Accuracy of step count measured by physical activity monitors: The effect of gait speed and anatomical placement site.

INTRODUCTION: Consumer-based physical activity monitors (PAMs) are becoming increasingly popular, with multiple global organisations recommending physical activity levels that equate to 10,000 steps per day for optimal health. We therefore aimed to compare the step count of five PAMs to a visual step count to identify the most accurate monitors at varying gait speeds, along with the optimal anatomical placement site.

METHODS: Participants completed 3min on a treadmill for five speeds (5.0km/h, 6.5km/h, 8.0km/h, 10km/h, 12km/h). An Actigraph wGT3XBT-BT was placed on the waist and wrist, a FitBit One on the waist, and a Fitbit Flex, Fitbit Charge HR and Jawbone UP24 on both wrists. A video of participant's lower limbs was recorded for visual count. Analyses of variance (ANOVAs) were conducted to examine the effects of gait speed and device placement site on step count accuracy.

RESULTS: Thirty-one participants (mean age 24.3±5.2yrs) took part. Step count error ranged from 41.3±13.8% for the wrist-worn Actigraph to only 0.04±4.3% and -0.3±4.0% for the waist-worn Fitbit One and Actigraph, respectively. Across all gait speeds, waist-worn devices achieved better accuracy than those on the wrist (p<0.001). The Jawbone was the most accurate wrist-worn consumer-based device at slower speeds (p=0.026), with the Fitbit Flex, and Fitbit Charge HR increasing in accuracy to match the Jawbone at higher speeds.

CONCLUSION: The accuracy and reliability of consumer-based PAMs and the Actigraph is affected by anatomical placement site and walking speed. The Fitbit One and Actigraph on the waist were the strongest performers across all speeds.