Table 2 Overview of analytical strategy and key results

Assessment of the pandemic impact on changes of the AFAR-based symptom ratings (i.e., Adaptive living skills, lower-order RRB, higher-order RRB, activity/inattention, oppositional, anxiety/affect, sleep problems) and services (lost or modified at and outside school)

[Table 1, Additional file 1: Table S6]

Pandemic outcome subgroups identification and characterization

Hierarchical clustering

Four outcome subgroups with differing profiles of change relative to the aggregate’s average change:

 -Broad symptom worsening only (20%)

 -Primarily modified services (23%)

 -Primarily lost services (6%)

 -Average symptom/service changes (53%)

[Figs. 4 and 5, Additional file 1: Table S7, Fig. S3]

Whole (aggregate) sample main effects analyses

One-way repeated measures MANCOVA (within-subject factor time; covariate: contributing sample); Post hoc one-way repeated measures ANCOVA for each symptom factor

Significant effect of time was driven by worsening sleep problems ratings, other symptoms did not reach statistical significance [Fig. 5]

Central tendency descriptive measures

On average, lost 1 service and continued 1 other at and outside school

[Additional file 1: Table S7]

Prediction of outcome subgroup membership across 20 features including pre-pandemic variables (e.g., service at and outside school, child, and family’s characteristics), pandemic-related experiences (e.g., COVID worry) and environment (containment measures)

[Fig. 2, Additional file 1: Table S2, Fig. S4]

Random Forest classification, ranking feature importance indexed by out-of-bag-error (OOBE)

81% classification accuracy. Pre-pandemic services in and outside school, Sringency index, Lifestyle Stress, COVID Worries, new COVID infections and age were top predictors (OOBE 16–1%). Other features had negligible importance (< 1%). Each outcome subgroup had distinct profiles of increases or decreases across the top predictors [Fig. 6, Table 3, Additional file 1: Table S7]