Inter- and intrarater reliability of an international consensus-based gait classification system in children with cerebral palsy

INTRODUCTION In CP many patterns of pathological gait can be identified, which all require a specific treatment approach. Recently, an international expert panel from eight research groups across the US and Europe have identified gait patterns for children diagnosed with spastic CP in a Delphi consensus study [1]. A total of 48 gait patterns distributed over the lower limb joints and based on kinematic and kinetic data were defined. This study aimed to introduce the classification system to experienced and inexperienced clinicians and assess inter- and intrarater reliability. MATERIALS AND METHODS One expert rater involved in the Delphi study, classified a retrospective sample of 76 gait trials belonging to 57 CP patients between 4-17 years old (GMFCS I-III, bilateral CP n=38). Two master students physiotherapy, specializing in pediatric rehabilitation, were trained to apply the classification system through educational presentations and use of a training dataset. Both students then classified all 76 gait trials twice within 30 days (3-day interval after first classification) to allow for intrarater reliability estimations. The first classification was used to assess interrater reliability. Cohen’s and Light’s Kappa as well as percentage of agreement were calculated with SPSS Statistics for Windows, version 20.0 (Armonk, NY) [2,3]. RESULTS Most patterns reached almost perfect agreement (tables 1 and 2). For gait patterns in the sagittal plane of the hip, knee in stance, and ankle in stance, interrater agreement was substantial. DISCUSSION The excellent agreement results indicate that, given a thorough training phase, inexperienced clinicians can reliably apply the classification system that was identified by an international group of experts. Future research could use weighted Kappa to identify individual patterns on which agreement might have been low. Additional training and clarified definitions of those patterns could further improve agreement results. REFERENCES [1] Nieuwenhuys et al. submitted to Dev Med Child Neurol. [2] Cohen J, Educ. Psychol. Meas.1960 [3] Light RJ, Psychol. Bull.1971Nominated for Best Paper Awardstatus: publishe