Intraindividual Variation of Objectively Measured Physical Activity in Children

MATTOCKS, CALUM1; LEARY, SAM1; NESS, ANDY4; DEERE, KEVIN1; SAUNDERS, JOANNE1; KIRKBY, JOANNE2; BLAIR, STEVEN N.3; TILLING, KATE1; RIDDOCH, CHRIS5

Medicine & Science in Sports & Exercise:
doi: 10.1249/mss.0b013e318030631b
BASIC SCIENCES: Epidemiology
Abstract

Purpose: This study examined the seasonal and intraindividual variation in objectively measured physical activity in 11- to 12-yr-olds.

Methods: Children were asked to wear a uniaxial accelerometer for 7 d four times throughout the course of about a year. A random-intercepts model was used to separate the inter- and intraindividual components of physical activity. Gender, age, body mass index (BMI), height, and month of measurement were fitted to the model as potential confounders.

Results: A total of 315 children had valid data for at least two measurement occasions, and 244 had data for all four measurement occasions. The unadjusted intraclass correlation coefficient (ICC) for total activity (counts per minute) was 0.54; 0.49 after adjusting for gender, age, and BMI; and 0.53 after adjusting for gender, age, BMI, and month. Further adjustment for pubertal status at baseline had no effect on the ICC. Restricting the analysis to only those with data for all four measurement occasions (N = 244), or to measurements taken on schooldays only, had no effect on the ICC. The fully adjusted ICC was 0.51 for weekdays only and 0.39 for weekend days only. For minutes of moderate to vigorous physical activity, minutes of vigorous activity, minutes of sedentary behavior, and number of 30-min blocks of sedentary behavior, the fully adjusted ICC were 0.45, 0.37, 0.59, and 0.39, respectively. The analysis was repeated for boys and girls separately, but the differences in ICC were small.

Conclusion: There was substantial intraindividual variation in the objectively measured physical activity of these children. Studies using single a measurement occasion where physical activity is the exposure should take this into account to adjust for regression dilution.

Author Information

1Department of Social Medicine, University of Bristol, UNITED KINGDOM; 2Institute of Diet, Exercise and Lifestyle University of Glasgow, Glasgow, UNITED KINGDOM; 3University of South Carolina, Columbia, SC; 4Department of Oral and Dental Science, University of Bristol, UNITED KINGDOM; 5Department of Sport and Exercise Science, School for Health, University of Bath, UNITED KINGDOM

Address for correspondence: Calum Mattocks, Department of Social Medicine, University of Bristol, 24 Tyndall Avenue, Bristol, BS8 1TQ United Kingdom; E-mail: c.mattocks@bristol.ac.uk.

Submitted for publication June 2006.

Accepted for publication November 2006.

©2007The American College of Sports Medicine