Many children with cerebral palsy walk with excessive knee flexion, an inefficient locomotion pattern known as crouch gait that progressively worsens over time without intervention. Patients often receive surgeries to lengthen tight or spastic muscles or to correct bone malalignments. These treatments are intended to reduce excessive knee flexion and enhance mobility, but surgical outcomes are variable, largely because there are no standardized protocols for determining what treatment a particular patient should receive. This dissertation sought to answer two main questions. First, can we objectively identify the set of biomechanical factors that cause a patient to walk with excess knee flexion? Second, can we use these factors to predict whether a patient’s crouch gait will improve after receiving treatment? We addressed these questions by analyzing the dynamics of crouch gait and developing a multi-variable regression model to predict whether subjects’ crouch gait would improve or deteriorate between visits to the gait analysis laboratory.

We first determined the influence of abnormal bone geometry and crouched gait postures on the function of muscles during walking using a three-dimensional model of the musculoskeletal system. Our analysis revealed that a tibial torsion deformity, a common bone malalignment, reduces the capacity of lower limb muscles to generate extension of the knee and hip joints. Excess tibial torsion may thus be a significant contributor to crouch gait and warrant surgical correction. In addition, our analysis showed that a crouched gait posture markedly reduces the capacity of most lower limb muscles to extend the knee and hip joints. A crouched gait posture also increases the hip and knee flexion accelerations induced by gravity. These findings help explain the altered muscle activations, increased energy requirements, and increased joint loading associated with walking in a crouch gait.

We also developed a multi-variable linear regression model to estimate how much subjects’ crouch gait would change between hospital visits by analyzing patients’ gait kinematics and surgical histories. The regression model was able to explain 49% of the variance in the change in knee flexion between gait visits for the 353 subject-limbs that we analyzed. Further, the regression model classified subject-limbs as ‘Improved’ or ‘Unimproved’ with approximately 70% accuracy, in contrast with the observed improvement rate of 48% among subject-limbs in the study. We further demonstrated that more improvement in crouch gait was expected when subjects had i) adequate hamstrings lengths and velocities, ii) normal tibial torsion, and iii) greater extensor muscle strength, three variables drawn from our knowledge about the biomechanics of crouch gait.

This dissertation quantified the changes in muscle function that occur as a consequence of a tibial torsion deformity and walking in a crouched posture. Using this information, we developed guidelines for identifying good candidates for a tibial derotation osteotomy. Next, drawing on our understanding of the biomechanics of crouch gait, we developed and tested a statistical model that was able to predict whether a subjects’ crouch gait would improve given their gait kinematics and treatment plan. This work establishes a new framework—combining biomechanical modeling and statistical analysis—for understanding gait pathology and objectively planning treatment.