Introduction:
Both monitoring and promoting PA in children, two important areas in epidemiologic research, require the use of accurate and feasible measurement tools. The Fitbit activity monitor is one of the most widely commercialized, consumer-based activity monitors and its validity and reliability has been determined in adults; however, little research has determined the validity of Fitbit activity monitor in measuring sedentary behavior and PA in preschool-age children.
Hypothesis:
We hypothesized that, when comparing to the direct observation (DO), the Fitbit Flex (FF) activity monitor would accurately measure the time spent in sedentary, moderate-to-vigorous PA (MVPA), and Total (light-to-vigorous) PA in preschool-age children.
Methods:
A total of 28 preschool age children (Female: 46%, Age: 4.6 ± 1.0 yrs, BMI: 16.4 ± 1.5 kg·m2) wore the FF (on the wrist) and were directly observed while performing a set of unstructured and structured activities with varying intensity levels such as watching TV, drawing, playing with toys, playing soccer, and shooting basket. Breath-by-breath oxygen consumption were also measured using Oxycon Mobile indirect calorimetry (OM). Data from the FF (PA counts) and the OM (VO2 ml·kg·min-1) were summarized in terms of minute-by-minute basis. The validity of FF for measuring the time spent in sedentary behavior, MVPA, and Total PA against DO was examined using three ways: (1) Pearson correlation examining the relatively agreement between FF and DO; (2) Mean absolute percent error (MAPE) examining the measurement errors of FF; and (3) One-sided equivalence test examining the absolute agreement between FF and DO by comparing the 90% confidence intervals (CI) of the estimates from the FF with the respective equivalence zone (EZ; ± 10% of the mean estimates from the DO).
Results:
On average, oxygen consumptions for sedentary, light, MVPA were 9.0 ± 1.6, 14.9 ± 3.9, 23.5 ± 5.5, and 33.8 ± 5.6 ml·kg·min-1, respectively. Correlations between FF and DO were consistently high for sedentary (r = 0.81, P <.001), MVPA (r = 0.62, P <.001), and Total PA (r = 0.81, P <.001). MAPEs were 8.2%, 21.1%, and 8.2% for sedentary, MVPA, and Total PA, respectively. The estimates from the FF were not significantly equivalent to those from the DO; sedentary (FF: Mean (M) = 7.8 min, 90% CI: 7.3 - 8.3 min vs. DO: M = 6 min, EZ: 5.4 - 6.6 min), MVPA (FF: M = 7.2 min, 90% CI: 6.2 - 8.2 min vs. DO: M = 12 min, EZ: 10.8 - 13.2 min), and Total PA (FF: M = 14.0 min, 90% CI: 13.6 - 14.4 min vs. DO: M = 16 min, EZ: 14.4 - 17.6 min).
Conclusions:
The validity of the FF was supported by the consistently high correlations with the criterion measure. However, inconsistent results from equivalence testing warrants subsequent assessment of the validity of the FF as it has a high potential for replacing research-based PA monitors in preschool-age children.