Purpose: To examine the convergent validity of the ActiGraph and activPAL accelerometers with the Bouchard Activity Record (BAR) in adults. Sedentary behavior and walking were evaluated in all instruments; standing and moderate-to-vigorous physical activity (MVPA) was evaluated only in those that detected such variables.
Methods: Thirty-two participants wore the accelerometers and completed the BAR concurrently for 1 d. Descriptive statistics and delta values were reported for all instruments. Summary time spent in sedentary behavior and walking was compared between all instruments using repeated-measures ANOVA. Dependent t-tests were used to analyze summary time in 1) standing between activPAL and BAR and 2) MVPA between ActiGraph and BAR. Bland-Altman plots were interpreted for systematic bias. On a detailed level, concurrent time interval data were compared using mean percent agreement and κ statistics.
Results: There was a significant difference found in summary time spent in sedentary behavior apparent between ActiGraph and activPAL as well as between ActiGraph and BAR. There was also a significant difference detected in time spent in walking, apparent between ActiGraph and activPAL, and between ActiGraph and BAR. In the time interval analysis, mean percent agreement ranged from 54.0% (for walking detected by ActiGraph and activPAL) to 86.7% (for MVPA by ActiGraph and BAR). κ values ranged from 0.25 (for walking by ActiGraph and activPAL) to 0.70 (for sedentary behavior between activPAL and BAR). Differences were also found in standing and MVPA.
Conclusions: The activPAL and BAR showed convergence on both summary and concurrent time interval levels in both sedentary behavior and walking. The comparative discordance between activPAL and BAR with ActiGraph was likely a function of different approaches used to distinguish sedentary behavior from walking.