0779 WRIST-WORN ACTIVITY MONITORING DEVICES OVERESTIMATE SLEEP DURATION AND EFFICIENCY IN HEALTHY ADULTS

R Jumabhoy, PD Dao, S Maskevich, JC Stout, SP Drummond

Research output: Contribution to conferenceAbstractOtherpeer-review

Abstract

Introduction:The use of activity monitors for sleep measurement purposes has increased in research and consumer settings. However, validation of such monitors is lacking. This study examined agreement on total sleep time (TST) and sleep efficiency (SE) between polysomnography and three activity monitors - Actiwatch Spectrum Pro (ACT), Fitbit One (FB) and Jawbone UP2 (JB). Differences between polysomnography and each activity monitor, and differences between research-grade ACT and commercial devices FB and JB, were examined.Methods:Twenty-two healthy adults (Mage =29.3, SDage=11.4) had one night of sleep measured by polysomnography and each activity monitor simultaneously in a laboratory. Minute-by-minute data were extracted and compared. Wilcoxon signed-ranks tests assessed statistical differences between measures, and Bland-Altman analyses examined clinically meaningful differences between measures, using cut-offs of ±30 minutes for TST and ±5% for SE.Results:Compared to polysomnography, all activity monitors significantly overestimated TST and SE. Differences between polysomnography and each monitor were also clinically meaningful, as Bland-Altman upper and lower limits of agreement for TST exceeded clinical cut-offs for ACT (-64.7min-+166.1min), FB (-66.1min-+189.2min), and JB (-103.6min-+186.2min). Similarly, upper and lower limits of agreement for SE exceeded clinical cut-offs for ACT (-13.0%-+34.0%), FB (-13.1%-+38.5%), and JB (-20.7%-+37.9%). Compared to ACT, only FB significantly overestimated TST and SE. However, differences between ACT and each of FB and JB were clinically meaningful. For FB, Bland-Altman upper limits of agreement exceeded clinical cut-offs for TST (-18.7min-+40.2min) and SE (-3.8%-+8.1%). For JB, upper and lower limits of agreement exceeded clinical cut-offs for TST (-100.4min-+79.3min) and SE (-20.1%-+16.0%). Agreement between devices decreased as TST and SE decreased. All monitors demonstrated poor specificity (18.8–35.6%), but high sensitivity (94.2–99.2%).Conclusion:Results suggest these models of ACT, FB, and JB cannot be used interchangeably with polysomnography. When activity monitors must be used, such as in field settings, FB and JB cannot replace research-grade ACT. Overall, users should account for each monitor’s potential to overestimate or underestimate TST and SE to an unacceptable degree. Future research should examine within-subject variability over time to determine whether monitors can be used to track long-term sleep patterns.Support (If Any):N/A.
Original languageEnglish
PagesA288-A289
Number of pages2
DOIs
Publication statusPublished - 28 Apr 2017
EventAnnual Meeting of the Associated Professional Sleep Societies 2017 - Boston, United States
Duration: 3 Jun 20177 Jun 2017
Conference number: 31st

Conference

ConferenceAnnual Meeting of the Associated Professional Sleep Societies 2017
Abbreviated titleSLEEP 2017
CountryUnited States
CityBoston
Period3/06/177/06/17

Cite this

Jumabhoy, R., Dao, PD., Maskevich, S., Stout, JC., & Drummond, SP. (2017). 0779 WRIST-WORN ACTIVITY MONITORING DEVICES OVERESTIMATE SLEEP DURATION AND EFFICIENCY IN HEALTHY ADULTS. A288-A289. Abstract from Annual Meeting of the Associated Professional Sleep Societies 2017, Boston, United States. https://doi.org/10.1093/sleepj/zsx050.778
Jumabhoy, R ; Dao, PD ; Maskevich, S ; Stout, JC ; Drummond, SP. / 0779 WRIST-WORN ACTIVITY MONITORING DEVICES OVERESTIMATE SLEEP DURATION AND EFFICIENCY IN HEALTHY ADULTS. Abstract from Annual Meeting of the Associated Professional Sleep Societies 2017, Boston, United States.2 p.
@conference{e2568f7ddc7740938efd3dfe8cd79c8c,
title = "0779 WRIST-WORN ACTIVITY MONITORING DEVICES OVERESTIMATE SLEEP DURATION AND EFFICIENCY IN HEALTHY ADULTS",
abstract = "Introduction:The use of activity monitors for sleep measurement purposes has increased in research and consumer settings. However, validation of such monitors is lacking. This study examined agreement on total sleep time (TST) and sleep efficiency (SE) between polysomnography and three activity monitors - Actiwatch Spectrum Pro (ACT), Fitbit One (FB) and Jawbone UP2 (JB). Differences between polysomnography and each activity monitor, and differences between research-grade ACT and commercial devices FB and JB, were examined.Methods:Twenty-two healthy adults (Mage =29.3, SDage=11.4) had one night of sleep measured by polysomnography and each activity monitor simultaneously in a laboratory. Minute-by-minute data were extracted and compared. Wilcoxon signed-ranks tests assessed statistical differences between measures, and Bland-Altman analyses examined clinically meaningful differences between measures, using cut-offs of ±30 minutes for TST and ±5{\%} for SE.Results:Compared to polysomnography, all activity monitors significantly overestimated TST and SE. Differences between polysomnography and each monitor were also clinically meaningful, as Bland-Altman upper and lower limits of agreement for TST exceeded clinical cut-offs for ACT (-64.7min-+166.1min), FB (-66.1min-+189.2min), and JB (-103.6min-+186.2min). Similarly, upper and lower limits of agreement for SE exceeded clinical cut-offs for ACT (-13.0{\%}-+34.0{\%}), FB (-13.1{\%}-+38.5{\%}), and JB (-20.7{\%}-+37.9{\%}). Compared to ACT, only FB significantly overestimated TST and SE. However, differences between ACT and each of FB and JB were clinically meaningful. For FB, Bland-Altman upper limits of agreement exceeded clinical cut-offs for TST (-18.7min-+40.2min) and SE (-3.8{\%}-+8.1{\%}). For JB, upper and lower limits of agreement exceeded clinical cut-offs for TST (-100.4min-+79.3min) and SE (-20.1{\%}-+16.0{\%}). Agreement between devices decreased as TST and SE decreased. All monitors demonstrated poor specificity (18.8–35.6{\%}), but high sensitivity (94.2–99.2{\%}).Conclusion:Results suggest these models of ACT, FB, and JB cannot be used interchangeably with polysomnography. When activity monitors must be used, such as in field settings, FB and JB cannot replace research-grade ACT. Overall, users should account for each monitor’s potential to overestimate or underestimate TST and SE to an unacceptable degree. Future research should examine within-subject variability over time to determine whether monitors can be used to track long-term sleep patterns.Support (If Any):N/A.",
author = "R Jumabhoy and PD Dao and S Maskevich and JC Stout and SP Drummond",
year = "2017",
month = "4",
day = "28",
doi = "10.1093/sleepj/zsx050.778",
language = "English",
pages = "A288--A289",
note = "Annual Meeting of the Associated Professional Sleep Societies 2017, SLEEP 2017 ; Conference date: 03-06-2017 Through 07-06-2017",

}

Jumabhoy, R, Dao, PD, Maskevich, S, Stout, JC & Drummond, SP 2017, '0779 WRIST-WORN ACTIVITY MONITORING DEVICES OVERESTIMATE SLEEP DURATION AND EFFICIENCY IN HEALTHY ADULTS' Annual Meeting of the Associated Professional Sleep Societies 2017, Boston, United States, 3/06/17 - 7/06/17, pp. A288-A289. https://doi.org/10.1093/sleepj/zsx050.778

0779 WRIST-WORN ACTIVITY MONITORING DEVICES OVERESTIMATE SLEEP DURATION AND EFFICIENCY IN HEALTHY ADULTS. / Jumabhoy, R; Dao, PD; Maskevich, S; Stout, JC; Drummond, SP.

2017. A288-A289 Abstract from Annual Meeting of the Associated Professional Sleep Societies 2017, Boston, United States.

Research output: Contribution to conferenceAbstractOtherpeer-review

TY - CONF

T1 - 0779 WRIST-WORN ACTIVITY MONITORING DEVICES OVERESTIMATE SLEEP DURATION AND EFFICIENCY IN HEALTHY ADULTS

AU - Jumabhoy, R

AU - Dao, PD

AU - Maskevich, S

AU - Stout, JC

AU - Drummond, SP

PY - 2017/4/28

Y1 - 2017/4/28

N2 - Introduction:The use of activity monitors for sleep measurement purposes has increased in research and consumer settings. However, validation of such monitors is lacking. This study examined agreement on total sleep time (TST) and sleep efficiency (SE) between polysomnography and three activity monitors - Actiwatch Spectrum Pro (ACT), Fitbit One (FB) and Jawbone UP2 (JB). Differences between polysomnography and each activity monitor, and differences between research-grade ACT and commercial devices FB and JB, were examined.Methods:Twenty-two healthy adults (Mage =29.3, SDage=11.4) had one night of sleep measured by polysomnography and each activity monitor simultaneously in a laboratory. Minute-by-minute data were extracted and compared. Wilcoxon signed-ranks tests assessed statistical differences between measures, and Bland-Altman analyses examined clinically meaningful differences between measures, using cut-offs of ±30 minutes for TST and ±5% for SE.Results:Compared to polysomnography, all activity monitors significantly overestimated TST and SE. Differences between polysomnography and each monitor were also clinically meaningful, as Bland-Altman upper and lower limits of agreement for TST exceeded clinical cut-offs for ACT (-64.7min-+166.1min), FB (-66.1min-+189.2min), and JB (-103.6min-+186.2min). Similarly, upper and lower limits of agreement for SE exceeded clinical cut-offs for ACT (-13.0%-+34.0%), FB (-13.1%-+38.5%), and JB (-20.7%-+37.9%). Compared to ACT, only FB significantly overestimated TST and SE. However, differences between ACT and each of FB and JB were clinically meaningful. For FB, Bland-Altman upper limits of agreement exceeded clinical cut-offs for TST (-18.7min-+40.2min) and SE (-3.8%-+8.1%). For JB, upper and lower limits of agreement exceeded clinical cut-offs for TST (-100.4min-+79.3min) and SE (-20.1%-+16.0%). Agreement between devices decreased as TST and SE decreased. All monitors demonstrated poor specificity (18.8–35.6%), but high sensitivity (94.2–99.2%).Conclusion:Results suggest these models of ACT, FB, and JB cannot be used interchangeably with polysomnography. When activity monitors must be used, such as in field settings, FB and JB cannot replace research-grade ACT. Overall, users should account for each monitor’s potential to overestimate or underestimate TST and SE to an unacceptable degree. Future research should examine within-subject variability over time to determine whether monitors can be used to track long-term sleep patterns.Support (If Any):N/A.

AB - Introduction:The use of activity monitors for sleep measurement purposes has increased in research and consumer settings. However, validation of such monitors is lacking. This study examined agreement on total sleep time (TST) and sleep efficiency (SE) between polysomnography and three activity monitors - Actiwatch Spectrum Pro (ACT), Fitbit One (FB) and Jawbone UP2 (JB). Differences between polysomnography and each activity monitor, and differences between research-grade ACT and commercial devices FB and JB, were examined.Methods:Twenty-two healthy adults (Mage =29.3, SDage=11.4) had one night of sleep measured by polysomnography and each activity monitor simultaneously in a laboratory. Minute-by-minute data were extracted and compared. Wilcoxon signed-ranks tests assessed statistical differences between measures, and Bland-Altman analyses examined clinically meaningful differences between measures, using cut-offs of ±30 minutes for TST and ±5% for SE.Results:Compared to polysomnography, all activity monitors significantly overestimated TST and SE. Differences between polysomnography and each monitor were also clinically meaningful, as Bland-Altman upper and lower limits of agreement for TST exceeded clinical cut-offs for ACT (-64.7min-+166.1min), FB (-66.1min-+189.2min), and JB (-103.6min-+186.2min). Similarly, upper and lower limits of agreement for SE exceeded clinical cut-offs for ACT (-13.0%-+34.0%), FB (-13.1%-+38.5%), and JB (-20.7%-+37.9%). Compared to ACT, only FB significantly overestimated TST and SE. However, differences between ACT and each of FB and JB were clinically meaningful. For FB, Bland-Altman upper limits of agreement exceeded clinical cut-offs for TST (-18.7min-+40.2min) and SE (-3.8%-+8.1%). For JB, upper and lower limits of agreement exceeded clinical cut-offs for TST (-100.4min-+79.3min) and SE (-20.1%-+16.0%). Agreement between devices decreased as TST and SE decreased. All monitors demonstrated poor specificity (18.8–35.6%), but high sensitivity (94.2–99.2%).Conclusion:Results suggest these models of ACT, FB, and JB cannot be used interchangeably with polysomnography. When activity monitors must be used, such as in field settings, FB and JB cannot replace research-grade ACT. Overall, users should account for each monitor’s potential to overestimate or underestimate TST and SE to an unacceptable degree. Future research should examine within-subject variability over time to determine whether monitors can be used to track long-term sleep patterns.Support (If Any):N/A.

U2 - 10.1093/sleepj/zsx050.778

DO - 10.1093/sleepj/zsx050.778

M3 - Abstract

SP - A288-A289

ER -

Jumabhoy R, Dao PD, Maskevich S, Stout JC, Drummond SP. 0779 WRIST-WORN ACTIVITY MONITORING DEVICES OVERESTIMATE SLEEP DURATION AND EFFICIENCY IN HEALTHY ADULTS. 2017. Abstract from Annual Meeting of the Associated Professional Sleep Societies 2017, Boston, United States. https://doi.org/10.1093/sleepj/zsx050.778