Reproducibility of the Motor Optimality Score–Revised in infants with an increased risk of adverse neurodevelopmental outcomes

Aim: To determine reproducibility of the Motor Optimality Score–Revised (MOS-R) to assess infants at high risk of adverse neurodevelopmental outcomes, including cerebral palsy (CP), autism, and developmental delays.

Method: Thirty infants (18 males, 12 females, gestational age mean [range] = 32.5 [23–41] weeks) were randomly selected, according to 2-year outcome (typically developing; CP; or adverse neurodevelopmental outcome [ad-NDO]) from a prospective cohort. Participants had two General Movements videos between 12 weeks and 15 + 6 weeks corrected age. Six assessors, masked to history and outcomes, independently scored the MOS-R from videos. Assessors scored either one (Group 1; n = 3) or two videos for each infant (Group 2; n = 3). Intraclass correlation coefficient (ICC), Gwet's agreement coefficient, and limits of agreement were calculated.

Results: Combined interassessor reliability (IRR) over six assessors for total MOS-R was ‘fair’ (ICC = 0.56, 95% confidence interval [CI] 0.41–0.72), and ‘excellent’ with consensus agreement (ICC = 0.99, 95% CI 0.98–0.99). Analyses demonstrated a mean interrater difference of 0.316 (95% limits of agreement −11.51, 12.14) over 450 comparisons (15 pairs). IRR was ‘moderate’ to ‘almost perfect’ across subcategories, with the highest reliability ‘movement patterns’ (Gwet's agreement coefficient = 0.73–1.00) and the lowest ‘postural patterns’ (0.45–0.73). Assessors who scored two videos (Group 2) demonstrated higher reproducibility. IRR for total MOS-R was ‘excellent’ when infants were typically developing (ICC = 0.90), and ‘good’ for CP (0.74) and ad-NDO (0.68).

Interpretation: The MOS-R is a highly reproducible tool for assessing infants at high risk of ad-NDOs and is feasible for implementation in clinical settings. Reproducibility is best when the tool is used by experienced assessors to gain consensus agreement.