Abstract
To develop a simple screening tool, the Thumbs-Out Classification (TOC), for detecting thumb-in-palm deformity during clinical follow-ups of children with cerebral palsy (CP), and to evaluate its reliability. The TOC was developed through expert consensus and consists of three levels of radial thumb abduction: normal (green), slightly reduced (yellow), and severely reduced (red). Classification cut-offs were established using goniometric measurements from 20 typically developing children. These cut-offs were then applied to classify active and passive radial thumb abduction in 61 children with CP (median age 9 years). Inter-rater, intra-rater, and test-retest reliability were assessed using Cohen's weighted kappa (κw) and percentage agreement. Agreement between goniometric measurements and TOC levels was moderate for both active (κw = 0.59) and passive (κw = 0.54) radial abduction, with substantial percentage agreement (79% and 81%, respectively). Inter-rater reliability was almost perfect for active (κw = 0.93) and passive (κw = 0.89) radial abduction. Test-retest reliability was also almost perfect (active κw = 0.84; passive κw = 0.93). The TOC is a reliable screening tool for monitoring thumb positioning in children with CP. It may facilitate detection of thumb-in-palm deformity and support timely interventions aimed at optimising hand function.