Prediction of gait outcome with the knee-ankle-foot orthosis with medial hip joint in patients with spinal cord injuries: A study using recursive partitioning analysis

Study design: Retrospective study of the degree of gait independence achieved by persons with spinal cord injury (SCI) using knee-ankle-foot orthosis with a medial single hip joint (MSH-KAFO). Objective: To examine the effects of the neurological level, degree of paresis, age, and inhibitory physical/other factors on the gait with a MSH-KAFO in patients with SCIs. Setting: Three university hospitals and two rehabilitation hospitals in Japan. Methods: The 45 patients (36 men, nine women) examined included 10 with injuries in the cervical cord between C6 and C8 (group C), 20 with injuries in the upper-middle thoracic cord between T4 and T10 (group UT), and 15 with injuries in the lower thoracic-lumbar cord between T12 and L1 (group TL). Mean age was 34.0 years (range 16-68 years). Of these patients, 13 used the Walkabout, four used the gear joint, and 28 used the Primewalk as the medial hip joint. Recursive partitioning, which predicted the final status of gait from the level, degree of paresis, age, and inhibitory factors, was performed, and a decision tree for gait was constructed. Inhibitory factors were spasticity, involuntary spasms or muscle contractions, pain, contracture, weakness of the upper extremities, and decreased motivation to perform gait exercise. The degree of gait independence was rated on the following five-point scale: outdoor independent gait (5 points), indoor independent gait (4 points), indoor supervised gait (3 points), indoor assisted gait (2 points), and gait within parallel bars (1 point). New branches were added to the decision tree for gait based on the clinical experience, thereby constructing a new decision tree. Results: The coincident ratio between the value predicted on the basis of the decision tree of gait and the value actually observed was 53.3%. The coincident ratio between the value predicted on the basis of the modified decision tree of gait and the actually observed value was 68.9%. Conclusion: The results provide valuable information to medical teams that may assist prescription of gait orthoses.