Direct comparison of measured and calculated total knee replacement force envelopes during walking in the presence of normal and abnormal gait patterns

Abstract

Knee joint forces measured from instrumented implants provide important information for testing the validity of computational models that predict knee joint forces. The purpose of this study was to validate a parametric numerical model for predicting knee joint contact forces against measurements from four subjects with instrumented TKRs during the stance phase of gait. Model sensitivity to abnormal gait patterns was also investigated. The results demonstrated good agreement for three subjects with relatively normal gait patterns, where the difference between the mean measured and calculated forces ranged from 0.05 to 0.45 body weights, and the envelopes of measured and calculated forces (from three walking trials) overlapped. The fourth subject, who had a “quadriceps avoidance” external moment pattern, initially had little overlap between the measured and calculated force envelopes. When additional constraints were added, tailored to the subject’s gait pattern, the model predictions improved to complete force envelope overlap. Coefficient of multiple determination analysis indicated that the shape of the measured and calculated force waveforms were similar for all subjects (adjusted coefficient of multiple correlation values between 0.88 and 0.92). The parametric model was accurate in predicting both the magnitude and waveform of the contact force, and the accuracy of model predictions was affected by deviations from normal gait patterns. Equally important, the envelope of forces generated by the range of solutions substantially overlapped with the corresponding measured envelope from multiple gait trials for a given subject, suggesting that the variable strategic processes of in vivo force generation are covered by the solution range of this parametric model.

Introduction

Detailed knowledge of in vivo knee contact forces and the contribution from muscles, ligaments, and other soft-tissues to knee joint function is essential for evaluating total knee replacement (TKR) designs. Laboratory tests and computational models of TKRs and natural knee joints require accurate force inputs in order to physiologically replicate in vivo conditions. If available, patient-specific knee contact forces and muscle forces could be used to determine testing protocols that are truly representative of specific TKR designs, design rehabilitation protocols or predict the safety of recreational activities, and monitor recovery progress after surgery.

Knee joint forces are difficult to obtain; currently, in vivo force data from instrumented total knees are only available for a few subjects for walking, chair rising/sitting, stair ascent/descent, and other activities (D'Lima et al., 2008, D'Lima et al., 2006, Heinlein et al., 2009, Kutzner et al., 2010, Mündermann et al., 2008). Consequently, computational models are necessary to bridge the knowledge gap between the available data from the few patients with a specific implant type to patient-specific knee joint contact forces for a larger patient population and multiple TKR designs. Numerical models can be used to calculate muscle and passive structure forces simultaneously with contact forces, and thus allow a more comprehensive and systematic evaluation of knee joint loading.

The unknown validity and sensitivity of modeling assumptions to different gait patterns is illustrated by results from previous models where calculated knee joint contact forces range from 1.7 to 4.3 body weights during walking (Komistek et al., 1998, Komistek et al., 2004, Komistek et al., 2005, Morrison, 1970, Paul, 1976, Wimmer and Andriacchi, 1997). With the recent availability of data from instrumented TKRs, direct comparisons to numerical models are now possible. We have previously developed a numerical model which calculates a range or envelope of possible three-dimensional contact forces for both the medial and lateral compartments of the tibial plateau (Lundberg et al., 2009). The force envelope is intended to represent the natural physiological variability in gait, as any number of strategies could be used to balance the external moments and forces measured during gait analysis. The purpose of this study was to test the validity of the knee joint contact forces predicted by the parametric numerical model. Model validity is tested by direct comparison of the predicted contact forces to measurements from four subjects with instrumented TKRs during the stance phase of gait. Model sensitivity to abnormal gait patterns is also discussed.