Assessing the validity and reliability and determining cut-points of the Actiwatch 2 in measuring physical activity

Abstract

Objective: The Actiwatch 2 (AW2) is a wrist-worn accelerometer typically used to measure sleep. Although it can measure physical activity, there is limited evidence supporting its validity. We assessed the validity and reliability of the AW2 to measure sedentary behavior and physical activity (light, moderate, vigorous intensities), and reported their respective count cut-points. Approach: Twenty-eight males and 22 females completed a task battery comprising three sedentary tasks and six randomized physical activity tasks at varying intensities, whilst wearing the AW2, a reference accelerometry device (Actigraph GT3X) and a cardiopulmonary gas analyzer on two separate occasions. Validity was assessed using correlations (AW2 counts versus GT3X counts and metabolic equivalent (MET) values), reliability using Bland–Altman analyses, and cut-points were determined using receiver operating characteristic (ROC) area under the curve (AUC) analyses. Main results: AW2 counts were positively correlated with GT3X counts (rho = 0.902, p < 0.001) and METs (rho = 0.900, p < 0.001). AW2-derived counts were comparable across independent assessment periods. Sedentary (AUC = 0.99, cut-point: 256 cpm) and vigorous activity (AUC = 0.95, cut-point: 720 cpm) were strongly characterized, and moderate activity (AUC = 0.66, cut-point: 418 cpm) was weakly characterized. Significance: The use of the AW2 in physical activity monitoring looks promising for sedentary behavior, moderate and vigorous activity, however, further validation is needed.