The convergent validity of Actiwatch 2 and ActiGraph Link accelerometers in measuring total sleeping period, wake after sleep onset, and sleep efficiency in free-living condition

Physical activity (PA) and sleep are important to health; thus, it is important for researchers to have valid tools to measure them. Accelerometers have been proven valid for measuring PA and sleep, but only one device does this simultaneously: the ActiGraph Link (ActiGraph, LLC); however, the sleep-monitoring function has not been validated. This study aimed to evaluate the predictive power of ActiGraph Link sleep parameters against a validated accelerometer (Actiwatch 2, Phillips Respironics Mini-Mitter). A total of 49 Hong Kong adults aged 18-64 provided valid data on both accelerometers on their non-dominant wrist for seven consecutive days. Epochs from both accelerometers were classified as either sleep or awake using seven established algorithms (Cole-Kripke, Sadeh, Sazonov, high sensitivity threshold, medium sensitivity threshold, low sensitivity threshold, and neural network model), and these data were transformed to total sleeping period, wake after sleep onset, and sleep efficiency. The non-zero count data for both accelerometers (331,103 observations) were strongly correlated with a Spearman correlation of 0.83 (p < 0.001). The total sleeping period was highly correlated (Spearman correlation ranged from 0.74 to 0.90) regardless of the algorithms used. All algorithms yielded insignificant difference in total sleep time measured by the two accelerometers (p > 0.05) with a negligible effect size of d < 0.2. The agreement of sleep/wake status was high for all algorithms, with accuracy ranging from 93.05 % (Sadeh's algorithm) to 96.13 % (Cole-Kripke's algorithm). Results showed that the sleep function of the ActiGraph Link performs similar to a validated accelerometer (Actiwatch 2) and provides an opportunity to measure both sleep and PA simultaneously.