Statistical analysis
The primary exercise measures were average daily steps, average daily MVPA measured by the pedometer, and the PAQ questionnaire score. Activity values that were extreme (2 standard deviations above or below average) were discarded to account for any technical errors or induced bias.
Associations between the average daily steps, average MVPA and PAQ score, with PFT and MRI measures, were conducted using Spearman correlations. We interpreted the absolute values of correlation coefficients as the following: 0.10-0.39 (weak), 0.40-0.69 (moderate), 0.70-0.89 (strong), and >0.9 (very strong) associations26. All statistical analyses were carried out using R version 4.0.3 (R Core Team, Vienna, Austria).
Separate multiple linear regression analyses were conducted to assess the association between each of our primary outcomes (average daily steps and average daily MVPA), respectively, with predictors of interest, including history of BPD (yes/no), FEV1% predicted and MRI total proton density. Sensitivity analyses examining the impact of recruitment site were conducted for each multiple linear regression. Unadjusted and adjusted estimates with 95% CI were reported.
To explore the impact of ADHD medication use on daily step count, we conducted an exploratory descriptive analysis comparing daily step count between children taking ADHD medication and those not taking ADHD medications, as well as a subgroup analysis only within the group with BPD.