COMPARISON OF ACTIWATCH SENSORS AND CARS ON ASSESSMENT OF PHYSICAL ACTIVITY IN CHILDREN.

Abstract

1577 The Children's Activity Rating Scale (CARS) is a rating scale that is used in direct observation of physical activity in children. Direct observation is costly, tedious, and accuracy may decrease as the observation period lengthens. Motion sensors have gained acceptance for assessment of physical activity. Actiwatch (Mini-Mitter) is a motion sensor with an omni-directional accelerometer. The purpose of this study is to compare 6-hr of activity levels using simultaneous monitoring of pre-school aged children with CARS and Actiwatch motion sensors. A total of 41 children had 3-6 hours of direct observation during waking hours while wearing a sensor on the waist. Simultaneous 10-min mean CARS scores and 10-min sensor activity counts were matched for each subject. The mean (+/− SD) CARS for the 41 children over the 3 to 6 hour period was low (1.76+/−0.28), with a range of 1.20 to 2.23. The range for the mean 10-min CARS scores was 1.0 to 4.49. The 10-min sensor counts ranged from 0 to 2,752 with a mean of 312 (median 230). Overall, 10-min sensor readings correlated with 10-min CARS (r=0.60, P<0.001). The within child correlations between the 10-min CARS score and the 10-min sensor readings from −0.09 to 0.96 with a median of 0.71. We found that the correlation between 10-min sensor readings and the CARS was dependent on the child's activity level for the day (r=0.38, P=0.01): a higher correlation was observed in those children who were more active, probably due to the larger ranges in the CARS scores. Using mixed model repeated measures, sensor readings were significantly associated with CARS (P<0.001). The results indicate that the 10-min CARS score correlates with 10-min sensor counts, favoring the use of the sensors in assessing physical activity in preschool-aged children. Supported by NIH RO1 AR 45310