0153 FATIGUE RISK MANAGEMENT BY PRIOR SLEEP WAKE MODEL (PSWM): TOO EASY TO BE RELIABLE?

Abstract

Introduction: Fatigue from sleep loss and circadian misalignment degrades performance and jeopardizes workplace productivity and safety. Mathematical models have been developed to help predict and manage fatigue, but data needed to inform model predictions are often not available. This constraint has inspired a simple rule of thumb approach, the Prior Sleep Wake Model (PSWM), to determine when fatigue becomes an unacceptable risk in the workplace. The PSWM is based on three simple criteria: <5h sleep in prior 24h; <12h sleep in prior 48h; current time awake exceeds amount of sleep in prior 48h. When these criteria are met, an individual is assumed to be operating with an unsafe level of fatigue. We investigated the effectiveness of the PSWM in a naturalistic field study of commercial motor vehicle drivers. Method(s): Sleep and performance were measured in N=105 truck drivers (ages 24-69; 6 females) for two consecutive duty cycles. Sleep was measured continuously by means of wrist actigraphy. Performance was measured three times per day with a 3min Psychomotor Vigilance Test (PVT-B), for which lapses (RT>355ms) were determined. Mixedeffects ANOVA was used to compare PVT-B performance by whether or not drivers were operating with an unsafe level of fatigue according to the PSWM. Result(s): There were 2,589 performance measurements, of which 6.8% were classified as unsafe due to fatigue per the PSWM. Performance was significantly worse under conditions classified as unsafe due to fatigue (F=8.14, P=0.004). However, for 20.6% of performance measurements not classified as unsafe, performance was worse than the mean of performance measurements classified as unsafe. Conversely, for 51.2% of performance measurements classified as unsafe, performance was better than the mean of performance measurements not classified as unsafe. Conclusion(s): Based on PVT-B performance during a naturalistic field study of commercial motor vehicle drivers, the PSWM exhibited low sensitivity and poor specificity to differentiate relatively safe from unsafe levels of fatigue. The rule of thumb approach of the PSWM does not offer a suitable alternative to mathematical models developed to help predict and manage fatigue.