Interobserver agreement in the classification of cerebral palsy.

Interobserver variation in diagnosis is thought to be an important source of bias in studies of cerebral palsy. Kappa (kappa) statistics were used as a measure of interobserver diagnostic agreement for two case series of 20 children attending an institution for the motor handicapped. kappa increased threefold after standardisation of diagnostic terms. Sources of diagnostic variation are discussed and possibilities and benefits of its further reduction explored.