Step Count and Sedentary Time Validation of Consumer Activity Trackers and a Pedometer in Free-Living Settings

in Journal for the Measurement of Physical Behaviour
Restricted access

Purchase article

USD  $24.95

Student 1 year subscription

USD  $37.00

1 year subscription

USD  $50.00

Student 2 year subscription

USD  $71.00

2 year subscription

USD  $93.00

We evaluated the accuracy and precision of wearable activity trackers and a pedometer (ATPs) in estimating steps and sedentary time (ST) in free-living settings. Thirty-two healthy men and women (M ± SD: age = 32.3 ± 13.3 years; BMI = 24.4 ± 3.3 kg·m−2) were directly observed during three, 2-hour sessions on different days while wearing 10 devices and a biometric shirt. A validated direct observation (DO) system provided criterion measures for steps and ST. For steps, bias ranged from −753 steps/2-hrs (Fitbit Flex) to −57 steps/2-hrs (Polar Loop) and CIs ranged from [−1,144, −365] (Fitbit Flex) to [−291,175] (Polar Loop) steps/2-hrs. For all devices, step estimates were strongly correlated (r = 0.90 [Fitbit Flex] to r = 0.97 [New Lifestyles pedometer model 1000]) with DO counted steps. Estimates of ST were not accurate and were weakly correlated (r = −0.06 and r = 0.06 for Fitbit Flex and Fitbit One, respectively) with DO ST. Most ATPs were not accurate and varied in precision in estimating steps and ST in free-living settings. Implications from this study are that although point estimates of steps from ATPs are not accurate, ATPs’ ranking of step counts among individuals was high. However, the Fitbit Flex and Fitbit One are not recommended for estimating ST. This study advances our understanding of the performance of ATPs in estimating steps and ST in free-living settings, and significantly advances activity tracker and pedometer validation studies.

Mendoza is with the Department of Kinesiology, California State University, East Bay, Hayward, CA. Lyden is with KAL Research, Consulting, LLC, Denver, CO. Sirard, Tudor-Locke, and Freedson are with the Department of Kinesiology; Staudenmayer is with the Department of Mathematics and Statistics; University of Massachusetts, Amherst.

Freedson (psf@kin.umass.edu) is corresponding author.

Supplementary Materials

    • Supplementary Figure 1 (PDF 140 KB)
    • Supplementary Figure 2 (PDF 126 KB)
    • Supplementary Figure 3 (PDF 127 KB)
    • Supplementary Figure 4 (PDF 114 KB)
    • Supplementary Table 1 (PDF 217 KB)