Cluster analysis to identify foot motion patterns in children with flexible flatfeet using gait analysis—A statistical approach to detect decompensated pathology?

Abstract

IntroductionThe paediatric flexible flatfoot constitutes the major cause of clinic visits for orthopaedic foot problems. It shows variations of deformities in different planes and locations of the foot and its indication for treatment have been extensively discussed. Despite its high prevalence there exists no classification of flatfeet during walking as a prerequisite for treatment decision. Therefore, the aim of this study is to classify flexible flatfeet based on 3D foot kinematics during walking. MethodsPatients age 7–17 years with flexible flatfeet (N = 129, 255 feet) of non-neurogenic or syndromic origin, were retrospectively included. Patients underwent gait analysis using the Oxford Foot Model after standard clinical examination. A k-means cluster analysis was performed on 3 scores derived from the principal component analysis of the foot kinematic waveforms over the gait cycle. Gait and clinical parameters were then statistically tested between clusters. ResultsCluster analysis revealed two groups of flexible flatfeet that were discriminated best by the inversion at push-off during walking. Cluster 2, including 110 feet, showed an average eversion instead of an inversion at push-off and a lower number of heel rises in the clinical test. Both was significantly different between clusters (p < 0.001). DiscussionBased on the findings, the resultant clusters can be interpreted as describing compensated and decompensated feet, with the latter presenting a group that may require surgical interventions, even if they are not yet present with pain. The hindfoot inversion capability at push-off is the most important variable in the 3D gait analysis to classify flexible flat feet.