No abstract

Objective: To assess (1) step count accuracy of the Fitbit Zip, compared to manual step count, in people receiving outpatient rehabilitation, in indoor and outdoor conditions, and (2) impact of slow walking speed on Fitbit accuracy. Design: Observational study. Setting: A metropolitan rehabilitation hospital. Subjects: Adults ( n = 88) attending a subacute rehabilitation outpatient clinic with walking speeds of between 0.4 and 1.0 m/s. Interventions: Two 2-minute walk tests, one indoors and one outdoors, completed in random order. Main measures: Step count recorded manually by observation and by a Fitbit Zip, attached to the shoe on the dominant or non-affected side. Subgroup analysis included assessment accuracy for those considered limited community walkers (slower than 0.8 m/s) and those considered community walkers (faster than 0.8 m/s). Results: The Fitbit significantly ( P < 0.05) undercounted steps compared to manual step count, indoors and outdoors, with percentage agreement slightly higher outdoors (mean 92.4%) than indoors (90.1%). Overall, there was excellent consistent agreement between the Fitbit and manual step count for both indoor (ICC 0.83) and outdoor (ICC 0.88) walks. The accuracy of the Fitbit was significantly ( P < 0.05) reduced in those who walked slower than 0.8 m/s outdoors (ICC 0.80) compared to those who walk faster than 0.8 m/s (ICC 0.90). Conclusions: The Fitbit Zip shows high step count accuracy with manual step count in a mixed subacute rehabilitation population. However, accuracy is affected by walking speed, with decreased accuracy in limited community walkers.

Grading reliability improves statistically with some experience, although perhaps not to a clinically meaningful extent. No added benefit can be derived from supplemental training.

No abstract