A Novel Approach To True Free-living Validation Of Accelerometer-measured Movement Behaviors

Accelerometer validation studies often include small, non-representative samples who complete simulated free-living activities in a laboratory. This setting may not reflect how people behave or move in real life. PURPOSE: To pilot a new approach to capturing true free-living accelerometer validation data, and to compare to activities performed in a simulated free-living laboratory setting. METHODS: Participants (N=61; age=21.3±4.1 y) wore a triaxial accelerometer at each wrist and the right hip during a 1.5-hr simulated free-living laboratory visit. Participants completed ~10 activities that were representative of their daily activities. Next, participants wore the accelerometers for 24-hr and reported start and stop times of activities via text message. A mixed effects model was used to compare the mean and standard deviation (SD) of Euclidian Norm Minus One (ENMO) for each activity between lab and free-living (fixed effects) with participant as a random effect. RESULTS: Mean ENMO was significantly different (p<0.05) between laboratory and free-living settings for stairs, brisk walk, sports, and run, but not for all wear locations. SD ENMO differed for clean, cook, eat, laundry, resistance exercise, run, sit, walk, and brisk walk for most wear locations. Computer, cycle, lying, self-care, television, video games, and yoga did not differ for any outcome or wear location. Activities like child care, driving, hiking, shopping, and vehicle maintenance could not be replicated in the laboratory. CONCLUSIONS: While overall activity intensity (mean ENMO) was similar for most activities, there was more variability (SD ENMO) during free-living, supporting that individuals behave differently in free-living compared to the laboratory setting. Furthermore, the proposed method of collecting true free-living data allowed for a larger variety of activities to be included than what is practical in a laboratory (e.g., flying, rock climbing).