Accelerometers are widely used to quantify physical activity (PA) in preschoolers, yet no ‘best practice’ method for data treatment exists. The purpose of this study was to develop a robust method for data reduction using contemporary statistical methods and apply it to preschoolers’ accelerometer data.
Children age 2 to 5 years were recruited in Auckland, New Zealand, and asked to wear accelerometers over 7 days. Average daily PA rates per second were derived for participants, estimated using negative binomial generalized estimating equation (GEE) models. Overall participant rates were derived and compared using normal GEE models. Descriptive information for data analyzed were compared with that derived using traditional data inclusion approaches.
Data were gathered from 78 of the 93 enrolled children over a median of 7 days. Daily PA rates ranged from 1.27 to 17.64 counts per second (median 5.70). Compared with traditional approaches, this method had many advantages, including improved data retention, the computation of a continuous measure, and facilitating powerful multivariable regression analyses, while providing similar descriptive information to existing methods.
PA rates were successfully calculated for preschoolers’ activity description and advantages of the approach identified. This method holds promise for future use and merits further application and enhancement.
Oliver and Schofeld are with the Centre for Physical Activity and Nutrition Research, Auckland University of Technology, Auckland, New Zealand. Schluter is with the School of Public Health and Psychosocial Studies, Auckland University of Technology, Auckland, New Zealand.