Modeling fatigue-related truck accidents : prior sleep duration, recency and continuity

A review of fatigue-management approaches proposed an alternative to hours of service (HOS) regulations based on prior sleep and wakefulness (PSW). This approach states that an individual must have ≥X h and ≥Y h sleep in the 24-h and 48-h periods prior to work, respectively; and that the continuous period for which they have been awake by the end of the shift must be less than or equal to the amount of sleep they have had in the prior 48 h. An a priori model with X = 5 and Y = 12 was suggested. This study investigated the ability to predict fatigue-related accidents using: (a) the a priori model; (b) an optimized model (manipulating X and Y); (c) models with continuous or split sleep; (d) HOS indicators; and (e) combined PSW and HOS models. Sleep and work histories were collected from 107 truck accidents (male drivers, aged 23–66 y), of which 62 had a probable cause of fatigue. The a priori model correctly classified 65%, and the optimized model (6.5 h in the prior 24 h and 8 h in the prior 48 h) nearly 71% of accidents. A simplified model of 6.5 h continuous sleep in the prior 24 h correctly classified nearly 75%. HOS indicators correctly classified