Estimating sleep duration: performance of open-source processing of actigraphy compared to in-laboratory polysomnography in the community

Kelly Sansom; Amy Reynolds; Joanne McVeigh; Diego R Mazzotti; Satvinder S Dhaliwal; Kathleen Maddison; Jennifer Walsh; Bhajan Singh; Peter Eastwood; Nigel McArdle

doi:10.1093/sleepadvances/zpad028

Estimating sleep duration: performance of open-source processing of actigraphy compared to in-laboratory polysomnography in the community

Sleep Adv. 2023 Jul 20;4(1):zpad028. doi: 10.1093/sleepadvances/zpad028. eCollection 2023.

Authors

Kelly Sansom^{1

2

3}, Amy Reynolds³, Joanne McVeigh^{4

5}, Diego R Mazzotti^{6

7}, Satvinder S Dhaliwal^{8

9

10

11}, Kathleen Maddison^{1

2

12}, Jennifer Walsh^{1

2

12}, Bhajan Singh^{1

2

12}, Peter Eastwood³, Nigel McArdle^{1

2

12}

Affiliations

¹ Centre for Sleep Science, School of Human Sciences, University of Western Australia, Perth, WA, Australia.
² West Australian Sleep Disorders Research Institute, Sir Charles Gairdner Hospital, Perth, WA, Australia.
³ Flinders Health and Medical Research Institute, College of Medicine and Public Health, Flinders University, Adelaide, SA, Australia.
⁴ Curtin School of Allied Health, Faculty of Health Sciences, Curtin University, Perth, WA, Australia.
⁵ Movement Physiology Laboratory, School of Physiology, University of Witwatersrand, South Africa.
⁶ Division of Medical Informatics, Department of Internal Medicine, University of Kansas Medical Center, KS, USA.
⁷ Division of Pulmonary, Critical Care and Sleep Medicine, Department of Internal Medicine, University of Kansas Medical Center, KS, USA.
⁸ Curtin Health Innovation Research Institute, Faculty of Health Sciences, B305, Curtin University, Bentley, WA, Australia.
⁹ Office of the Provost, Singapore University of Social Sciences, Clementi Road, Singapore.
¹⁰ Duke-NUS Medical School, National University of Singapore, 8 College Road, Singapore.
¹¹ Institute for Research in Molecular Medicine (INFORMM), Universiti Sains Malaysia, Pulau Pinang, Malaysia.
¹² Department of Pulmonary Physiology and Sleep Medicine, Sir Charles Gairdner Hospital, Perth, WA, Australia.

Abstract

Comparisons of actigraphy findings between studies are challenging given differences between brand-specific algorithms. This issue may be minimized by using open-source algorithms. However, the accuracy of actigraphy-derived sleep parameters processed in open-source software needs to be assessed against polysomnography (PSG). Middle-aged adults from the Raine Study (n = 835; F 58%; Age 56.7 ± 5.6 years) completed one night of in-laboratory PSG and concurrent actigraphy (GT3X+ ActiGraph). Actigraphic measures of total sleep time (TST) were analyzed and processed using the open-source R-package GENEActiv and GENEA data in R (GGIR) with and without a sleep diary and additionally processed using proprietary software, ActiLife, for comparison. Bias and agreement (intraclass correlation coefficient) between actigraphy and PSG were examined. Common PSG and sleep health variables associated with the discrepancy between actigraphy, and PSG TST were examined using linear regression. Actigraphy, assessed in GGIR, with and without a sleep diary overestimated PSG TST by (mean ± SD) 31.0 ± 50.0 and 26.4 ± 69.0 minutes, respectively. This overestimation was greater (46.8 ± 50.4 minutes) when actigraphy was analyzed in ActiLife. Agreement between actigraphy and PSG TST was poor (ICC = 0.27-0.44) across all three methods of actigraphy analysis. Longer sleep onset latency and longer wakefulness after sleep onset were associated with overestimation of PSG TST. Open-source processing of actigraphy in a middle-aged community population, agreed poorly with PSG and, on average, overestimated TST. TST overestimation increased with increasing wakefulness overnight. Processing of actigraphy without a diary in GGIR was comparable to when a sleep diary was used and comparable to actigraphy processed with proprietary algorithms in ActiLife.

Keywords: accelerometry; methodology; non-proprietary; triaxial.