Predictors of neurological recovery in spinal epidural abscess: the impact of radiographic compression, ASIA scores, and surgical delay

Background – Controversy remains whether spinal epidural abscess (SEA) should be managed conservatively or by early surgical decompression. It is essential to understand how clinical and radiological parameters correlate with neurological deficits and how they predict recovery trajectories. Objective – Our aim was to evaluate whether the preoperative magnetic resonance imaging (MRI) features, the clinical management factors, and the neurological status on admission can predict neurological recovery in patients with SEA. Methods – We retrospectively analyzed patients diagnosed with SEA between 2021 and 2024 with a minimum follow-up of 12 months. Collected variables included demographics, duration of preoperative antibiotic therapy, the American Spinal Injury Association (ASIA) motor scores, and the Numerical Rating Scale (NRS) pain scores. Quantitative MRI parameters were assessed, including cord T2 hyperintensity, ossification of the ligamentum flavum (OLF), and maximum spinal cord compression (MSCC). Neurological recovery was calculated using the ASIA recovery rate, which was dichotomized as good (≥50%) or poor (<50%). Logistic regression and receiver operating characteristic (ROC) analyses were performed. Results – A total of 125 patients were included, of whom 24 were treated conservatively and 101 underwent surgical decompression. Cord T2 hyperintensity (β = −13.79, p < 0.01), OLF (β = −16.59, p < 0.01), >50% contrast enhancement (β = −8.64, p = 0.002), and MSCC (β = −0.3, p < 0.001) were significantly correlated with worse admission ASIA scores. Cord T2 hyperintensity (β = 0.37, p = 0.038) and MSCC (β = −0.01, p = 0.043) were also associated with higher admission NRS scores. Prolonged preoperative antibiotic therapy exceeding 3 weeks was associated with poorer neurological recovery (p = 0.04). An admission ASIA score of >88 demonstrated excellent prognostic performance for good recovery (AUC = 0.927, 95%CI = 0.876–0.977; sensitivity = 0.81, specificity = 0.925). Conclusion – Both the clinical and radiological parameters upon presentation are predictors of neurological recovery rates in SEA. Delayed surgical decompression beyond 3 weeks is associated with poorer neurological recovery. Clinical trial registration – [website], identifier GWLCZXEC-SOP-K-2025-137.