File(s) not publicly available
Two for one? : Validating an energy expenditure monitor against polysomnography for sleep/wake measurement
conference contributionposted on 06.12.2017, 00:00 by Anastasi Kosmadopoulos, Charli Sargent, Xuan Zhou, David Darwent, Gregory Roach
Gold standard sleep measurement, polysomnography (PSG), is not always practical for field-based research due to time and expense. Wristwatch-like activity monitors that estimate sleep from wrist movement are often used instead. The Actical is an activity monitor designed to estimate energy expenditure. While not specifically designed to measure sleep, it records and stores data in a similar fashion to devices that are. Dual functionality could be useful for research spanning sleep and exercise behaviours, and thus the aim was to evaluate the Actical as an alternative to PSG for field research. Methods: Ten participants spent one night in bed (22:00-07:30 h) in temperature-controlled (21 1C) bedrooms. Each wore an Actical, set to record activity in 30-s epochs, on their non-dominant wrist, and electrodes were placed in the 10-20 arrangement for PSG. PSG data were manually scored as sleep/wake in 30-s epochs. Software provided sleep/wake estimates for Actical epochs at a medium- (default) and a low-activity threshold for wake. Actical sleep/wake data were compared with data from PSG on an epoch-by-epoch basis and in terms of total sleep time (TST), sleep efficiency (SE), and wake after sleep onset (WASO). Results: Actical identified 97 2% (M SD) and 98 1% of PSG-determined sleep epochs, at the low threshold and medium threshold for wake, respectively. Identification of PSG-determined wake was 31 13% and 22 9%, and overall agreements were 89 10% and 88 11%, respectively. TST was overestimated by 25 47 and 36 52 min, and SE by 5 8% and 6 9%; WASO was underestimated by 25 52 and 36 57 min, respectively. Discussion: Results suggest that overall epoch agreement between Acticals and PSG is good and identification of sleep is equally high at both thresholds; however, the lower threshold is better than the default for identifying wake. In addition, the biases of sleep parameters calculated using the Actical were smaller at the low threshold. The results indicated that the Actical is a satisfactory alternative tool for measuring sleep, particularly in field research.