Participants wore the Actigraph GT3X-BT or GT9X-Link accelerometers on both wrists for the three cohorts, with methods described previously (32 (link)). Briefly, tri-axial acceleration data are sampled at 30 Hz for 24 or more hours continuously. Once the accelerometers were returned to the lab, data were uploaded, visually inspected, and processed using Actilife 6 (Actigraph Corp., Pensacola, FL) proprietary software. For most variables, data were band-pass filtered (0.25 and 2.5 Hz) and down sampled into 1-s epochs with ActiLife proprietary software, where each second is the sum of the 30 Hz values in that second and converted to activity counts (1 count = 0.001664 g). For a few variables (see Table 1 ), calculations were done directly on the 30 Hz data (6 (link), 24 (link)–26 ). Similar to previous work (7 (link), 12 (link), 19 (link), 21 (link), 43 (link)), accelerometry data was processed using custom written software in MATLAB (Mathworks, Inc., Natick, MA) to calculate UL performance variables which qualify various aspects of UL activity in everyday life. Table 1 presents the 12 UL performance variables included in the analysis along with their description and the source of accelerometer data for calculation (1 vs. 30 Hz). The variables independently measure duration, magnitude, variability, symmetry and quality of movement of one or both ULs.
Free full text: Click here
