Accuracy Of The Geneactiv And Actigraph To Classify Posture-based Sedentary Behavior Estimates In Free-living Settings

Introduction. The most common method to score accelerometer (ACC) data for physical activity and sedentary behaviors (SB) measurement is the use of cut-points, this method is satisfactory for locomotion, but poses several limitations in classifying time spent in SB. As cut-points rely on the magnitude of acceleration, classifications for time spent in sitting/standing is similar as there is no movement. Hence, misclassification can occur regardless of postural differences. The sedentary sphere (SS) method (Rowlands 2013, 2016) allows to use an inclinometer feature present in commonly used tri-axial accelerometers that provides the possibility of adding posture allocations to the cut-points approach when assessing SB. 
Purpose. To test the accuracy of posture-based sedentary time estimates made by the SS method using data obtained from the GENEActiv (GA) and the ActiGraph GT3x+ (AG) accelerometers during free-living conditions. 
Materials and Methods. 20 healthy adults participated in the study, 50% female, 30.25±6.43 years of age, and body mass index=22.7±3.1 kg/m2. Participants wore four monitors simultaneously, two on each wrist (GA + AG). The criterion SB variable was obtained from direct observation with focal sampling and duration coding during free-living conditions. Several analyses were used to assess the accuracy of the SS, namely: Equivalency testing, Percent error (PE), Bland-Altman plots (BA), Kappa statistic (K), sensitivity (S), and specificity (SP). The accuracy of two alternative configurations of the SS were also tested (SS and SS ). a b 
Results. A total of 40 sessions and 241.32 hours of free-living direct observation were observed. None of the estimates (SS, SS ) were equivalent to the criterion regardless the ACC and location. SS a , SS b showed higher PE for the non-dominant wrist than dominant wrist regardless the ACC. SS results for K, S, and SP were similar with fair agreement for K (0.26-0.30) and moderate S (0.61-0.65) and SP (0.65-0.72). BA showed no trends in the error distribution, and negative slope in the regression lines regardless of the configuration (SS, SS ) location or ACC; however, the overall range between the a , SS b 95% limits of agreement were considerably wider (124 to -206). As for SSa, the AG non-dominant showed the lowest PE (-0.49), with fair agreement for K (0.31) and moderate S (0.63) and SP (0.68). As for SSa, the GA non-dominant showed lower PE (-1.09), however, K, S, and SP were lower as compared to SS. 
Conclusion. The SS is an accurate method to classify sedentary time in free-living settings from wrist-worn GA and AG. For the GA, the accuracy was highest when the ACC was worn on the dominant wrist using the original SS configuration. For the AG, the accuracy was highest when the ACC was worn on the non-dominant wrist using an alternative SS configuration (SSa). 

Acessar Arquivo