Clinician- versus caregiver-rated scales as outcome measures of repetitive-restricted behaviors in clinical trials of autism: A systematic review and meta-analysis.

The agreement between treatment effects measured by clinician- and caregiver-ratings of repetitive-restricted behaviors (RRBs) is important for clinical practice and research but is still unclear. Therefore, we conducted a post-hoc meta-analysis of placebo-controlled randomized-controlled trials (RCTs) investigating pharmacological and dietary-supplement treatments for autism that reported both clinician- and caregiver-ratings of RRBs. Treatment effects between medications and placebo were quantified with standardized mean differences (SMDs). The agreement between clinician- and caregiver-rated SMDs was investigated with an intraclass correlation coefficient (ICC) and random-effects meta-analysis of their difference (Δg). A meta-regression investigated the association between clinician (dependent) and caregiver-rated SMDs (independent variable). Certainty in the evidence was evaluated using the GRADE approach. We identified 15 eligible placebo-controlled RCTs with 1567 participants, from which 13 included children/adolescents and 9 reported data for the pair of the clinician-rated Yale-Brown Obsessive Compulsive Scale (YBOCS) and the caregiver-rated Aberrant Behavior Checklist-Stereotypic Behavior (ABC-S). There was on average a good agreement between clinician- and caregiver-rated SMDs (ICC=0.84, 95% confidence intervals [0.55, 0.95]), no clear difference between them (Δg=0.08, 95%CI[-0.06, 0.21], 95% prediction intervals [-0.16, 0.31]), and the beta of the meta-regression was 0.62, 95%CI[0.27, 0.97]. The certainty of the evidence was low due to concerns in imprecision and inconsistency. Our analysis showed on average a good agreement between clinician- and caregiver-rated treatment effects in RRBs, yet discordance could be expected in future RCTs, given the wide prediction intervals. It is also not certain that these results could be generalizable to other rating scales and intervention modalities. Ethics committee approval: Not applicable as a meta-analysis of previously published studies..