Validation of Uniaxial and Triaxial Accelerometers for the Assessment of Physical Activity in Preschool Children

in Journal of Physical Activity and Health
Restricted access

Purchase article

USD $24.95

Student 1 year subscription

USD $115.00

1 year subscription

USD $153.00

Student 2 year subscription

USD $218.00

2 year subscription

USD $285.00

Purpose:

Given the unique physical activity (PA) patterns of preschoolers, wearable electronic devices for quantitative assessment of physical activity require validation in this population. Study objective was to validate uniaxial and triaxial accelerometers in preschoolers.

Methods:

Room calorimetry was performed over 3 hours in 64 preschoolers, wearing Actical, Actiheart, and RT3 accelerometers during play, slow, moderate, and fast translocation. Based on activity energy expenditure (AEE) and accelerometer counts, optimal thresholds for PA levels were determined by piecewise linear regression and discrimination boundary analysis.

Results:

Established HR cutoffs in preschoolers for sedentary/light, light/moderate and moderate/vigorous levels were used to define AEE (0.015, 0.054, 0.076 kcal·kg−1·min−1) and PA ratio (PAR; 1.6, 2.9, 3.6) thresholds, and accelerometer thresholds. True positive predictive rates were 77%, 75%, and 76% for sedentary; 63%, 61%, and 65% for light; 34%, 52%, and 49% for moderate; 46%, 46%, and 49% for vigorous levels. Due to low positive predictive rates, we combined moderate and vigorous PA. Classification accuracy was improved overall and for the combined moderate-to-vigorous PA level (69%, 82%, 79%) for Actical, Actiheart, and RT3, respectively.

Conclusion:

Uniaxial and triaxial accelerometers are acceptable devices with similar classification accuracy for sedentary, light, and moderate-to-vigorous levels of PA in preschoolers.

Adolph, Puyau, Vohra, Nicklas, and Butte are with the Dept of Pediatrics, USDA/ARS Children’s Nutrition Research Center, Baylor College of Medicine, Houston, TX. Zakeri is with the Dept of Epidemiology and Biostatistics, Drexel University, Philadelphia, PA.