HOW MANY DAYS OF MONITORING ARE REQUIRED FOR REPRESENTATIVE MEASUREMENTS OF PHYSICAL ACTIVITY IN CHILDREN?

Abstract

Accelerometry is a popular objective method to measure physical activity (PA) in children. Currently there is limited evidence on how long (number of days) and which days (weekdays/weekends) monitoring should be carried out to be confident of a representative PA measurement. Previous work has addressed some of these questions in certain populations of children (Trost et al, 2000). Further research is required to verify these findings in other populations of children. PURPOSE To determine the number of days of monitoring required to obtain a representative measurement of daily PA in young children and which days of the week should be included in this monitoring period. METHODS CSA 7164 uniaxial accelerometers were worn by 76 healthy children (40 male(M) and 36 female(F)), mean age 5.8 years(SD 0.5), for a minimum of 7 consecutive days. The daily total physical activity for each child was expressed as counts per minute (cpm) averaged over the entire daily monitoring period. RESULTS Logarithms of the daily cpm were used to reduce the effect of increasing within child day-to-day variability with increasing daily activity. Using Spearman-Brown Prophecy Formula, reliability co-efficients and 95% Confidence Intervals for 1, 4 and 7 days of monitoring were established as 0.35 (0.26,0.46), 0.69 (0.58,0.77) and 0.79 (0.71,0.86) respectively. Repeated Measures Analysis of Variance showed that the children were slightly and significantly more active on weekends (M mean = 900.2cpm, F mean = 833.0cpm) than weekdays (M mean = 858.2cpm, F mean = 747.2cpm) (p = 0.02). Boys were slightly and significantly more active than girls on both weekdays and weekends (p = 0.011). CONCLUSION Consistent with previous work, these results show that when using accelerometers, adoption of a monitoring period of around 7 days will ensure a representative measurement of habitual PA of young children. This monitoring period will allow for the effect of greater PA of 5-year-old children at weekends than weekdays. This work is funded by SPARKS 0GLW/O1