Abstract
Cervical spinal cord atrophy occurs after spinal cord injury. The atrophy and how level of injury affects atrophy differs between studies. A systematic review and metaanalysis were done after systematic searches of PubMed, CINAHL, APA PsycInfo and Web of Science. English language original studies analyzing MRI cervical spinal cord cross-sectional area in adults with spinal cord injury were included. Atrophy and correlation between injury level and atrophy were estimated with random-effects models, standardized mean differences, and 95% confidence intervals. 24 studies were identified. 13/24 studies had low risk of bias. Cord atrophy meta-analysis of 18 articles corresponded to a standardized mean difference of -1.48 (95% CI -1.78 to -1.19) with moderate to large interstudy heterogeneity. Logarithmic time since injury influenced heterogeneity. Longitudinal atrophy was best described by a logarithmic model, indicating that rate of spinal atrophy decreases over time. Meta-correlation of eight studies indicated more severe atrophy in more rostral injuries (0.41, 95% CI 0.20-0.59). Larger and preferably longitudinal studies, data sharing, and standardized protocols are warranted.