Charli Sargent1, Michele Lastella2, Shona L Halson3, Gregory D Roach2. 1. Appleton Institute for Behavioural Science, Central Queensland University, Australia. Electronic address: charli.sargent@cqu.edu.au. 2. Appleton Institute for Behavioural Science, Central Queensland University, Australia. 3. Department of Physiology, Australian Institute of Sport, Australia.
Abstract
OBJECTIVES: There is a growing interest in monitoring the sleep of elite athletes. Polysomnography is considered the gold standard for measuring sleep, however this technique is impractical if the aim is to collect data simultaneously with multiple athletes over consecutive nights. Activity monitors may be a suitable alternative for monitoring sleep, but these devices have not been validated against polysomnography in a population of elite athletes. DESIGN: Participants (n=16) were endurance-trained cyclists participating in a 6-week training camp. METHODS: A total of 122 nights of sleep were recorded with polysomnography and activity monitors simultaneously. Agreement, sensitivity, and specificity were calculated from epoch-for-epoch comparisons of polysomnography and activity monitor data. Sleep variables derived from polysomnography and activity monitors were compared using paired t-tests. Activity monitor data were analysed using low, medium, and high sleep-wake thresholds. RESULTS: Epoch-for-epoch comparisons showed good agreement between activity monitors and polysomnography for each sleep-wake threshold (81-90%). Activity monitors were sensitive to sleep (81-92%), but specificity differed depending on the threshold applied (67-82%). Activity monitors underestimated sleep duration (18-90min) and overestimated wake duration (4-77min) depending on the threshold applied. CONCLUSIONS: Applying the correct sleep-wake threshold is important when using activity monitors to measure the sleep of elite athletes. For example, the default sleep-wake threshold (>40 activity counts=wake) underestimates sleep duration by ∼50min and overestimates wake duration by ∼40min. In contrast, sleep-wake thresholds that have a high sensitivity to sleep (>80 activity counts=wake) yield the best combination of agreement, sensitivity, and specificity.
OBJECTIVES: There is a growing interest in monitoring the sleep of elite athletes. Polysomnography is considered the gold standard for measuring sleep, however this technique is impractical if the aim is to collect data simultaneously with multiple athletes over consecutive nights. Activity monitors may be a suitable alternative for monitoring sleep, but these devices have not been validated against polysomnography in a population of elite athletes. DESIGN:Participants (n=16) were endurance-trained cyclists participating in a 6-week training camp. METHODS: A total of 122 nights of sleep were recorded with polysomnography and activity monitors simultaneously. Agreement, sensitivity, and specificity were calculated from epoch-for-epoch comparisons of polysomnography and activity monitor data. Sleep variables derived from polysomnography and activity monitors were compared using paired t-tests. Activity monitor data were analysed using low, medium, and high sleep-wake thresholds. RESULTS: Epoch-for-epoch comparisons showed good agreement between activity monitors and polysomnography for each sleep-wake threshold (81-90%). Activity monitors were sensitive to sleep (81-92%), but specificity differed depending on the threshold applied (67-82%). Activity monitors underestimated sleep duration (18-90min) and overestimated wake duration (4-77min) depending on the threshold applied. CONCLUSIONS: Applying the correct sleep-wake threshold is important when using activity monitors to measure the sleep of elite athletes. For example, the default sleep-wake threshold (>40 activity counts=wake) underestimates sleep duration by ∼50min and overestimates wake duration by ∼40min. In contrast, sleep-wake thresholds that have a high sensitivity to sleep (>80 activity counts=wake) yield the best combination of agreement, sensitivity, and specificity.
Authors: Melanie Knufinke; Arne Nieuwenhuys; Kamiel Maase; Maarten H Moen; Sabine A E Geurts; Anton M L Coenen; Michiel A J Kompier Journal: J Sports Sci Med Date: 2018-11-20 Impact factor: 2.988