Statistical analyses

For logistical reasons and for improving compliance, no patients or control participants were included into the study during the holidays; thus PA measurements were recorded only during school periods. Similarly to precedent paper,²⁴ patients and controls were matched according to age (±6 months), gender (male/female) and season. The participants characteristics used for matching were compared between groups using an unpaired Student’s t-test. A Shapiro-Wilk test was used to reject the normality distribution of variables relative to PA. Thus, an unpaired Wilcoxon test was used to compare PA levels between cases and controls. Results were reported as mean±SD for all parameters. The imputation of the missing values was used only if the rate of missing values was lower than 10%. Otherwise, rows were excluded to limit bias due to missing values. Data analyses were performed using STATA V.9.2 (StataCorp).