Fatigue, sleep restriction, and performance in automobile drivers: a controlled study in a natural environment.
Sleep. 2003-05-01; 26(3): 277-80
DOI: 10.1093/sleep/26.3.277
Abstract
OBJECTIVES:
To test the neurobehavioral consequences of sleep restriction combined with fatigue from long-distance driving (1000 Km/600 miles).
DESIGN:
Counterbalanced study involving 3 experimental conditions: laboratory after controlled habitual sleep (8.5 hours), driving after controlled habitual sleep (8.5 hours) (Road 1), and driving after reduced sleep (2 hours) (Road 2).
SETTING:
Sleep laboratory and open French highway.
PARTICIPANTS:
10 male participants (mean age 22 years, range 18-24 years, mean driving distance per year 15000 Km/9000 miles) free of sleep disorders.
MEASUREMENTS:
Simple reaction time, prospective self-assessment of performance, and instantaneous fatigue and sleepiness ratings measured at 2-hour intervals.
RESULTS:
A two-way repeated ANOVA with time of day and condition indicated a significant main effect for time of day (p < 0.05). The interaction between the two factors (condition * time of day) was also significant (p < 0.05). The effects of time of day were significant only in the condition of driving after sleep restriction, (p < 0.05). Under sleep restriction, some drivers presented an increase of 650 milliseconds compared to the laboratory condition, representing an increase of 23 meters in breaking distance at a speed of 75 miles per hour. Correlation analyses showed a significant linear correlation between self-assessment and reaction time in the laboratory condition (r = -0.58, p < 0.01) but not in the road conditions. Self-ratings during the breaks showed a significant increase in instantaneous self-rated fatigue and sleepiness between Road 1 and Road 2 conditions (Wilcoxon’s test, Z = – 6.47, p < 0.0001 and Z = – 6.26, p < 0.0001).
CONCLUSIONS:
Sleep restriction combined with fatigue significantly affects reaction time. The lack of correspondence between reaction time and prospective self-evaluation of performance suggests that self-monitoring in real conditions is poorly reliable.