SLEEP, SLEEP RESTRICTION, AND PERFORMANCE
The Maintenance of Wakefulness Test and Driving Simulator Performance
Siobhan Banks, PhD1,2; Peter Catcheside, PhD1; Leon C. Lack, PhD1,3; Ron R. Grunstein, MD, PhD4; R. Doug McEvoy, MD1,2
Adelaide Institute for Sleep Health, Repatriation General Hospital – Daw Park, Adelaide; 2School of Medicine and 3School of Psychology, Flinders
University of South Australia, Adelaide; 4Woolcock Institute of Medical Research, Sydney, Australia
Study Objectives: It has been suggested that the Maintenance of Wake- inversely correlated with braking reaction time. During the partial sleep
fulness Test (MWT) may be clinically useful to assess ﬁtness to drive, yet deprivation and alcohol condition, the number of microsleeps during the
little is known about the actual relationship between sleep latency and driving task, steering deviation, braking reaction time, and crashes all
driving performance. This study examined the ability of 2 MWT trials to negatively correlated with the MWT1 sleep latency. Additionally, construc-
predict driving-simulator performance in healthy individuals. tion of a receiver-operator characteristic curve revealed that MWT1 sleep
Design: Experimental latency in the partial sleep deprivation plus alcohol condition signiﬁcantly
Setting: NA. discriminated subjects who had a crash from those who did not.
Patients or Participants: Twenty healthy volunteers (mean age 22.8 Conclusions: These results indicate that sleep latency on the MWT is
years; 9 men). a reasonable predictor of driving simulator performance in sleepy, alco-
Interventions: NA. hol-impaired, normal subjects. Further research is needed to examine the
Measurements and Results: The MWT and driving-simulator perfor- relationship between daytime MWT results and driving simulator perfor-
mance were examined under 2 conditions—partial sleep deprivation and mance in sleepy patients (eg, those with obstructive sleep apnea) and in
a combination of partial sleep deprivation and alcohol consumption. Each experimentally sleep-deprived normal subjects.
subject was studied a week apart, with the order randomly assigned. Sub- Keywords: MWT, Driving, Sleepiness, Alcohol
jects completed a nighttime 70-minute AusEd driving simulation task and Citation: Banks S; Catcheside P; Lack LC et al. The maintenance of
two 40-minute MWT trials, 1 before (MWT1) and 1 after (MWT2) the driv- wakefulness test and driving simulator performance SLEEP 2005;28(11):
ing task. In the sleep-deprived condition, the MWT1 sleep latency was 1381-1385.
INTRODUCTION tion with other tests, to assess fitness to drive in patients with
sleep disorders (eg, obstructive sleep apnea) who are considered
WHILE ESTIMATES VARY ACCORDING TO COUNTRY at risk of falling asleep while driving. The MWT has also been
AND METHOD OF POLICE REPORTING,1-3 IT IS THOUGHT recommended to assess the alertness of United States Air Force
THAT APPROXIMATELY 20% OF MOTOR VEHICLE crashes pilots suffering from hypersomnia.7
are caused by sleepiness or fatigue. People with obstructive sleep Despite the recent recommendations by regulatory authori-
apnea, a disorder characterized by excessive daytime sleepiness, ties and the theoretical attraction of the MWT as a test of driv-
are 2- to 7-fold more likely to have a crash than are healthy driv- ing safety, little is known about the actual relationship between
ers.4, 5 One of the most important challenges facing the sleep med- MWT sleep latency and driving performance. A small prelimi-
icine community is how to accurately measure individuals’ levels nary study examining only 5 patients with obstructive sleep apnea
of sleepiness and accident risk. Currently there is no agreed-upon found no relationship between the MWT and driving simulator
test or tests of driving safety for sleepy patients. One test that has performance.8 A recent study of drivers who had crashed their
been used both in research and clinical settings and that has sub- cars demonstrated that the drivers who crashed had significantly
stantial face validity as a measure of sleepiness is the maintenance more sleepiness, slower reaction times, and a trend for shorter
of wakefulness test (MWT). The MWT requires subjects to stay MWT sleep latencies compared with control subjects.9 A study
awake while in a soporific environment for several periods over a that examined the multiple sleep latency test (MSLT), another
day. It can be readily performed in most clinical sleep laboratories objective test of sleepiness, and driving simulator performance
and, in fact, is currently recommended in Australia,6 in conjunc- found that, while there was a reasonable relationship between the
2 tests, MSLT sleep latency explained little of the error in the
Disclosure Statement Today’s society is very demanding on time, both professional-
This was not an industry supported study. Dr. Grunstein has received re- ly and socially, resulting in a high prevalence of sleep restriction.
search support from Cephalon, ResMed, Sanoﬁ-Synthelabo, GSK, and Neu- The associated impact on the ability to perform complex tasks is
rocrine Pharmaceuticals. Dr. McEvoy has received research support and potentially devastating. More investigation is needed to ascertain
equipment from ResMed and Massimo. Drs. Banks, Catcheside, and Lack the validity of electroencephalogram (EEG)-based sleep labora-
have indicated no ﬁnancial conﬂicts of interest. tory tests of sleepiness to predict driving performance. The aim
of this study was to examine the ability of the MWT sleep latency
Submitted for publication April 2005 to predict driving simulator performance in healthy young indi-
Accepted for publication July 2005
viduals who were sleep restricted with and without the presence
Address correspondence to: Siobhan Banks, PhD, Unit for Experimental
of low-dose alcohol. Results from the same experiment showing
Psychiatry, Division of Sleep and Chronobiology, Department of Psychia-
the effects of sleep restriction and low-dose alcohol on driving
try, University of Pennsylvania School of Medicine, 1013 Blockley Hall, 423
Guardian Drive, Philadelphia, PA, USA 19104-6021; Tel: (215) 898-9665;
performance have been reported previously.11
Fax: (215) 573-6410
SLEEP, Vol. 28, No. 11, 2005 1381 MWT and Driving Simulator Performance—Banks et al
METHODS puter program devised to monitor driving impairment from a
number of variables, including position on the road, speed de-
The relationship between MWT sleep latency and driving viation over time, reaction time to a braking task (appearance of
simulator performance was examined under 2 conditions—partial trucks), and crashes (driving off the road, stoppage events and
sleep deprivation (PSD) and a combination of PSD and alcohol crashing into the back of a truck). Early work suggests that this
consumption (PSD-A). Each subject was studied a week apart, test is sensitive to varying degrees of sleep deprivation and sleep-
with the order randomly assigned. The MWT and driving simula- iness.11,13-15 Subjects were required to “drive” the AusEd simulator
tor performance were examined under 2 conditions. The research using a steering wheel and pedals. The view, seen from a front-
and ethics committee at the Repatriation General Hospital - Daw seat perspective, was of a dual-carriage rural road at night, with
Park, Adelaide approved the study, and all subjects gave written the usual lane divisions and the road edges marked by reflective
informed consent. posts. A speedometer was displayed in the top left corner of the
computer screen. Subjects were asked to maintain their position
Participants in the left-hand lane on the road (in accordance with Australian
Advertisements were posted at the Flinders University of South driving code), to maintain speed within 60 to 80 kilometers per
Australia campus. Young adults were recruited for this study due hour, and to react by braking firmly and as quickly as possible to
to the high number of 18- to 30-year olds who have sleep-related any trucks that appeared ahead in the driving lane. The simulator
car accidents in the early hours of the morning. Subjects were was programmed to present 4 trucks at approximately 10-minute
excluded if they had a history of sleep disorders (eg, self-reported intervals during the 70-minute task.
snoring or difficulty sleeping), were taking any medication or had
a history of motion sickness. All subjects received an honorarium Experimental Conditions
of $100. The subjects were required to keep a detailed diary of their
sleep habits and to wear an activity monitor (Gaehwiler Electronic,
Familiarization Session Hombrechtikon, Switzerland), which measured their sleep-wake
During the first visit to the laboratory, which occurred in the activity for 1 week prior to the experimental conditions. This was
daytime, subjects were introduced to the testing equipment and done to verify that the subjects had regular sleep habits in the
driving simulator. Subjects underwent three 10-minute practice week prior to testing, followed the sleep-deprivation protocols,
sessions on the driving simulator and were randomly assigned and did not nap during the day of testing. Subjects participated
to condition order. After three 10-minute sessions, the driving in the 2 experimental conditions in a repeated-measures design,
simulator performance learning effect had become asymptotic attending the laboratory twice with a week separating each visit.
(defined as less than a 10% change in performance between the In the PSD condition subjects were restricted to 5 hours time in
practice trials). The subjects then completed the Epworth Sleepi- bed on the night prior to testing (1 AM -6 AM). They were required
ness Scale.12 The subjects took home an activity monitor and a to telephone a time- and date-stamped answering machine before
sleep diary to be completed in the week prior to testing. Subjects going to bed and after rising in the morning to ensure compli-
were not required to obtain a specific amount of sleep during the ance.
testing period (except on the PSD night). They were instructed to In the PSD-A condition, subjects were required to restrict sleep
keep to their normal sleep-wake pattern. according to the protocol above and to consume alcohol prior to
the driving task to produce a blood alcohol concentration (BAC)
Maintenance of Wakefulness Test of approximately 0.04 g/dL, which is just below the legal BAC
limit for driving in Australia, 0.05g/dL. To achieve this, subjects
All MWT trials were performed in a similar setting using a consumed 1 mL of 50% alcohol per kg of body weight (in a car-
simplified recording montage (C3/A2, O1/A2, electromyogram, bonated, noncaffeinated beverage) at 10:30 PM. At 12:15 AM, they
and electrooculogram). The testing room was sound attenuated, consumed another drink with 0.5 mL of 50% alcohol per kg of
insulated from external light, and equipped with dimmer lights body weight. Blood alcohol levels were estimated immediately
overhead. Ambient temperature in the room was approximately before and after the MWT trials and driving test using a calibrated
22oC. Bedroom doors were closed, and all monitoring was per- Breathalyzer (Dräger, Alcotest 7410Plus Lübeck, Germany) accu-
formed external to the bedroom to keep noise to a minimum. rate to 0.005 g/dL. Subjects were not blinded to alcohol presen-
During each MWT trial, subjects sat semiupright (10o to 30o back tation, as our aim in this study was to test the subjects in a real-
from vertical) in a comfortable lounge chair that had a high back world common situation in which the subjects would be aware of
to support the head and neck. Prior to each trial, subjects were alcohol consumption.
instructed to “keep your eyes open and try not to fall asleep.”
Subjects were asked not to use any extraordinary mental or physi- Experimental Procedure
cal measures (eg face slapping to avoid sleep). The recordings
were then started, and the lights dimmed to an illumination of 1 Subjects arrived at the laboratory at 9:00 PM. Their BAC was
lux. Each trial was terminated at the first occurrence of sustained ascertained with the Breathalyzer, a urine sample was taken to
sleep (3 consecutive 30-second epochs of stage 1 sleep or 1 epoch test for habitual drugs of abuse (eg, opioids, cannabinoids, and
of any other stage) or after 40 minutes if there was no sleep. amphetamine—all tested negative), and activity monitors were
downloaded to ensure that the subjects had complied with the
AusEd Driving Simulator study protocol requirements. The subjects’ timepieces were re-
moved so that they had no external time cues. Standard surface
The AusEd driving simulation task used in this study is a com- electrodes were applied for monitoring: EEG (C3/A2, C4/A1),
SLEEP, Vol. 28, No. 11, 2005 1382 MWT and Driving Simulator Performance—Banks et al
submental electromyogram, left and right eye movements, and did not. A ROC curve is a plot of the true positive rate against
electrocardiogram. All parameters were recorded using the Sleep- the false positive rate for all possible cut points of a diagnostic
watch (Compumedics, Melbourne, Australia) data-acquisition test. An area of 0.84 under the curve, for example, means that a
system. Subjects then completed the Stanford Sleepiness Scale.16 randomly selected individual from the positive group has a test
Subjects were allowed a short practice run on the driving simu- value larger than that of a randomly chosen individual from the
lator and given a standardized snack (150 calories; dry biscuits negative group 84% of the time. The closer the curve follows the
and cheese) and alcohol or an equivalent volume of the carbon- left-hand and the top border of ROC curve space, the more ac-
ated noncaffeinated beverage at 10:30 PM and at 12:15 AM. The curate the test. The closer the curve is to the 45-degree diagonal
protocol of this study had to allow for the consumption of alco- of ROC curve space, the less accurate the test. The area under a
hol (10 minutes to consume and 30 minutes to stabilize for BAC ROC curve therefore provides a measure of the performance of
readings) and other tests (eg, the Psychomotor Vigilance Task, the the test, in this case the utility of MWT sleep latency to predict
results of which have been reported previously).17 driving simulator crashes. Mann-Whitney U tests were used to
The first MWT (MWT1) commenced at 11:30 PM. Subjects determine if the area under the ROC curves differed from 0.5 (ie,
started the 70-minute driving simulation at 1:00 AM and were a test that discriminates no better than pure chance).
prompted every 4.5 minutes during the driving task to answer
simple questions about their perception of level of driving perfor- RESULTS
mance and crash risk (results reported elsewhere).11 Subjects were
Twenty healthy subjects (11 women, mean age 21.9±2.2; 9
told that the probes would sound at random intervals. The driving
men, mean age 23.8±4.8) participated. All subjects had normal
task took place in a private, semidark (10 lux), and sound-attenu-
body mass index (23.3±3.1). All participants were university stu-
ated room. The second MWT (MWT2) commenced at 02:15 AM.
dents, who consumed 2 or fewer servings of caffeine (2 small
The experiment concluded at approximately 3 AM, and subjects
cups of regular coffee or tea, 2 chocolate bars [50 grams each],
were driven home by taxi cab.
2 cans of cola, etc) per week and 6 or fewer standard alcoholic
DATA ANALYSIS beverages a week (eg weekend social drinkers). In addition, all
subjects had current driver’s licenses, and all were experienced
Maintenance of Wakefulness Test at computer game controls and formats. The mean Epworth
Sleepiness Scale score at familiarization for the whole group was
Sleep latency was defined as the first appearance of 3 epochs of 6.4±3.9. All subjects had a regular sleep-wake cycle in the week
stage 1 sleep or 1 epoch of any other sleep stage. Subjects with no prior to testing, with activity-monitor data showing subjects were
sleep onset were assigned a value of 40 minutes. Both the individ- inactive for an average of 411±37 minutes per night. They sub-
ual trial sleep latencies and the mean sleep latency were recorded jectively reported (sleep diary kept for 7 days) that they obtained
for each condition. an average of 423.6±45.8 minutes of sleep per night. No subjects
were excluded on the basis of the amount of sleep obtained in the
AusEd Driving Simulator week before testing. All subjects had zero BAC and a negative
This study examined mean steering deviation (subject’s me- urine drug test on arrival at the laboratory on experimental nights.
dian position on the road, excluding crashes), reaction time (in Data from the activity monitors showed that subjects complied
response to trucks on the road ahead), and number of driving sim- with the sleep-restriction protocol. They were inactive for 270±20
ulator crashes (off-road, truck collision, or stoppage events). The minutes on the night before testing. Subjects rated themselves as
mean number of crashes was determined for each 4.5-minute bin moderately sleepy according to the Stanford Sleepiness Scale on
and for the whole task. both experimental nights (PSD mean 4.0±1.2 and PSD-A mean
EEG Microsleep Analysis MWT mean sleep latency was significantly lower at the 2:15
am trial than at 11:30 PM trial in both the experimental conditions
The EEG (C3-A2) during the driving simulation task was as- (P<.001). There was a trend for a reduction in the mean sleep
sessed for the appearance of microsleeps. A microsleep was de- latency with the consumption of alcohol, but the difference was
fined as a burst of EEG theta activity greater than 3 seconds in not statistically significant (P=.07; See Table 1). Forty percent
duration.18,19 The cumulative theta time (seconds) was determined and 80% of subjects achieved sleep onset in the PSD and PSD-A
for each subject. conditions, respectively.
The subjects’ mean BACs on the alcohol night at the start and
Correlations end of the 70-minutes driving simulation were 0.037±0.011g/dL
Pearson correlations were conducted between the driving sim- and 0.021±0.009g/dL. Before and after MWT1, subjects’ mean
ulator parameters (mean steering deviation, reaction time, and
crashes), mean cumulative microsleep, and MWT sleep latency Table 1—Sleep Latency on the Maintenance of Wakefulness Test
for both conditions.
Sleep Deprivation Sleep Deprivation and Alcohol
MWT1 MWT2 Mean MWT1 MWT2 Mean
ROC Curve Analysis 31.7±11.9 25.2±16.2 28.9±13.2 25.4±16.1 19.3±14.6 22.3±14.8
Receiver operating characteristic (ROC) curves were con-
Sleep latency is presented in minutes, mean + SD; Maintenance of
structed to assess the ability of MWT to distinguish those sub-
Wakefulness Test 1 (MWT1) occurred at 11:30 PM and MWT2 oc-
jects who had a crash during the simulation task from those who curred at 2:15 AM.
SLEEP, Vol. 28, No. 11, 2005 1383 MWT and Driving Simulator Performance—Banks et al
Table 2—Correlations Between Driving Simulator Parameters and
Sleep Latency on the Maintenance of Wakefulness Test curve= 0.81
Sleep Deprivation Sleep Deprivation
MWT1 MWT2 MWT1 MWT2
(True +ve rate)
Microsleeps -0.02 0.00 -0.45* -0.43
Reaction time -0.52* -0.47 -0.51* -0.15
Mean steering deviation -0.24 -0.27 -0.59** -0.25
Number of crashes -0.19 -0.14 -0.54* -0.27
Maintenance of Wakefulness Test 1 (MWT1) occurred at 11:30 PM
and (MWT2) occurred at 2:15 AM.
BACs were 0.031±0.01g/dL and 0.026±0.012g/dL, and, before
and after MWT2, subjects’ mean BACs were 0.021±0.009g/dL 1-Specificity
and 0.012±0.008g/dL. (False +ve rate)
In the PSD condition only driving simulator reaction time to Figure 1—The true positive rate (short Maintenance of Wakefulness
the appearance of trucks correlated with MWT1 (see Table 2). Test [MWT] sleep latency and presence of a driving simulator crash)
In the combined PSD-A condition, however, the driving simu- versus false positive rate (short MWT sleep latency without subjects
lator parameters and the duration of microsleeps correlated with having crashed) of the MWT directly before the driving task during
MWT sleep latency but only at the 11:30 PM trial, with the MWT1 the combined sleep restriction and alcohol condition.
explaining between 20% to 35% of the variance of the driving but the difference was not statistically significant, consistent with
simulator parameters (see Table 2). increased variance in sleep-latency after alcohol consumption.
A ROC curve (see figure 1) revealed that the MWT trial di- We postulate that increased variance in subjects’ sleepiness after
rectly before the driving task in the PSD-A condition was able to alcohol consumption allowed the relationship between MWT and
discriminate between subjects who crashed and those who did not driving simulator performance to become evident. The increased
(Mann-Whitney U test, P=.006). The area under the curve was variance in sleepiness after alcohol consumption was most like-
0.81±0.10. Neither the MWT before or after the driving task in ly due to more subjects falling asleep and less truncation of the
the PSD condition (P=.22 and P=.32, respectively) nor the MWT sample to 40 minutes. This relationship needs further examina-
after the driving task in the PSD-A condition (P=.23) significantly tion in patient populations (eg, those with obstructive sleep apnea
discriminated between the 2 conditions (crash, no-crash). and narcolepsy) who have existing trait sleepiness rather than ex-
perimentally induced sleepiness in normal subjects with alcohol.
DISCUSSION Alcohol may have also affected MWT sleep latency by reduc-
ing subject compliance and motivation to follow instructions and
This study found that MWT1 correlated with reaction time in
remain vigilant. There is some literature to suggest that alcohol
both conditions, with a slightly greater magnitude of relationship
reduces subjects’ general motivation to perform.21
in the PSD-A condition. Other parameters such as microsleeps,
The MWT and driving simulator reaction-time results corre-
steering deviation, and crashes, only correlated with MWT1 in
lated well in both conditions, with MWT sleep latency accounting
the PSD-A condition. The MWT was found to explain a fifth to a
for just under one third of the variance in reaction time. Reaction
third of the variance in the driving-performance measures in the
time has been found to be very sensitive to sleep loss and alcohol
PSD-A condition when the subjects were more sleepy and im-
consumption20,22 and is recognized as a good measure of perfor-
paired due to the alcohol consumption. The MWT trial prior to the
mance.23-29 Reaction time is a vital component of driving perfor-
driving task in the PSD-A condition was able to discern subjects
mance. The correlation between MWT sleep latency and reaction
who had a crash during the driving task from those who did not.
time increases the validity of the MWT as an indicator of driving
These results suggest that there is validity in using the MWT to
predict driving performance.
A limitation of this study was that polysomnography was not
The subjects’ BACs during the first MWT on the PSD-A night
conducted in the subjects to completely rule out the presence of a
were close to those during the driving task (0.031±0.010g/dL be-
sleep disorder. However, every effort was made to ensure that the
fore MWT1 and 0.037±0.011g/dL before the driving simulation
results of all screening tools used (eg, the Pittsburg Sleep Qual-
task). However, during the second MWT, the average BACs were
ity Inventory and Epworth Sleepiness Scale) were within normal
much lower (0.021±0.009g/dL before MWT2). The higher BACs
limits. Additionally, the subjects were young, had normal body
likely contributed to the increased degree of relationship between
mass indexes, and wore actigraphy for several weeks, confirming
the first MWT and the subjects’ performance during the driving
that all subjects had normal sleep-wake times. It remains pos-
sible, however, that a subject may have had a sleep disorder that
The consumption of alcohol was an important factor in this
effected their MWT results and driving simulation performance.
study. Low doses of alcohol have been found to increase sleepi-
In conclusion, these results indicate that MWT sleep latency
ness and performance impairment in the already sleep-restricted
is a reasonable predictor of driving simulator performance in
individual.11,20 In the current study, there was a trend for a reduc-
sleepy, alcohol-impaired, normal subjects. This study represents
tion in the mean sleep latency with the consumption of alcohol,
a first step in examining the relationship between MWT and driv-
SLEEP, Vol. 28, No. 11, 2005 1384 MWT and Driving Simulator Performance—Banks et al
ing performance. We reasoned that conducting the MWT trials in 2001;24:A260.
close proximity to a driving task in the early hours of the morning 16. Hoddes E, Zarcone V, Smythe H, Phillips R, Dement WC. Quan-
when sleep-restricted subjects are under the most circadian pres- tification of sleepiness: a new approach. Psychophysiology
sure for sleep would strengthen relationships between MWT sleep 1973;10:431-6.
17. Banks S, Lack L, McEvoy RD. Perception of drowsiness while
latency and measures of performance. Normal sleep laboratory
driving: Preliminary analysis. Sleep 2003;26:A190-1.
practice, however, is to conduct 4 evenly spaced MWT trials in 18. Itoi A, Cilveti R, Voth M, et al. Can drivers avoid falling asleep at
the daytime. It will be important in future studies, therefore, to ex- the wheel? Relationship between awareness of sleepiness and abil-
plore the relationship between daytime MWT mean sleep latency ity to predict sleep onset. Washington: AAA Foundation For Traffic
and driving simulator performance in sleepy patient populations Safety; 1993:1-22.
and to compare results in older versus younger subjects. Ideally 19. Guilleminault C, Billiard M, Montplaisir J, Dement WC. Altered
the MWT should also be directly compared to real-life, on-road, states of consciousness in disorders of daytime sleepiness. J Neurol
driving performance. Sci 1975;26:377-93.
20. Roehrs T, Zwyghuizen-Doorenbos A, Knox M, Moskowitz H, Roth
T. Sedating effects of ethanol and time of drinking. Alcohol Clin
Exp Res 1992;16:553-7.
Study funded by the Brewers’ Foundation of Australia 21. Farquhar K, Lambert K, Drummond GB, Tiplady B, Wright P. Ef-
fect of ethanol on psychomotor performance and on risk taking be-
haviour. J Psychopharmacol 2002;16:379-84.
22. Philip P, Taillard J, Quera-Salva MA, Bioulac B, Akerstedt T. Sim-
1. Lyznicki JM, Doege TC, Davis RM, Williams MA, Sleepiness, ple reaction time, duration of driving and sleep deprivation in young
driving, and motor vehicle crashes. Council on Scientific Affairs, versus old automobile drivers. J Sleep Res 1999;8:9-14.
American Medical Association. JAMA 1998;279:1908-13. 23. Lubin RA. Influences of alcohol upon performance and perfor-
2. Dobbie, K. Fatigue related crashes: An analysis of fatigue related mance awareness. Percept Mot Skills 1977;45:303-10.
crashes on Australian roads using an operational definition of fa- 24. Horne JA, Gibbons H. Effects on vigilance performance and sleepi-
tigue. Australian Transport Safety Bureau; 2002:30. ness of alcohol given in the early afternoon (‘post lunch’) vs. early
3. Horne JA, Reyner LA. Sleep related vehicle accidents. BMJ evening. Ergonomics 1991;34:67-77.
1995;310:565-7. 25. Roehrs T, Beare D, Zorick F, Roth T. Sleepiness and ethanol effects
4. George CF, Nickerson PW, Hanly PJ, Millar TW, Kryger MH. on simulated driving. Alcohol Clin Exp Res 1994;18:154-8.
Sleep apnoea patients have more automobile accidents. Lancet 26. Powell NB, Riley RW, Schechtman KB, Blumen MB, Dinges DF,
1987;2:447. Guilleminault CA. Comparative model: reaction time performance
5. Findley LJ, Unverzagt ME, Suratt PM. Automobile accidents in- in sleep-disordered breathing versus alcohol-impaired controls. La-
volving patients with obstructive sleep apnea. Am Rev Respir Dis ryngoscope 1999;109:1648-54.
1988;138:337-40. 27. Dinges DF, Powell JW. Microcomputer analyses of performance on
6. AUSROADS, Assessing fitness to drive for commercial and private a portable, simple visual RT task during sustained operations. Be-
vehicle drivers; Guidelines and standards for health professionals in hav Res Meth Instrum Comp 1985;17:652-5.
Australia, 3rd ed. Sydney: Ausroads; 2003. 28. Jewett ME, Dijk DJ, Kronauer RE, Dinges DF. Dose-response rela-
7. Grossman A, Barenboim E, Azaria B, Sherer Y, Goldstein L. The tionship between sleep duration and human psychomotor vigilance
maintenance of wakefulness test as a predictor of alertness in air- and subjective alertness. Sleep 1999;22:171-9.
crew members with idiopathic hypersomnia. Aviat Space Environ 29. Van Dongen HP, Maislin G, Mullington JM, Dinges DF. The cu-
Med 2004;75:281-3. mulative cost of additional wakefulness: dose-response effects on
8. Wolfe LF, Reid KJ, Jankelowitz LM, Zec P. Maintenance of Wake- neurobehavioral functions and sleep physiology from chronic sleep
fulness Testing (MWT) and driving simulator performance in Ob- restriction and total sleep deprivation. Sleep 2003;26:117-26.
structive Sleep Apnea (OSA). Am J Respir Crit Care Med 2002;165:
9. Kingshott RN, Cowan JO, Jones DR, et al. The role of sleep-disor-
dered breathing, daytime sleepiness, and impaired performance in
motor vehicle crashes-a case control study. Sleep Breath, 2004;8:61-
10. George CF, Boudreau AC, Smiley A. Simulated driving perfor-
mance in patients with obstructive sleep apnea. Am J Respir Crit
Care Med 1996;154:175-81.
11. Banks S, Catcheside P, Lack L, Grunstein RR, McEvoy RD. Low
levels of alcohol impair driving simulator performance and reduce
perception of crash risk in partially sleep deprived subjects. Sleep
12. Johns MW. A new method for measuring daytime sleepiness: the
Epworth sleepiness scale. Sleep 1991;14:540-5.
13. Howard M, Gora J, Swann P, Pierce R. Evidence for poor perception
of sleepiness in professional drivers. Sleep 2002;25:A146.
14. Howard M, Gora J, Swann P, Pierce R. Alpha and theta activity and
slow eye closure are related to driving performance in professional
drivers. Sleep 2002;25:A148.
15. Newcombe J, Desai A, Joffe D, Engleman H, Seale J, Grunstein R.
Modafinil improves alertness and driving simulator performance in
sleep-deprived mild obstructive sleep apnoea (OSA) subjects. Sleep
SLEEP, Vol. 28, No. 11, 2005 1385 MWT and Driving Simulator Performance—Banks et al