Validity, reliability, and calibration of the Tritrac accelerometer as a measure of physical activity

Purpose: The purposes of this study were to assess the validity and reliability of the Tritrac R3D accelerometer during treadmill walking and running and then to calibrate the instrument. Methods: The Tritrac was assessed on 60 young adults (23.4 +/- 2.9 yr) during treadmill walking and running at 3.2, 6.4, and 9.7 km.h(-1). The calibration was carried out by identifying ranges of Tritrac raw data (vector magnitude) values corresponding to light (2-3.9 MET), moderate (4-7 MET), and vigorous (>7 MET) physical activity. Energy expenditure (EE), measured by indirect calorimetry, served as the criterion measure. Results: Interinstrument intraclass reliability coefficients for Tritracs worn on the right and left hip ranged from 0.73-0.87, while intersession coefficients demonstrated high reliability for all speeds (R = 0.87-0.92). Paired t-tests comparing mean accelerometer counts at 6.4 km.h(-1), 0% grade (2647 +/- 456), and 6.4 km.h(-1), 5% grade (2635 +/- 435) demonstrated no significant difference (P > 0.05). Mean differences between EE measured by indirect calorimetry and that estimated by the Tritrac ranged from 0.0082 kcal.kg(-1).min(-1) at 3.2 km.h(-1) to 0.0320 kcal.kg(-1).min(-1) at 9.7 km.h(-1), with the Tritrac consistently overestimating EE during horizontal treadmill walking. The relationship between vector magnitude and EE across all speeds was highly linear (R-2 = 0.90, SEE = 0.014 kcal.kg(-1).min(-1)), with little overlap between light, moderate, and vigorous categories. The mean vector magnitudes at 2, 4, and 7 MET were 650, 1772, and 3455, respectively. Conclusions: These data indicate that the Tritrac is highly reliable from day to day and is sensitive to changes in speed but not grade. Furthermore, the Tritrac accurately distinguishes various intensities of walking and jogging on level ground. With limitations, these cut-points can be used to categorize light, moderate, and vigorous physical activity and to estimate EE.