Reliability Of Accelerometry-based Activity Monitor Output During Treadmill Walking

Abstract

While reliability is a necessary prerequisite to measurement validity, reliability is rarely assessed for accelerometry-based activity monitors (AC) prior to use by researchers. PURPOSE: The study evaluated the test-retest reliability of an AC worn at different positions on the body during controlled walking trials on a treadmill. METHODS: Twenty subjects (10 men, Mean±SD: 23±3 yrs, 181.7±8.4 cm, 77.8±14.0 kg, 23.5±3.9 kg/m2; 10 women: 22±3 yrs, 167.2±5.7 cm, 64.6±13.8 kg, 23.0±4.2 kg/m2) walked at six speeds (40.2, 53.6, 67.0, 80.4, 93.8, 107.2 m/min) on a treadmill. The AC was worn on the left and right hip just anterior to the iliac crest and in line with the mid-axillary line of the thigh. AC monitors were also worn on the lateral aspect of the left and right ankles just proximal to the medial malleolus. Subjects walked three trials of 1 min at each treadmill speed while the order of speeds tested was counterbalanced across subjects. Using the raw AC output (counts/min) for each of the three trials and four wearing locations, reliability measures included the standard error of measurement (SEM), the coefficient of variation (CV, %), and the intraclass correlation (ICC). Standard formulae were used to compute the SEM and CV, while the ICC for AC output at each walking speed and wearing location were determined using a 2-factor RMANOVA. RESULTS: ICC values across all treadmill speeds and AC monitor locations were high when evaluated across three trials (0.97-0.99) and remained high when extrapolated to a single trial (0.91-0.99). Both SEM (0.9-2.3 counts/min) and CV (1.7-6.3%) values were consistently low for the ankle monitors across all walking speeds, though CV values tended to be highest (4.7-6.3%) at the slowest speed. SEM values were similarly low (1.0-2.7 counts/min) for the hip monitors, but CV values ranged from 1.8% to 14.0% with the highest values observed at the two lowest speeds. CONCLUSIONS: Reliability was acceptable for output from both the ankle and hip worn AC monitors during the five fastest walking trials. Indicators of low reliability (high CV) were present for the slowest walking trial, especially for the monitors worn on the hip. These results indicate that the AC monitor can be considered a reliable tool for assessing physical activity.