Measuring ambulation with wrist-based and hip-based activity trackers for children with cerebral palsy

To assess the accuracy of consumer available wrist-based and hip-based activity trackers in quantitatively measuring ambulation in children with cerebral palsy (CP).

Thirty-nine children (23 males, 16 females; mean age [SD] 9y 7mo [3y 5mo]; range 4-15y) with CP were fitted with trackers both on their wrist and hip. Each participant stood for 3 minutes, ambulated in a hallway, and sat for 3 minutes. The number of steps and distance were recorded on trackers and compared to manually counted steps and distance. Pearson correlation coefficients were determined for the number of steps during ambulation from each tracker and a manual count. Mean absolute error (MAE) and range of errors were calculated for steps during ambulation for each tracker and a manual count and for distance for each tracker and hallway distance.

For the number of steps, a weak inverse relationship (r=-0.033) was found for the wrist-based tracker and a strong positive relationship (r=0.991) for the hip-based tracker. The MAE was 88 steps for the wrist-based and seven steps for the hip-based tracker. The MAE for distance was 0.06 miles for the wrist-based and 0.07 miles for the hip-based tracker.

Only the hip-based tracker provided an accurate step count; neither tracker was accurate for distance. Thus, ambulation of children with CP can be accurately quantified with readily available trackers.

Consumer available activity trackers accurately measure ambulation in children with cerebral palsy (CP). The hip-based tracker is more accurate than the wrist-based tracker for children with CP. The hip-based Fitbit activity tracker accurately measures step counts of children with CP during ambulation.