Sleep measurement in flight crew: comparing actigraphic and subjective estimates to polysomnography.

INTRODUCTION: Flight crew sleep is recognized as critical to alertness and safety, but few studies have compared the reliability of the different types of sleep monitoring that are commonly used. The present study compared actigraphic and subjective estimates of sleep to the gold standard, polysomnography (PSG).
METHODS: In-flight (25 episodes) and layover sleep (21 episodes) of 21 flight crew were measured with PSG and actigraphy. Subjective reports were made 30 min after in-flight sleep episodes. Actigraphy data were analyzed at low, medium, and high activity thresholds.
RESULTS: Actigraphic and subjective estimates of sleep duration correlated highly with PSG (range 0.84-0.95), regardless of sleep location or activity threshold. Mean differences were relatively small (-36-20 min), but the 95% confidence intervals of the differences were wide (+/- 71- +/- 103 min), particularly for subjective estimates (+/- 112 min). Actigraphic estimates of sleep efficiency and latency showed moderate to poor correlation with PSG values. Epoch-by-epoch comparisons showed the actigraph was sensitive to sleep (83-95%), but not very specific (34-62%). Kappa values indicated only weak agreement (< 0.4). DISCUSSION: For estimating mean sleep duration, both actigraphic and subjective estimates are sufficiently close to PSG values, but the amount of random error must be considered. Any single estimate may vary by more than 1 h from the mean difference. Neither actigraphy nor subjective estimates are suitable for estimating sleep efficiency and latency. Findings indicate that the performance of the actigraph is not altered in flight, other than the predictable effects associated with shorter, more disturbed sleep.