A convenience sample of children (N = 282) who were participating in the Finnish Health in Teens survey (Fin-HIT) was recruited during spring 2013. In total, 17 schools were contacted and of these schools, 12 schools participated in this sub-study.
After schools’ willingness to participate, the parent and the child gave their informed consent to participate in this sub-study. The introductions of using accelerometer and completing diary for the participating children were given in a school lesson and in written form. The study was approved by the Coordinating Ethics Committee of the Helsinki and Uusimaa Hospital District.
The participating children completed an out-of-school PA questionnaire twice (approximately 30 days apart). Two out-of-school PA questions were asked. The out-of-school PA was defined as moderate-to-vigorous PA (MVPA) that child was doing alone, in sport clubs, and with family or friends. PA in schools and during school trips were not asked to take into account. In the first question, children were asked to evaluate how many hours per week they were physically active in out-of-school-hours (hours/per week). 10 response options were possible, ranging from ‘one hour or less per week’ to ‘ten hours or more per week’. For the analyses, two types of variables were formed: a) the weekly MVPA was transformed into minutes and divided by seven to generate the average daily activity time (min/day) called as MVPA Duration (continuous variable), b) the average daily activity time was divided into the quartiles called as the quartiles of MVPA duration.
Secondly, children were asked to evaluate how many times they were physically active in their weekly out-of-school-hours (times per week). 10 response options were possible, ranging from ‘I’m not physically active at all’ to ‘seven times or more per week’. For the analyses, two types of variables were formed: a) the continuous MVPA frequency based on the original answers, and b) the categorized MVPA frequency, which was recoded from the original variable so that answer options from ‘never’ to ‘1-3 times per month’ was coded to 0.5. The other answer options were coded from 1 (‘one time per week’) to 7 (‘seven time or more per week’).
Accelerometer data management
MVPA was assessed by the Actigraph GTX3 (LLC, Florida, USA) accelerometer, an validated construct measure of MVPA . The accelerometer was worn on the waist seven consecutive days except when in water. Actigraph data was analyzed by separating out-of-school-hours from the sleeping times and school times by Actilife 5.1. The epoch length was set at 15 seconds. Non-wear time was defined as 60 minutes of consecutive zeroes. We chose to use Evenson’s cut-points  which are recommended for use with school-aged children . That is, the cut-point for at least moderate activity is 2296 counts per minute .
A valid day for the analysis was defined as at least eight hours of data in the child’s out-of-school-hours. Each child had to have four days of valid data with one weekend day. The total minutes of vigorous and moderate activity in out-of-school-hours indicated by the accelerometer were combined to form a moderate-to-vigorous-activity measure (MVPA). The average minutes of MVPA in children’s out-of-school-hours per day was calculated by dividing the total amount of MVPA in the selected four days by four.
All the analyses were conducted by using SPSS software, version 19.0 (SPSS, Chicago USA). To test the repeatability of questions, intra-class correlations (ICC) with 95 % confidence intervals were calculated using a two-way random model with an absolute agreement type .
To test the validity, the self-reported daily MVPA duration and frequency were compared to the accelerometer MVPA minutes and by calculating the Spearman’s correlations with 95 % confidence intervals . Bland-Altman plots with 95 % limits of agreement were calculated to measure the agreement between and within average daily MVPA duration (min/day) according to the questionnaire and the average daily MVPA (min/day) according to the accelerometer.
The Kruskal-Wallis tests with pairwise comparisons using the Dunn-Bonferroni correction were done for testing if the questionnaire was able to categorize children according to their levels of MVPA. The objectively measured MVPA was compared separately to a) the quartiles of self-reported MVPA duration and b) to the categorized self-reported MVPA frequency.