Evaluation of In-Car Countermeasures to Sleepiness: Cold Air and Radio L.A. Reyner, J.A. Horne Sleep Research Laboratory, Loughborough University, UK Summary: The efficacy of putative in-car countermeasures to driver sleepiness is unknown. Sixteen young adult drivers within the normal range for the Epworth Sleepiness Scale (ESS), had their sleep restricted to 5 hours the night before, and drove an interactive car simulator in the afternoon for 2.5 hours, under monotonous conditions. After 30 minutes of driving they were exposed to: (1) cold air to the face (AIR) from the vehicle s air conditioning vents, (2) listening to the vehicle s radio/tape (RADIO) according to subjects choice, or (3) NIL treatment. The active treatments typified those experienced under real driving conditions. Drifting over lane markings were incidents. EEGs were recorded and spectrally analyzed in the alpha and theta range. Subjects responded to the Karolinska Sleepiness Scale (KSS) every 200 seconds. Overall, RADIO and AIR had no significant effects on incidents, although there was a trend for RADIO to reduce incidents, particularly during the first 30 minutes, when AIR also had some effect. KSS scores were significantly lower for RADIO for most of the drive, whereas AIR had only transient and non-significant effects. The EEG showed no significant effects of the active treatments. Compared with other countermeasures such as caffeine and a brief nap, which we have previously shown to be more effective (using the same equipment and protocols), AIR and RADIO are at best only temporary expedients to reduce driver sleepiness, perhaps enabling drivers to find a suitable place to stop, take a break and avail themselves of caffeine and a nap. Key words: Driver sleepiness; countermeasures; EEG; driving performance; subjective sleepiness. Accepted for publication November 1997. Address correspondence and reprint requests to L.A. Reyner, Human Sciences Department, Loughborough University, Leicestershire LE11 3TU, United Kingdom A RECENT and substantive survey by Maycock 1 on sleepiness and driving in 4621 car drivers reported that opening the vehicle s window or listening to the radio/tape player were among the most commonly used helpful methods employed by the drivers for countering the effects of sleepiness. Unfortunately, there is little scientific evidence to support the efficacy of these putative in-car countermeasures to sleepiness. 2 The only study on the effects of a radio on driving performance and fatigue 3 was in the context of personality types and reaction time. No driving data were given and the other findings were inconclusive. Authoritative reviews of sleep loss 4-7 give no guidance to the efficacy of cold air, music, conversation or acceptable levels of noise for keeping people awake, and neither do manuals advising military commanders on how to sustain wakefulness in sleep deprived personnel. 8,9 The latter rely on scheduled naps and psychostimulants. The two non-driving studies on the effects of noise on performance in sleepy subjects 10,11 administered this in the context of a stressor, where the noise was unpleasantly loud (white noise at 90-100 db). Both investigations found that this prevented a performance decline during a 30-minute vigilance task, but such a method is clearly unsuitable for drivers. Prudent drivers stop and take a break when sleepy. As to what they do during this break, Maycock s 1 findings showed that drivers believed that taking a walk is the best countermeasure to sleepiness. However, laboratory findings of sleepy subjects 12 have shown this to be ineffective, as the alerting effect of even heavy (10 minutes at 70% VO2 maximum) exercise is only transient. In contrast, drinking coffee (150 mg caffeine) and a (<15 minute) 46
Mean and se of incidents for AIR, NIL and RADIO conditions Figure 1. Mean number of incidents per 30 minutes for all (N=16) subjects and the three treatment conditions. There was a significant effect of time, with the first post-treatment hour being less than the second hour (see text for further details). There was no significant difference between treatments, although there was a trend for RADIO to reduce incidents during the post-treatment period, particularly for the first half hour, when AIR also had some (non-significant) effect. nap 13 or better still the two combined, 14 are very effective countermeasures that can be utilized feasibly during a 30-minute break from driving. Unfortunately, few of Maycock s 1 sleepy drivers resorted to these latter methods. The present investigation assessed the usefulness of two commonly utilized in-vehicle countermeasures to sleepiness: blowing cold air on the face and listening to the car radio/tape player (vs nil treatment). Subjects drove an instrumented car simulator under dull and monotonous road conditions for a total of 2.5 hours, at a time of day propitious for daytime sleepiness and when sleep-related vehicle accidents are prevalent. 15,16 METHOD Subjects As most sleep-related vehicle accidents occur in young adults, 15 we targeted this age group. Sixteen subjects (8 male, 8 female, average 23 years; standard deviation 2.0 years), all of whom were healthy (medicationfree), experienced drivers (driving for >2 years, averaging >3 hours per week), good sleepers, sleeping regular hours and infrequent daytime nappers (< once a month), were 47 recruited by advertisement. They completed the Epworth Sleepiness Scale (ESS), 17 and all were within the normal range of 0-10. None complained of daytime sleepiness, nor indicated potential sleep disorders in a screening questionnaire. They had the procedures fully explained to them, signed consent forms, and were paid to participate. Design and Procedure On an initial preparatory day subjects took a 2-hour practice drive on an interactive car simulator (see below). They later underwent all treatment conditions (see below) a week apart, in a randomized design, with the subjects afternoon sleepiness enhanced by sleep being restricted to 5 hours (delayed bedtime) the night before. They slept at home, with sleep monitored by wrist actimeters. 18 A simulated drive (see below) commenced at 1415 and spanned 2.5 continuous hours. The first 30 minutes was for adaptation purposes, after which one of the following treatments was applied for the remaining 2 hours: AIR cold air (10 C) from the car s air conditioning system, blown onto the face (wind speed 1.4 meters/second)
from vents in the dashboard. RADIO use of the car radio/tape player. Subjects had been instructed that they could bring with them any of their own tapes which they might typically play on a long drive, or they could choose any radio station that they would normally select on such journeys. Volume was at their discretion. NIL a control condition of neither AIR nor RADIO. Car simulator. The simulator was an immobile car with an interactive computer-generated projection of a roadway. 13 Subjects drove within lateral markings either side of a specific road lane. Typically, a car being driven by a driver falling asleep weaves and/or drifts across the road. We designated lane-drifting as an incident, which was determined from continuously logged steering data and identified when a car-wheel crossed a lane marking. Approximately hourly, at a fixed time, a slower vehicle had to be overtaken. This very infrequent event was introduced to facilitate a collision. EEG. Using the A1-C3 derivation, data were logged at 128 Hz and spectrally analyzed ( Rhythm software, Stellate Systems-Quebec) in 4-second epochs and averaged per minute. The raw data were high and low band-pass filtered to remove slow eye movements and muscle artifact. Sleepiness during driving is problematic, as the psychophysiological processes accompanying falling asleep at the wheel do not typify sleep onset when subjects lie down, expect to sleep, and fall sharply into obvious sleep. Driving motivates sleepy subjects to put considerable effort into remaining awake and keeping their eyes open. In this context, we have found 13 that the summation of alpha and theta (ie, slower EEG activity in the range 4-11 Hz) shows more consistent EEG findings in relation to sleepiness than either of these alone. Unavoidable blink artifact on the EEG does not bias the EEG outcomes. 13 As EEG power shows large individual differences, and to allow these data to be averaged across subjects, it was standardized for each subject (using the 1-minute epochs) by the following transformation 13 prior to averaging across subjects: Difference from mean of first 30 minutes transformed epoch = Standard deviation of first 30 minutes Subjective responses. Subjects verbally reported their sleepiness every 200 seconds on the 9-point Karolinska Sleepiness Scale 19 (eg, 1 = extremely alert, 3 = alert, 5 = neither alert nor sleepy, 6 = some signs of sleepiness, 7 = sleepy, no effort to stay awake, 8 = sleepy, some effort to stay awake, 9 = very sleepy, great effort to keep awake, fighting sleep). The printed scale was attached to Mean and SE of KSS scores for each condition (N = 16) Figure 2.. Mean (with s.e. bars) subjective sleepiness trends (KSS scores - 200-second intervals) under all conditions. RADIO significantly reduced subjective sleepiness for the first, third and fourth 30 min post-treatment periods. 48
Mean and se of standardized EEG scores for each condition (N = 16) Standardized EEG values Figure 3. Mean (with s.e. bars) of standardized EEG power (4-11 Hz) in 1 min intervals, for all conditions. The zero line approximates a level of theta + alpha, above which indicates increasing EEG sleepiness. There were no significant effects of AIR or RADIO. Note the alerting effects on the three occasions when the subjects overtook another vehicle (see text). the dashboard, in clear view of the driver. Subjects gave the appropriate number when asked to do so. Data analyses. For the preliminary pre-treatment half hour, incidents, subjective responses, and EEG data were compared across conditions to ensure that there were no significant differences. This was the case for all indices (see Results). Post-treatment data were averaged into 4 5 half-hour blocks per subject. Two-way (time 5 treatment) repeated-measures ANOVA (using the Hunyh- Feldt epsilon, e ) were performed, followed by Tukey post-hoc tests when appropriate. Any significant time effect could be confounded by the treatment, inasmuch as the treatment could become less effective with time. Accordingly, any interaction effects would be important in these respects. Post hoc tests were only undertaken on significant time effects when there was no accompanying significant interaction. When both time and the interaction were significant, post hoc tests were performed only on the interaction. RESULTS Neither AIR nor RADIO had any statistically significant effects in reducing incidents, although it can be seen 49 from Fig. 1 that there were clear trends for RADIO to reduce these compared with NIL, particularly during the first post-treatment half hour, when AIR also caused some reduction. However, incidents did show a significant effect with time (F=3.84 [3,45] p<0.02, e=0.72), and posthoc Tukey tests were significant (p<0.05) for the first two half-hour periods vs both of the second two periods (in effect, between the first and second post-treatment hours). There was no significant interaction. Figure 2 shows the KSS trends for all conditions. The ANOVA was significant for condition (F=7.19 [2,30] p<0.003, e=0.77), time (F=46.3 [3,45] p<0.005; e=0.74), and with the interaction (F=2.60 [6,90] p<0.03, e=0.76) still being significant when applying epsilon. Post-hoc tests on conditions showed a significant (p<0.05) effect only between NIL and RADIO. Post-hoc tests on conditions within time showed that RADIO was significantly lower than NIL for the first, third and last half-hour epochs. There were no such significant findings between AIR and NIL. It is clear from Fig. 2 that the effect of RADIO was most apparent in the first half-hour epoch. EEG revealed a significant time effect (F=6.94 [3,45] p<0.001; e=0.7), but no other significant effects. From Fig. 3 it can be seen that EEG power progressively
increases with time, and post-hoc tests show both the first two initial half-hours to be significantly (p<0.05) lower than both the last two. The zero line in Fig. 3 approximates a level of theta + alpha, above which indicates increasing EEG sleepiness. The alerting effect of overtaking another vehicle is apparent in both groups, shown by transient reductions in EEG power in the alpha + theta range. DISCUSSION The efficacy of AIR and RADIO in counteracting driver sleepiness fall below that of caffeine (150 mg) and/or a (<15 minute) nap, found by us using a similar experimental protocol. 13,14 However, although not significant in reducing incidents, RADIO seemed to have a marginally better and more sustained effect than did AIR. This is also reflected in the post-treatment KSS data, particularly in those half-hour blocks when KSS scores were significantly reduced by RADIO (see Results). It can be seen from Fig. 2 that any initially beneficial effects of AIR on subjective sleepiness only lasted around 15 minutes. Inasmuch as there were clear individual differences in the effects of AIR and RADIO on incidents, these techniques may offer transient benefits to some drivers. Nevertheless, neither treatment is recommended for the sustained counteraction of sleepiness in sleepy drivers. In practical terms, these treatments may only be sufficient to allow drivers to find a suitable stopping place to take a break and have their caffeine and/or nap. The EEG was obtained during driving, and although the task was monotonous, the situation was not constant (compared with, for example, staring at a blank wall or undertaking a simple auditory vigilance task in a dimly lit room). Hence, even when all EEG data are standardized and averaged together, there is still evident fluctuation within conditions, as can be seen in Fig. 3. Nevertheless, under these driving circumstances, this EEG measure of sleepiness has been shown by our other studies 13,14 to be sensitive to other countermeasures to sleepiness. There are independent findings 20 showing that alpha and especially theta EEG activities (as embodied in our EEG index) are valid measures of physiological sleepiness. Furthermore, the transient alerting effect on the EEG when the subjects overtook another vehicle (Fig. 3) indicates that this EEG index is potentially sensitive to changes in arousal. The half-hour EEG data blocks did incorporate the overtaking periods (which were at constant times), but when these averaged within the 30- minute blocks, they made no difference to the statistical outcome. It might be argued that had we utilized colder air or louder/more stimulating auditory output from to the radio/tape player, then beneficial effects might have been greater. As the cold air typified that produced by car air conditioning systems, we felt it unreasonable to make it 50 colder or increase air flow, as this would not be appropriate to the real world. Similarly, the selections of radio stations and tapes were at the choice of the driver and played at the volume they felt comfortable listening to, and were therefore felt to represent real world conditions. In summary, we find that common countermeasures used by drivers to combat sleepiness while driving, such as cold air and listening to the vehicle s radio/tape player, are of marginal and transient benefit. While sleepy drivers may believe that these countermeasures are effective in improving their alertness, this is not reflected to the extent of their belief in their otherwise deteriorating driving performance. ACKNOWLEDGEMENT This work was supported by the UK Department of Transport. We would like to thank Hilarie Kemp for her assistance with this project. REFERENCES 1. Maycock G. Sleepiness and driving: the experience of UK car drivers. J. Sleep Res 1996,5:229-237. 2. Mackie RR, Wylie CD. Countermeasures to loss of alertness in truck drivers - theoretical and practical considerations. In M Vallet (ed) Le Maintien de la Vigilance dans les Transports. Caen: Paridigme, 1991;113-141 3. Fagerstrom K-O, Lisper H-O. Effects of listening to car radio, experience and personality of the driver on subsidiary reaction time and heart rate in long term driving. In R R Mackie (ed), Vigilance, New York: Plenum Press, 1977;73-85. 4. Kleitman N. Sleep and Wakefulness. 1963, Chicago: Chicago University Press. 5. Wilkinson RT. Sleep deprivation. In OG Edholm and AL Bacharach (Eds) Physiology of Survival London: Academic Press, 1965;399-430. 6. Monk TH (ed). Sleep, Sleepiness and Performance. 1993, Chichester: Wiley. 7. Horne JA. Why We Sleep, 1988 Oxford: Oxford University Press 8. Belenky GL, Krueger GP, Balkin TJ, Headley DB, Solick RE. Effects of continuous operations (CONOPS) on soldier unit performance: review of the literature and strategies for sustaining the soldier in CONOPS. Walter Reid Army Institute of Research 1987, Report NoWRAIR-BB-87-1 9. Naitoh P, Kelly TL. Sleep management user s guide for special operations personnel. Naval Health Research Centre, 1992, Report No 92-28 10. Corcoran DWJ Noise and sleep loss. Quart J Exper Psychol 1962: 14,178-182. 11. Wilkinson RT. Interaction of noise with knowledge of results and sleep deprivation. J Exper Psychol 1963;66:332-337 12. Horne JA, Foster SC. Can exercise overcome sleepiness? Sleep Res 1995,24A437. 13. Horne JA, Reyner LA. Counteracting driver sleepiness: effects of napping, caffeine and placebo. Psychophysiology 1996;33:306-309 14. Reyner LA, Horne JA Suppression of sleepiness in drivers: combination of caffeine with a short nap. Psychophysiology, 1997 (in press) 15. Horne JA, Reyner LA. Sleep related vehicle accidents. Br Med J 1995;310:565-567. 16. Åkerstedt T, Czeisler CA, Dinges D, Horne JA. Accidents and sleepiness: a consensus statement. J Sleep Res 1994;3:195. 17. Johns MW. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep 1991;14:540-546.
18. Horne JA, Pankhurst FL, Reyner LA, Hume K, Diamond I. A field study of sleep disturbance: effects of aircraft noise and other factors on 5,742 nights of actimetrically monitored sleep in a large subject sample. Sleep, 1994;17:146-159. 19. Åkerstedt T, Gillberg M. Subjective and objective sleepiness in the active individual. Int J Neurosci 1990;52:29-37. 20. Rechtschaffen A, Kales A. A Manual of Standardized Terminology, Techniques, and Scoring System for Sleep Stages of Human Subjects, 1968, Los Angeles, UCLA Brain Information Service. 51