The objectives of this study were to explore whether triaxial is more accurate than uniaxial accelerometry and whether shorter sampling periods (epochs) are more accurate than longer epochs. Physical activity data from uniaxial and triaxial (RT3) devices were collected in 1-s epochs from 31 preschool children (15 males, 16 females, 4.4 ± 0.8 yrs) who were videoed while they engaged in 1-hr of free-play. Video data were coded using the Children’s Activity Rating Scale (CARS). A significant difference (p < .001) in the number of minutes classified as moderate to vigorous physical activity (MVPA) was found between the RT3 and the CARS (p < .002) using the cut point of relaxed walk. No significant difference was found between the GT1M and the CARS or between the RT3 and the CARS using the cut point for light jog. Shorter epochs resulted in significantly greater overestimation of MVPA, with the bias increasing from 0.7 mins at 15-s to 3.2 mins at 60-s epochs for the GT1M and 0 mins to 1.7 mins for the RT3. Results suggest that there was no advantage of a triaxial accelerometer over a uniaxial model. Shorter epochs result in significantly higher number of minutes of MVPA with smaller bias relative to direct observation.
Hislop, Bulley, and Mercer are with the School of Health Sciences, Queen Margaret University, Edinburgh, Scotland, UK. Reilly is with the Physical Activity for Health Group, University of Strathclyde, Glasgow, Scotland, UK.