Physical Activity Assessment Using Different Accelerometer Epoch Lengths During A Low-intensity Ballroom Dance Class

Abstract

Until recently accelerometry data has been commonly collected using 1-minute sampling intervals (epochs). However, 1-minute epochs may misclassify bouts of (in)activity when compared to more intermittent assessment (e.g. 2-second epochs). PURPOSE: To evaluate variability in physical activity intensity levels using different epoch lengths during a low intensity community dance programme. METHOD: Subjects were 66, mostly senior, adults with a self-assessed status of being physically inactive, who participated in a 2-hour beginners ballroom dance class. During class, physical activity was measured using an Actigraph GT1M accelerometer, set at 2-second epochs. Later, activity counts were transformed into 4, 6, 10, 20, 30 and 60-second epochs. Time spent in sedentary, light and moderate-to-vigorous (MTV) activity was calculated, using adjusted intensity thresholds for each epoch based on counts per minute values of <200, 200-1952 and >1952, respectively. RESULTS: Using 2-second epochs, participants spent 142 ± 9 mins, 32 ± 7 mins and 6 ± 3 mins of class time in sedentary, light and MTV physical activity, respectively. However, sedentary activity based on 4, 6, 10, 20, 30 and 60-second epochs underestimated totals by 5, 7, 10, 14, 17 and 20 mins, respectively. In contrast, time spent in light activity was overestimated by 5, 7, 10, 14, 18 and 22 mins when applying the same epoch periods. Considering MTV, time spent was underestimated by <1 min using 4, 6 and 10-second epochs, and by 1-2 mins using 20, 30 and 60-second epochs. All differences were statistically significant (p<0.01). CONCLUSION: In this case, where a large proportion of time was spent in sedentary and low intensity activity, it is unlikely that the differences obtained for time spent in MTV activity using varying epoch lengths are of biological significance. However, during low intensity activity evaluation, more sensitive (short duration) epochs may provide more accurate estimations. Further research is required to investigate the effect of different epoch lengths when assessing activity changes over time.