Statistical analyses:
Baseline clinical characteristics and BOT-2 measures were summarized as
either median (Interquartile range [QR]; min – max), mean ±
standard deviation (SD) or frequency (%). For group comparisons,
participants were dichotomized into two groups based on their percentile
score for TMC, Bilateral Coordination and Strength and Agility,
corresponding to clinically meaningful cutoffs of below 18% and below
25%. Associations between continuous measures were assessed via Pearson
correlation, continuous and categorical measures with one-way ANOVA and
categorical via Chi-square tests for independence. Studentized
jack-knife residuals were plotted to examine the assumption of linear
regression. When categorizations had intrinsic ordering, tests for
linear trends were conducted. When appropriate, non-parametric methods
such as Wilcoxson Rank Test and Fisher’s exact were used. Regression
coefficients and correlation estimates are presented with 95%
confidence intervals. No adjustments for multiple comparisons were made
as this project was viewed as retrospective and hypothesis generating.
All analyses were conducted in SAS version 9.3 (Cary, NC).