FREE-LIVING COMPARABILITY OF THIGH-AND POCKET-WORN ACTIVITY MONITORS

Accurate assessment of free-living postures and activities is important for assessing health and wellbeing. The thigh-worn (AP) and thigh- or pocket-worn (FIB) inclinometer-based accelerometers can classify posture into categories (sit, stand, step, cycle). PURPOSE: To assess the reliability of the FIB thigh- and pocket-monitors for classifying posture in a free-living setting and compare results to the AP monitor. METHODS: 25 participants (12 male; age 40 ± 15 yr) wore 2 FIB on each the left (L) and right (R) side thigh and pocket, and one AP on the L or R thigh for three free-living days during waking hours. Minute-by-minute sitting, standing, walking, and cycling outputs from the L and R FIB monitors (thigh and pocket) were compared to each other using correlations, percent agreement and kappa. Paired-samples t-tests compared L and R FIB monitors for daily time spent in activity type (sitting, standing, slow & fast walk, cycling, high-intensity), and activity intensity (light, moderate, vigorous). Repeated measures ANOVA compared daily sitting, standing, stepping, and cycling among FIB and AP monitors. RESULTS: Daily L and R (thigh and pocket) output variables were highly correlated; L and R thigh (r = 0.89-0.99) and pocket (r = 0.73-0.98) except for thigh: light intensity (r = 0.57) and cycling (r = 0.68) and pocket: standing (r = 0.66) and cycling (r = 0.04). There were no significant differences between FIB L and R pocket outcomes, and only 2 outcomes (stepping, and moderate intensity) were significantly higher on FIB R thigh than L thigh. When comparing all five monitors, sitting was significantly lower for FIB pocket (506 ± 131 min/day, <0.05) than AP and FIB thigh (551 ± 135 min/day). Standing was significantly lower for FIB thigh (180 ± 80 min/day, <0.05) than AP and FIB pocket (216 ± 99 min/day). Stepping was significantly more for FIB R thigh (160 ± 77 min/day, <0.05) than FIB pocket (150 ± 74 min/day), and AP (132 ± 67 min/day) was significantly lower than all. There were no differences in cycling. Across 68,141 total minutes, FIB L and R thigh had 85.6% agreement (kappa = 0.73), and L and R pocket had 81.8% agreement (kappa = 0.58; both <0.001). CONCLUSION: FIB thigh, pocket, and AP estimates of sitting, standing and stepping were dissimilar. FIB thigh and pocket monitors had similar and high inter-monitor reliability.