Validating The Latest Commercial Physical Activity Monitors

Abstract

PURPOSE: Completing 10,000 steps per day is recommended by the World Health Organisation, the US Centre for Disease Control and the Australian National Heart Foundation. Physical activity monitors (PAM) are popular and have become increasingly available, marketed for their ability to count steps. However, limited data is available on the newer monitors (Fitbits/Jawbone) in comparison to the Actigraph, a validated research-based accelerometer, or a visual step counts, at multiple speeds and placement sites. This study aimed to compare these newer PAMs to the Actigraph and a visual count 1) across different speeds; and 2) at different placement sites in a lab based treadmill test. METHODS: Participants wore an Actigraph wGT3X-BT and Fitbit One around the waist and each wrist, as well as a Fitbit Flex, Fitbit Charge HR and Jawbone UP24 on each wrist while performing 5 x 3 minute bouts on a treadmill at multiple speeds (5.0, 6.5, 8.0, 10.0 and 12.0km/h). A video of their lower limbs was recorded for a visual count (criterion measure) to be completed in duplicate. Paired sample t-tests to test for any significant differences and Spearman’s correlations for associations, between the device and visual count were completed. RESULTS: 31 participants (61% male) aged 24.3±5.2yrs ranging in BMI (18.6kg/m2 to 29.9kg/m2) were recruited. Across all speeds the Actigraph (waist) and Fitbit One (waist) were accurate when compared to visual counts (p<0.001), with the Fitbit One (waist) being the most accurate PAM at slower speeds (5.0, 6.5km/h). All PAMs worn around the wrist became more accurate at the faster speeds with the strongest associations at 8.0km/h=Fitbit Charge (r=0.87, p<0.01); 10.0km/h=Fitbit One (r=0.91, p<0.01) and 12.0km/h=Jawbone (r=0.90, p<0.01). Most PAMS, excluding the Fitbit One significantly undercounted at 5.0 and 6.5km/h (p>0.05). At the higher speeds (10.0, 12.0kph) there were no significant differences (p<0.05) between step counts for all devices. CONCLUSION: The step count accuracy of PAMs was affected by speed and placement site. The Fitbit One (waist) was the most accurate PAM across all speeds. Other PAMs worn on the wrist were accurate at the faster speeds only. Future studies should look at the effect of body mass index and outdoor free living environments on the accuracy of these newer PAMs.