, mean response time improved and mean tracking error improved. There are too few subjects to carry out a statistical analysis. However, the results to date show consistent improvement.

III – Corrélation between Clinical and Performance Measures

20Finally, we examined the correlation between clinical measures, performance measures, and daytime sleepiness measures. A chart of correlations is given in Table 3. As can be seen, the two performance measures, mean reaction time and tracking error, show some degree of correlation with each other. However, there is no correlation between the apnea hypopnoea index and performance measures, nor is there any correlation between the multiple sleep latency test and performance measures. The multiple sleep latency test has been the standard test for examining daytime sleepiness. The one correlation that does exist is between the Epworth sleepiness scale, a measure of subjective daytime sleepiness and mean reaction time. It is interesting to find no correlation between the clinical measures and the impaired performance on a performance test. This strongly suggests that clinical measures alone are unlikely to be good predictors of driving capability. It implies also that some sleep apnea patients can continue to be licensed safely, whereas others should not be driving. Further validation of the driving skills test using a sophisticated driving simulator on an on-road study is required and is anticipated prior to using it for assessment purposes.

IV – Conclusions

21Our study has demonstrated that patients with moderate to severe sleep apnea have poorer performance on a driving skills task than do healthy subjects or subjects who have consumed 0.08 alcohol. Four subjects were tested after receiving treatment for sleep apnea ; all showed improved performance both on visual search and tracking. There was little correlation found between performance and severity of sleep apnea, suggesting that currently used clinical measures are not likely to be good indicators of which sleep apnea patients should continue to be licensed.

Table 3. Correlation analysis