International Journal of Exercise Science 10(5): 764-773, 2017. The purpose of this study was to determine step-count accuracy of pedometers at different walking speeds. Ten recreationally active participants walked at five treadmill speeds (0.89, 1.11, 1.34, 1.56, and 1.79 m/s) for five minutes while wearing four wrist-worn activity trackers (Fitbit Charge HR®, Garmin Vivosmart HR®, Apple iWatch®, Jawbone UP3®) and the hip-worn Digi-Walker®. Each step was manually counted by a research technician (benchmark). Total step count at each speed was obtained for each device and compared to the benchmark using one-way MANOVA and Pearson correlation coefficient. For all five speeds, the Digi-Walker® yielded the most accurate values, averaging -0.4% difference from the benchmark counted steps, and showed the strongest correlation, r >.730, p <.05, at every speed. The Fitbit averaged the highest percent difference of -10.2% from the benchmark of counted steps, and underestimated steps at all speeds (p <0.05). Garmin averaged a -2.7% step difference, Jawbone averaged a -5.3% step difference, and the iWatch showed a -7.9% step difference. Specifically, the Fitbit, Garmin, and Jawbone got progressively worse with increasing speed, whereas the iWatch performed the worst at the slowest and fastest speeds. All wrist-worn devices tested tended to underestimate steps. These data indicate that wrist-worn pedometers are inaccurate even with a specific designed purpose: count steps in a controlled manner. Because these devices are inaccurate in this setting, they remain highly questionable for accuracy in a real-world setting in which the definition of a “step” becomes less finite.
Sears, Tiffany; Alvalos, Elmer; Lawson, Samantha; McAlister, Ian; Eschbach, L. Chris; and Bunn, Jennifer
"Wrist-Worn Physical Activity Trackers Tend To Underestimate Steps During Walking,"
International Journal of Exercise Science: Vol. 10
5, Pages 764 - 773.
Available at: http://digitalcommons.wku.edu/ijes/vol10/iss5/12