ActiGraph GT3X+ and Actical Wrist and Hip Worn Accelerometers for Sleep and Wake Indices in Young Children Using an Automated Algorithm: Validation With Polysomnography
Open Access
- 14 January 2020
- journal article
- research article
- Published by Frontiers Media SA in Frontiers in Psychiatry
- Vol. 10, 958
Abstract
Objectives: Our count-scaled algorithm automatically scores sleep across 24 hours to process sleep timing, quantity, and quality. The aim of this study was to validate the algorithm against overnight PSG in children to determine the best site placement for sleep. Methods: 28 children (5–8 years) with no history of sleep disturbance wore two types of accelerometers (ActiGraph GT3X+ and Actical) at two sites (left hip, non-dominant wrist) for 24-h. Data were processed using the count-scaled algorithm. PSG data were collected using an in-home Type 2 device. PSG-actigraphy epoch sensitivity (sleep agreement) and specificity (wake agreement) were determined and sleep outcomes compared for timing (onset and offset), quantity [sleep period time (SPT) and total sleep time (TST)], and quality metrics [sleep efficiency and waking after sleep onset (WASO)]. Results: Overall, sensitivities were high (89.1% to 99.5%) and specificities low (21.1% to 45.7%). Sleep offset was accurately measured by actigraphy, regardless of brand or placement site. By contrast, sleep onset agreed with PSG using hip-positioned but not wrist-positioned devices (difference ActiGraph : PSG 21 min, P < .001; Actical : PSG 14 min, P < .001). The ActiGraph at the wrist accurately detected WASO and sleep efficiency, but under (−34 min, P < .001) and overestimated (5.8%, P < .001) these at the hip. The Actical under- and over-estimated these variables respectively at both sites. Results for TST varied ranging from significant differences to PSG of −26 to 21 min (ActiGraph wrist and hip respectively) and 9 min (ns) to 59 min for Actical (wrist and hip respectively). Conclusion: Overall the count-scaled algorithm produced high sensitivity at the expense of low specificity in comparison with PSG. A best site placement for estimates of all sleep variables could not be determined, but overall the results suggested ActiGraph GT3X+ at the hip may be superior for sleep timing and quantity metrics, whereas the wrist may be superior for sleep quality metrics. Both devices placed at the hip performed well for sleep timing but not for sleep quality. Differences are likely linked to freedom of movement of the wrist vs the trunk (hip) during overnight sleep.Keywords
This publication has 41 references indexed in Scilit:
- Algorithms for using an activity-based accelerometer for identification of infant sleep–wake states during nap studiesSleep Medicine, 2012
- A Catalog of Rules, Variables, and Definitions Applied to Accelerometer Data in the National Health and Nutrition Examination Survey, 2003–2006Preventing Chronic Disease, 2012
- A comparison of actigraphy scoring rules used in pediatric researchSleep Medicine, 2011
- Longitudinal analysis of sleep in relation to BMI and body fat in children: the FLAME studyBMJ, 2011
- Helping children sleepArchives of Disease in Childhood, 2010
- Correlation, Agreement, and Bland–Altman Analysis: Statistical Analysis of Method Comparison StudiesAmerican Journal of Ophthalmology, 2009
- Newly developed waist actigraphy and its sleep/wake scoring algorithmSleep and Biological Rhythms, 2009
- Development of a WHO growth reference for school-aged children and adolescents2007
- The Sleep Disturbance Scale for Children (SDSC) Construct ion and validation of an instrument to evaluate sleep disturbances in childhood and adolescenceJournal of Sleep Research, 1996
- Bias, prevalence and kappaJournal of Clinical Epidemiology, 1993