Total sleep deprivation results in an increase in driving accidents as measured by a high fidelity driving simulator. The purpose of the present study was to determine the effects of schedules of varying amounts of sleep on measures of alertness and performance, including simulator driving performance. An analysis of the effects of cumulative partial sleep deprivation on total number of driving accidents and the extent to which individual accidents were immediately preceded by microsleep are reported here. Sixty-six commercial drivers (50 males/16 females, ages 24-62) participated in the 14-day residential study. The first three days involved training/practice on study measures and 8 hours of nightly sleep. For the following 7 experimental (Exp.) nights, subjects were randomly assigned to 9, 7, 5, or 3 hours in bed. For the 4 recovery (Rec.) nights, all had 8 hours of nightly sleep. Daily driving tests using the STISIM PC-based driving simulator were administered 4-7 times per day, depending on sleep group, for 45 minutes per session. STISIM provided a fully interactive audiovisual display responsive to the driver's steering, braking, and throttle inputs. The scenario allowed for multiple accidents to occur in each driving session. Crash events were time marked on the polysomnograph recordings. Off-road and on-road accidents across comparable driving sessions were combined for the purpose of analysis. Corresponding EEG records of the driving sessions during which accidents occurred were visually scored in their entirety for the presence of microsleep, defined as 1-15 sec. episodes of Stage 1 sleep. Microsleep scorers were blind to sleep condition, number of accidents, and accident times. Data for accident times and microsleep times were compared to determine any temporal correspondence between the two events. A total of 777 accidents occurred during Baseline, Exp. 1-7, and Rec. 1-3 for the 9, 7, 5, and 3 hour sleep groups (80, 136, 182, and 379 accidents, respectively). A chi-square analysis revealed a significant (p < .000) increase in the number of driving accidents as a result of cumulative partial sleep deprivation. The percentage of crashes that were immediately preceded by a microsleep event (ME) is given for the groups by study day in the table. During the 7-day experimental sleep phase, a total of 619 accidents occurred across the four groups. Of these accidents, only 8 or 1% directly corresponded to a microsleep event. When an entire minute prior to each accident was considered, then 85 or 14% of these accidents were preceded by a microsleep event. (A)
Abstract