Recovery Trajectories of Lower Limb Motor Function After Cervical Spinal Cord Injury: A Latent Class Mixed Model Analysis.

OBJECTIVE: To identify the latent recovery trajectories of lower limb motor function after cervical spinal cord injury using a latent class mixed model (LCMM) and to examine clinical factors associated with trajectory class membership. DESIGN: Retrospective longitudinal cohort study. SETTING: A single specialized spinal cord injury center in Japan. PARTICIPANTS: A total of 516 participants and 4,040 longitudinal observations were included. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Longitudinal Lower Extremity Motor Score (LEMS) data within one year of injury were analyzed using the LCMM. Candidate models with one to five latent classes were evaluated using information criteria and clinical interpretability. Factors associated with trajectory class membership were identified using multinomial logistic regression with Firth correction. RESULTS: The LCMM identified five distinct recovery trajectory classes: early high recovery, steep nonlinear recovery, moderate recovery with early plateau, low recovery, and slow sustained recovery. The mean posterior probabilities were 0.91-0.98, supporting the validity of the classification. LEMS at admission was consistently associated with trajectory class membership across all comparisons. Age, ASIA Impairment Scale classification, and bony injury were associated with specific trajectory classes. CONCLUSIONS: Lower limb motor function recovery after cervical spinal cord injury can be classified into multiple longitudinal trajectories. Trajectory-based analysis may contribute to understanding heterogeneous recovery patterns and support prognostic assessment in clinical practice.