Free-Living Accelerometer Calibration: A Novel Direct Observation System

Abstract

PURPOSE: To develop a direct observation (DO) system to serve as a criterion measure for model calibration using free-living (FL) accelerometer data. METHODS: Ten participants (19.4±0.8 yrs) were video-recorded during four, 1-hr FL sessions in different settings: school, home, community and physical activity (PA). For each setting, 10-min clips from three, randomly selected sessions were extracted and coded by a criterion coder and assistant coders (ACs) using the Observer XT software (Noldus, Wageningen, the Netherlands). Coders identified the movement/body position and four modifiers: locomotion (yes or no), activity type (e.g. reading), MET value, and intensity category. For intra-rater agreement, the criterion coder coded all 12 videos twice, separated by at least one week. For inter-rater agreement, all 12 videos coded by each AC were compared to the criterion coder. Intraclass correlation coefficients (ICCs) were calculated to assess agreement of intensity category for intra- and inter-rater comparisons. RESULTS: Intra-rater agreement ranged from 91% to 100% across all variables in all four settings. Inter-rater agreement between the criterion coder and the ACs ranged from 88±3.5% to 100±0% across all variables in all four settings. As shown below, ICCs for intensity category ranged from 0.74-1.00 and 0.81-1.00 for intra- and inter-rater comparisons, respectively. CONCLUSION: The DO system is reliable and feasible to serve as a criterion measure of FL physical activity in young adults. The DO system can serve as a standardized instrument to develop accelerometer models for estimating PA and sedentary behavior in FL settings. Supported by: NIH NIDDK 1R01DK110148-01