MEASURING POSTURE USING THIGH-AND POCKET-WORN ACCELEROMETERS IN A LABORATORY SETTING

It is important to have accurate measurements to better understand the health impacts of time spent in different postures/activities. Inclinometer-based accelerometers, most often worn on the front of the thigh, can classify postures into sitting, standing, walking, and cycling. PURPOSE: To assess the validity of a new monitor (FIB) and an established monitor (AP) for classifying postures in a laboratory setting, compared to direct observation. METHODS: 25 participants (12 male; age 40 ± 15 yr) wore one AP (left or right thigh), and four FIB (both thighs and front pants/shorts pockets) while performing 12 activities of daily living for 2 minutes each in a laboratory setting. Accuracy of postural classification (sit, stand, step, or cycle) from all devices was compared to direct observation using percent agreement and unweighted kappa. Posture classification was adjusted for certain activities: for sweeping, walking or standing was counted as a correct posture; for rowing, cycling or walking was counted as a correct posture. RESULTS: Average accuracy for the FIB thigh monitors was 86% (kappa = 0.8; p < 0.001) with high accuracy (90%-100%) for detecting lying, sitting, walking, stair climbing, jogging, and sweeping, but lower accuracy for cycling (84%), standing (82%), rowing (79%), and reclined sitting (75%). The average accuracy for the FIB pocket monitors was 75% (kappa 0.65; <0.001) with high accuracy (88%-100%) for detecting lying, sitting, walking, and stair climbing, but lower accuracy for jogging (86%), sweeping (78%), cycling (71%), standing (71%), reclined sitting (57%), and rowing (41%). The thigh-worn AP had an overall accuracy of 73.0% (kappa = 0.61; p < 0.001) with high accuracy (88%-100%) for detecting lying, sitting, standing, walking, jogging, and sweeping, but lower accuracy for stair climbing (84%), cycling (72%), rowing (12%), and reclined sitting (0%). AP classified reclined sitting as 100% standing, and rowing as 86% sitting. CONCLUSIONS: The FIB (thigh and pocket) and AP monitors were accurate for many postures, but differences were observed for certain activities (e.g., standing, reclined sitting, rowing). The FIB monitor shows promise as a tool for measuring sedentary and non-sedentary waking postures.