Utility of accelerometer thresholds for classifying sitting in office workers

To investigate the utility of a variety of Actical accelerometer count thresholds for determining sitting time in a sample of office workers.

Data were collected from 21 participants in Auckland, New Zealand, between December 2009 and January 2010. Participants wore a hip-mounted Actical accelerometer and thigh-mounted activPAL inclinometer (criterion) for a 48-h period. Raw inclinometer and accelerometer data for each 15s epoch of wear time were matched by date and time. Candidate accelerometer count thresholds for sitting classification were compared with the criterion measure using receiver operating characteristic analyses. Agreement in sitting time classification was determined using Bland-Altman methodology.

Significant differences in area under the curve (AUC) values by threshold criteria were found (p<0.001). A threshold of 0 counts provided the highest combined sensitivity and specificity (AUC 0.759, 95%CI 0.756, 0.761). The 95% limits of agreement for time spent sitting were wide, at 328min (range -30.8, 297.5).

A threshold of 0 counts/15s epoch with Actical accelerometers is likely to yield the most accurate quantification of sitting in office-based workers, however the wide limits of agreement found indicate limited utility of this threshold to accurately distinguish sitting time in office-based workers.