What is the Agreement Between Two Generations of Commercial Accelerometer in a Free-Living Environment for Young to Middle-Aged Adults?

Denise E. Jones, Harvi F. Hart, Kay M. Crossley, Ilana N. Ackerman, Joanne L Kemp

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Background: Rapid change in the commercial market can threaten consistency of activity data comparisons as devices are superseded. Purpose: To determine the level of agreement between two generations of Fitbit™ device for step count and activity level in a free-living environment. Methods: Thirty-seven healthy participants (17 women, 20 men; M ± SD: age 34 ± 8 y; body mass index 25.4 ± 3.9 kg/m2) wore a Fitbit Flex™ and Flex 2™ on their non-dominant wrist over two weeks in a free-living environment. A waist-mounted ActiGraph GT3X+ was also worn to provide a comparison of step count data obtained against a commercial device. Results: Comparison of step count between two generations of Fitbit™ device (Mean Absolute Percentage Error, 12%; Standard Error of Mean, 102.58 steps/d (p = .039); ICC = 0.955) showed closer inter-device agreement than comparison of step count data between commercial (Fitbit™) and research (ActiGraph GT3X+) grades of device (Mean Absolute Percentage Error, 31%; Standard Error of Mean, 124.6 steps/d (p < .001); ICC = 0.915). Statistically significant differences were identified for the Standard Error of Mean between generations of Fitbit™ device (p = .039) and grades of device (p < .001). A comparison of ‘fairly’ and ‘very’ active minutes showed no statistically significant difference between generations of Fitbit™ (p = .980); Mean Absolute Percentage Error, 38%; ICC = 0.908. The number of days of data captured for step count was comparable between to the two grades of device. Conclusion: Users should be aware of potential variations in data estimates from different generations of Fitbit™ device, with step count data providing a more consistent comparison metric.
Original languageEnglish
Pages (from-to)1-9
Number of pages9
JournalJournal for the Measurement of Physical Behaviour
DOIs
Publication statusAccepted/In press - 2019

Cite this

@article{94cc451ac29d45f8bbc6ae71e220af82,
title = "What is the Agreement Between Two Generations of Commercial Accelerometer in a Free-Living Environment for Young to Middle-Aged Adults?",
abstract = "Background: Rapid change in the commercial market can threaten consistency of activity data comparisons as devices are superseded. Purpose: To determine the level of agreement between two generations of Fitbit™ device for step count and activity level in a free-living environment. Methods: Thirty-seven healthy participants (17 women, 20 men; M ± SD: age 34 ± 8 y; body mass index 25.4 ± 3.9 kg/m2) wore a Fitbit Flex™ and Flex 2™ on their non-dominant wrist over two weeks in a free-living environment. A waist-mounted ActiGraph GT3X+ was also worn to provide a comparison of step count data obtained against a commercial device. Results: Comparison of step count between two generations of Fitbit™ device (Mean Absolute Percentage Error, 12{\%}; Standard Error of Mean, 102.58 steps/d (p = .039); ICC = 0.955) showed closer inter-device agreement than comparison of step count data between commercial (Fitbit™) and research (ActiGraph GT3X+) grades of device (Mean Absolute Percentage Error, 31{\%}; Standard Error of Mean, 124.6 steps/d (p < .001); ICC = 0.915). Statistically significant differences were identified for the Standard Error of Mean between generations of Fitbit™ device (p = .039) and grades of device (p < .001). A comparison of ‘fairly’ and ‘very’ active minutes showed no statistically significant difference between generations of Fitbit™ (p = .980); Mean Absolute Percentage Error, 38{\%}; ICC = 0.908. The number of days of data captured for step count was comparable between to the two grades of device. Conclusion: Users should be aware of potential variations in data estimates from different generations of Fitbit™ device, with step count data providing a more consistent comparison metric.",
author = "Jones, {Denise E.} and Hart, {Harvi F.} and Crossley, {Kay M.} and Ackerman, {Ilana N.} and Kemp, {Joanne L}",
year = "2019",
doi = "10.1123/jmpb.2018-0064",
language = "English",
pages = "1--9",
journal = "Journal for the Measurement of Physical Behaviour",
issn = "2575-6605",
publisher = "Human Kinetics Publishers",

}

What is the Agreement Between Two Generations of Commercial Accelerometer in a Free-Living Environment for Young to Middle-Aged Adults? / Jones, Denise E.; Hart, Harvi F. ; Crossley, Kay M.; Ackerman, Ilana N.; Kemp, Joanne L.

In: Journal for the Measurement of Physical Behaviour, 2019, p. 1-9.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - What is the Agreement Between Two Generations of Commercial Accelerometer in a Free-Living Environment for Young to Middle-Aged Adults?

AU - Jones, Denise E.

AU - Hart, Harvi F.

AU - Crossley, Kay M.

AU - Ackerman, Ilana N.

AU - Kemp, Joanne L

PY - 2019

Y1 - 2019

N2 - Background: Rapid change in the commercial market can threaten consistency of activity data comparisons as devices are superseded. Purpose: To determine the level of agreement between two generations of Fitbit™ device for step count and activity level in a free-living environment. Methods: Thirty-seven healthy participants (17 women, 20 men; M ± SD: age 34 ± 8 y; body mass index 25.4 ± 3.9 kg/m2) wore a Fitbit Flex™ and Flex 2™ on their non-dominant wrist over two weeks in a free-living environment. A waist-mounted ActiGraph GT3X+ was also worn to provide a comparison of step count data obtained against a commercial device. Results: Comparison of step count between two generations of Fitbit™ device (Mean Absolute Percentage Error, 12%; Standard Error of Mean, 102.58 steps/d (p = .039); ICC = 0.955) showed closer inter-device agreement than comparison of step count data between commercial (Fitbit™) and research (ActiGraph GT3X+) grades of device (Mean Absolute Percentage Error, 31%; Standard Error of Mean, 124.6 steps/d (p < .001); ICC = 0.915). Statistically significant differences were identified for the Standard Error of Mean between generations of Fitbit™ device (p = .039) and grades of device (p < .001). A comparison of ‘fairly’ and ‘very’ active minutes showed no statistically significant difference between generations of Fitbit™ (p = .980); Mean Absolute Percentage Error, 38%; ICC = 0.908. The number of days of data captured for step count was comparable between to the two grades of device. Conclusion: Users should be aware of potential variations in data estimates from different generations of Fitbit™ device, with step count data providing a more consistent comparison metric.

AB - Background: Rapid change in the commercial market can threaten consistency of activity data comparisons as devices are superseded. Purpose: To determine the level of agreement between two generations of Fitbit™ device for step count and activity level in a free-living environment. Methods: Thirty-seven healthy participants (17 women, 20 men; M ± SD: age 34 ± 8 y; body mass index 25.4 ± 3.9 kg/m2) wore a Fitbit Flex™ and Flex 2™ on their non-dominant wrist over two weeks in a free-living environment. A waist-mounted ActiGraph GT3X+ was also worn to provide a comparison of step count data obtained against a commercial device. Results: Comparison of step count between two generations of Fitbit™ device (Mean Absolute Percentage Error, 12%; Standard Error of Mean, 102.58 steps/d (p = .039); ICC = 0.955) showed closer inter-device agreement than comparison of step count data between commercial (Fitbit™) and research (ActiGraph GT3X+) grades of device (Mean Absolute Percentage Error, 31%; Standard Error of Mean, 124.6 steps/d (p < .001); ICC = 0.915). Statistically significant differences were identified for the Standard Error of Mean between generations of Fitbit™ device (p = .039) and grades of device (p < .001). A comparison of ‘fairly’ and ‘very’ active minutes showed no statistically significant difference between generations of Fitbit™ (p = .980); Mean Absolute Percentage Error, 38%; ICC = 0.908. The number of days of data captured for step count was comparable between to the two grades of device. Conclusion: Users should be aware of potential variations in data estimates from different generations of Fitbit™ device, with step count data providing a more consistent comparison metric.

U2 - 10.1123/jmpb.2018-0064

DO - 10.1123/jmpb.2018-0064

M3 - Article

SP - 1

EP - 9

JO - Journal for the Measurement of Physical Behaviour

JF - Journal for the Measurement of Physical Behaviour

SN - 2575-6605

ER -