Interrater reliability study of cerebral palsy diagnosis, neurological subtype, and gross motor function

Abstract

To evaluate the interrater reliability of the inclusion in registries and classification of children with cerebral palsy (CP). Two studies were conducted. In study 1, 12 paediatricians from 11 countries viewed video sequences of 12 children with or without CP (nine males, three females; median age 6y; range 2-16). In study 2, 19 professionals from eight countries participated in an online exercise. They had to classify the same children but based on written vignettes. All participants had to evaluate whether the child had CP, the neurological subtype (Surveillance of Cerebral Palsy in Europe classification system), and gross motor function level (Gross Motor Function Classification System [GMFCS]). Kappa (κ) coefficients were calculated for categorical variables and intraclass correlation coefficients (ICCs) for ordinal data. Reliability was excellent in assessing whether or not a child had CP in study 1 (κ=1.00) and substantial in study 2 (κ=0.73); 95% confidence interval [CI] 0.58-0.87). For the neurological subtype, overall κ between paediatricians was 0.85 (95% CI 0.68-0.98), with full agreement observed for eight children. In study 2, overall κ was 0.78 (95% CI 0.61-0.91) with full agreement seen for five children. For the GMFCS, the ICC was 0.88 (95% CI 0.78-0.95) in study 1 and 0.80 (95% CI 0.64-0.91) in study 2. Reliability was excellent for all characteristics classified by paediatricians viewing the videos and substantial for professionals reading vignettes.