Can a linear combination of gait principal component vectors identify hip OA stages?

Abstract

Hip osteoarthritis (OA) has been shown to affect gait patterns of lower extremities. However, until now, no specific identifying gait characteristics for the various disease stages of hip OA have emerged. The present study addresses the following questions: (1) does a vector-based principal component analysis (PCA) discriminate between various disease stages? And, is this analysis more robust than using discrete gait variables? (2) Does the elimination of differences in walking speed affect the discriminatory robustness of a vector-based PCA? De-identified data sets of forty-five unilateral hip OA patients with varying disease stages and twenty-three age-matched, healthy control subjects were obtained from an available repository. PCA was performed on trial matrices consisting of all external joint moments and sagittal joint angles of one full gait cycle. Group differences in sagittal angles, external moments and the linear combination of PC vectors were investigated using spatial parameter mapping (SPM), a statistical vector field test. Several individual gait variables (i.e. joint moments or angles) demonstrated differences between healthy and moderately and/or severely affected subjects. Only the hip adduction moment could discriminate between the healthy and the early-stage OA group. There was no variable that could distinguish between all OA disease stages. In contrast, the linear combination of PC vectors demonstrated significant group differences between all stages of osteoarthritis; furthermore, these group differences stayed significant when matched speeds were input to the model.