Knee osteoarthritis (OA) is a debilitating disease of the musculoskeletal system that is associated with high morbidity and disability across the world. There is yet no definitive cure for knee OA, and the development of early diagnostic and treatment strategies for the disease have been hindered by our lack of understanding of the complex, multifactoral and progressive nature of its pathophysiology. Mechanical factors have been implicated in the pathomechanics of knee OA, but the treatment of disease severity in previous study has been poor and there has been a lack of investigation into interactions between biomechanical factors. In this thesis, two models of knee OA disease severity are developed and a multivariate analysis approach is used to understand the mechanistic role of biomechanical factors in knee OA progression.

The kinematic and kinetic joint dynamics at the knee, hip and ankle joints, and the electromyographic activation patterns of seven major muscle sites surrounding the knee joint were measured during gait for 181 subjects categorized into three distinct clinical groups:​ i) asymptomatic, ii) moderate knee OA, and iii) severe knee OA. A cross-sectional knee OA severity model was used to identify distinctly different biomechanical mechanisms that discriminated between moderate knee OA and asymptomatic gait patterns and between moderate and severe patterns. Changes in the knee and hip loading patterns during gait separated moderate knee OA and asymptomatic patterns;​ knee kinematic changes and kinetic changes at the hip and ankle joints separated severe patterns from moderate. A second, continuous severity model captured the spectrum of radiographic and symptomatic disease severity within the moderate knee OA subject group. Symptomatic disease severity of the moderate group was not well predicted with biomechanical variables. Radiographic disease severity was well predicted with a multivariate model that included a combination of high overall magnitudes of knee adduction moment, high BMI, abnormal gastrocnemius muscle activation patterns, and abnormal kinematic and kinetic ankle biomechanics. This novel finding confirmed the multivariate nature of the role of biomechanical factors in the pathogenesis of knee OA, and suggested multifactoral pathways of disease progression that should be further investigated.