Physical Activity in Children using Accelerometers

Children’s physical activity level was measured objectively using a 7-day recording with a uniaxial accelerometer with pedometer function (GT1M Actigraph, Fort Walton, Florida) at three time-points: baseline, end-of- programme (3-months) and at 9-months. Recordings were not done at the two year follow-up in order to minimise respondent burden. A 7 day recording provides a reliable estimate of usual physical activity in children allowing for any differences between weekday and weekend (Trost et al 2005). The data collection interval was set to record both activity and pedometer counts at 60sec intervals. This time interval was chosen in order to allow the storage of a full week of data recording. Box 4.2 describes the other aspects for which the data collection was standardised.

Children (sometimes with help from parents) completed an activity diary alongside the accelerometer measurements, in order to enable interpretation of the accelerometer and pedometer data. The diary was adapted from an adult activity diary in use by Wilcock, Coventry University (personal communication) (Appendix XI). From the diary it was possible to determine if the child had participated in any activity which would not be picked up by a uniaxial (vertical) monitor, such as cycling.

For a day to be classified as a complete monitoring day and used in analysis, there has to be evidence of activity counts indicating that the accelerometer had been on for most of the day. To illustrate, on occasions where subjects forgot or chose not to put the monitor on to go to school, but put it on after school, these days were excluded. However, days (often weekends) where a subject had slept in until late (verified by the diary) and in which the monitor was put on around lunchtime, were included in the analysis.

Box 4.2 – Standardisation of the Data Collection using Accelerometers

 Same monitor (i.e. serial number) used for a child at each time- point (to remove ‘between-unit’ variation in accelerometer output)

 Worn on right hip, on an elastic belt around the waist

 7 consecutivedays, to include 2 weekend days

 Worn from waking to bedtime

 Attempt made to standardise for school or holiday period for a child at each time-point

 Children wore the monitor when it was given to them to allow habituation; data collection started the following day.

For a record to be included in the ‘before-and-after’ analysis, a minimum of 4 out of the 7 days of monitoring had to be available, because 4 is the minimum number of days needed to obtain a reliable measurement of habitual physical activity in children (reliability of 0.80) (Trost et al 2005).

At each time-point in the data collection two measures were calculated:- mean daily pedometer counts; and mean daily time spent in moderate and vigorous physical activity (MVPA). UK guidelines recommend an hour of at least moderate exercise per day for children and young people (Department of Health 2004b), though studies vary widely with regards to the proportion of children meeting this standard (Trost et al 2006, Riddoch et al 2007). It has recently been recognised that the differences in minutes of moderate and vigorous physical activity (MVPA) for children between studies is likely to be due to different cut-off points of accelerometer counts used to define MVPA (Riddoch et al 2007). I have therefore analysed the accelerometer recordings using two different methods to calculate MVPA in use for the Actigraph:

(i) Freedson Equation

The number of minutes per day undertaking moderate to vigorous physical activity (MVPA) was calculated by:-

1. Translating activity counts into METs using the Freedson equation for children (Freedson et al 2005, Trost et al 2006):-

METs = 2.757 + (0.0015 x counts.min-1) – (0.08957 x age [yr]) – (0.000038 x counts.min-1x age [yr])

2. Calculating the number of minutes each day at 4 METS or above (four times the standard resting metabolic rate) used as a cut-off point for MVPA. Then a

daily average time for MVPA was obtained for the record. (4 METS was used to define MVPA, rather than 3 METS which leads to over-estimation in children - personal communication with S Trost).

Using the Freedson equation the cut-off point of 4 METS equated to a mean activity count of 1834 counts.min-1 for the children at baseline. However, this ranged from 1510 counts.min-1 for the youngest child (7yrs) to 2515 counts.min-1 for the oldest child (13.7yrs), because the equation takes into account the age of the child.

(ii) Puyau

Puyau et al (2002) has more simply defined MVPA as an activity count of 3200 counts.min-1 or above in children, not taking into account age, derived from a calibration study in 6 to 14 year olds. The number of minutes each day at an activity count of 3200 counts.min-1 or above was determined and then a daily average time undertaking MVPA was calculated for the record.

Thus, the activity count which defines MVPA is much lower using the Freedson equation (mean: 1834 counts.min-1, at baseline) compared with the 3200 counts.min-1 recommended as the threshold by Puyau. Reilly et al (2008) has recently suggested that from current evidence the most appropriate cut-off point for MVPA in children is from 3000 to 3600 counts.min-1. This supports the use of the cut-off of 3200 counts.min-1 from Puyau et al (2002) rather than the Freedson (2005) equation. Both methods of calculating MVPA will however be

4.9.3.4 Further Assessment of Validity of Eating & Activity Measurements