Validity Of Accelerometer-derived Estimates Of Sedentary Behaviour

Accelerometers are gaining popularity as an objective measure of sedentary behaviour. Limited evidence exists however on the validity of different cut points used to define sedentary time, or on the inclinometer function seen in newer models of the ActiGraph accelerometer.

Purpose:
To determine the concurrent validity of the ActiGraph GT3X+ inclinometer and different counts per minute (cpm) cut points for detecting sedentary behaviour in free-living adults.

Methods:
52 participants (54% male, age = 26.3 ± 6.2 years) wore an activPAL3 inclinometer (the criterion) and an ActiGraph GT3X+ accelerometer whilst continuing with their normal routine for 1 day. The activPAL3 was attached to the right thigh, whilst the ActiGraph was positioned on the right hip using an elastic belt. Both devices displayed data in 15-second epochs. Accelerometer-determined sedentary time was calculated using the inclinometer function of the ActiGraph, and by multiple cut points (<50, 100, 150 and 200 cpm) applied to activity counts derived from the vertical axis. Accelerometer-determined sedentary times were compared to sedentary time measured by the activPAL3 using a repeated measures ANOVA and Bland-Altman plots. Associations between the different measures were examined using Pearson correlation coefficients.

Results:
Time spent sedentary did not differ significantly between the ActiGraph inclinometer function and the activPAL3 (532 ± 95 vs 522 ± 125 mins, p=0.51). There was a significant moderate correlation between these measures (r = 0.52), however the limits of agreement were large (-207 – 227 mins). Sedentary time was significantly overestimated, relative to the activPAL3, when calculated using the four cut points (p<0.001). The <50 cpm cut point provided the closest estimate of sedentary time (582 ± 86 mins). This cut point had a stronger correlation (r = 0.60) and narrower limits of agreement (-137 – 258 mins) relative to the criterion, when compared to data derived from the ActiGraph inclinometer.

Conclusion:
The findings suggest that the inclinometer function of the ActiGraph GT3X+ provides a good overall estimate of sedentary time when compared to the activPAL3. The present findings also suggest that the <50 cpm cut point may provide a better estimate of sedentary behaviour in comparison to the widely used <100 cpm cut point.