Quantifying Children's Self‐Paced Physical Activity: Rethinking Accelerometer Calibration

Abstract

Listen

Translate article

The purpose was to examine the ability of accelerometry (ACC) to estimate oxygen consumption (VO2) for self‐paced physical activity (PA). Children's (n=15; 9.3±1.2 yrs.) VO2 responses to paced treadmill (TM) activity and self‐paced PA using six games were determined using FITMATE. The vertical axis (V) and vector magnitude (VM) were used to quantify both paced and self‐paced PA with ACC (ActiGraph GT3X+ and expressed in counts/10sec). Linear regression and Bland‐Altman plots were used to compare VO2 for continuous paced (TM) and intermittent self‐paced (games) PA by assessing ACC vertical axis (ACC‐V), ACC vector magnitude (ACC‐VM) and the relative contribution of each axes (using ANOVA; p=0.05). Paced (TM) PA showed positive relationships (r) for VO2 plus ACC‐V and for VO2 plus ACC‐VM of 0.90±0.03 and 0.89±0.05, respectively (p>0.05). Results for measured VO2 showed higher VO2 for self‐paced vs. paced (TM) PA between 100‐1000 cnts/10sec (ACC‐VM) (i.e., VO2 at 300 cnts/10sec were 22 vs. 12 mLO2•kg‐1•min‐1, respectively (p<0.05). Thus, VO2 estimates from TM‐derived equations (both ACC‐VM and ACC‐V) for self‐paced PA were under‐estimated compared to measured VO2 for self‐paced PA (p<0.05); with low agreement (a dynamic bias especially as the intensity (>6METs) increased) as observed on Bland‐Altman plots. Comparing paced vs. self‐paced PA it was observed that the contribution of axis dominance to ACC‐VM for the two types of PA existed ‐ larger differences for paced (41±14%) and smaller differences for self‐paced (7±5%) (p<0.05). This study reveals that the poor estimates of VO2 for intermittent self‐paced PA using equations from continuous paced TM PA is attributed to the presence of a dominant axis.