Teun Remmers

Daily weather and children's PA patterns | 95 Discussion This study first investigated associations between weather elements and children’s daily PA patterns, and secondly the moderating effects of day of the week (weekdays vs. weekend days) and gender, age and BMI on the relationship between these weather elements and PA, using an extensive longitudinal design containing measurements across four seasons. Temperature and day type (weekdays versus weekends) were the strongest variables explaining MPA and VPA; followed by equal contributions from solar radiation and humidity. Temperature showed a curvilinear relationship, with highest engagement in MPA and VPA levels occurring at 22 and 20 degrees Celsius, respectively. Relative humidity was consistently related to lower MPA and VPA, and solar radiation was consistently related to higher MPA and VPA. Little rainfall (0.1–7.8mm) was associated with higher MPA compared to no rainfall, but no association was found for heavier rainfall (7.9-46.0mm). When compared against sedentary time, results were equally strong but in opposite direction, suggesting that the increments observed in MPA and VPA were accompanied with a similar decrease in sedentary time. The curvilinear relationship found between temperature and MPA and VPA was highly comparable with results from the study by Lewis et al. (21), which also reported an optimal temperature around 20-25 °C related to PA among children in Australia and Canada. For light PA however, we did not observe this typical curvilinear relationship. In addition, our study found that there was a notable curvilinear relationship between temperature and VPA during weekdays. During weekends, we found no evidence for such a curvilinear relationship, but still, higher temperatures were related to lower VPA. Lewis et al. (21) reported that the influence of weather was stronger on weekdays; while Duncan et al. (9) and Atkin et al. (3) in contrast reported that this was generally true for weekend days. Inconsistency regarding these findings may be explained by the possibility that on weekends with temperatures above 20 °C, more free-time may have been spent in water-related activities which likely resulted in misclassification of actual PA behavior as children were instructed to not wear the accelerometer during water-activities (21). Another explanation may be that the scheduling of structured PA (e.g. team sports) may have resulted in fewer PA opportunities in warmer periods of the year (26). However, by allowing for an individualized intercept and repeated measures for each participant across each day of measurement, it is more conceivable that daily changes in temperature influenced PA in our study, rather than the role of potential changes in structured PA. We found that solar radiation was consistently associated with higher MPA and VPA. Duncan et al. also examined solar exposure in their New Zealand study (9), but no associations were found with PA (9). This may be due to different definitions of solar radiation, as the present study defined solar radiation as the daily total in MJ/m 2 , while Duncan et al. (9) defined solar radiation as the duration of absent cloud cover. As the former provides more valid information about the actual intensity of solar radiation, this measure may be more likely to be related with PA behavior. In addition, although suggested by Duncan et al. (9), this is the first study that has investigated moderation