Effect of Epoch Length on Patterns of Stepping Cadence in Fourth-Grade School Children

Background: Children’s physical activity is often intermittent and sporadic. Shorter accelerometer epochs (e.g., 5 vs. 60 s) amplify children’s time spent at higher activity count-defined intensities; however, it remains unknown how accelerometer-determined estimates of time spent and steps accumulated within incremental cadence (steps/min) bands may be influenced when interpreted with shorter epochs.

Purpose: To evaluate the effect of epoch length on free-living estimates of time spent and steps accumulated within incremental cadence bands among fourth-grade school children.

Methods: A total of 491 children (41% boys; 9.9 ± 0.6 yrs) from 21 schools in Baton Rouge, Louisiana, participating in the International Study of Childhood Obesity, Lifestyle and the Environment, wore an ActiGraph GT3X+ accelerometer for 7 days. Step data were derived from raw files using the default filter and integrated to 5, 10, 15, 30, and 60 s epochs. Time spent and steps accumulated within incremental cadence bands (0, 1-19, 20-39, 40-59, 60-79, 80-99, 100-119, and 120+ steps/min) were computed for each of the 5 different epochs. Linear mixed-effects models were used to examine between-epoch differences in time spent and steps accumulated within each cadence band. Models accounted for repeated measures (epochs) and clustering of children within schools.

Results: Significant epoch effects were observed for time spent and steps accumulated within each cadence band (all p < 0.01). Compared to a 60 s epoch, changes in time spent within a given cadence band ranged from a 77% decrease (1-19 steps/min; 60 s = 442.4 min/day, 5 s = 99.9 min/day) to a 1,533% increase (120+ steps/min; 60 s = 0.6 min/day, 5 s = 9.8 min/day), while changes in steps accumulated within a given cadence band ranged from a 53% decrease (1-19 steps/min; 60 s = 2,551 steps/day, 5 s = 1,192 steps/day) to a 1,405% increase (120+ steps/min; 60 s = 86 steps/day, 5 s = 1,295 steps/day).

Conclusion: Estimates of time spent and steps accumulated within each cadence band can vary dramatically based upon epoch length. Further research is needed to identify which recording epochs optimize the measurement of behavior likely to have meaningful and translatable health-related impacts while also minimizing measurement error due to obscuration of movement and non-movement.

Supported by The Coca-Cola Company