BackgroundPatient demographic and clinical characteristics may be factors associated with the success of conservative treatment for pyogenic spondylitis. The ability of imaging findings at initial diagnosis to predict patient outcomes remains unclear. The aim of this study was to investigate the risk factors associated with conversion surgery after conservative treatment for pyogenic spondylitis, with a special focus on the initial computed tomography (CT) findings. MethodsThis study enrolled 35 patients with pyogenic spondylitis who underwent CT and magnetic resonance imaging (MRI) at diagnosis and were followed-up for more than 6 months. Patients were diagnosed by MRI, and the degree of bone destruction was assessed on CT cross-sectional images. Vertebral body destruction was classified as grades 0 (almost normal), 1 (endplate irregularity), 2 (vertebral body destruction not involving the posterior wall), and 3 (destruction involving the posterior wall). Patients were divided into four groups based on grade of bone destruction and their clinical characteristics were compared. Results1, 11, 11, and 12 patients were classified as grades 0, 1, 2, and 3, respectively. Univariate analysis showed no significant differences in the demographic and clinical characteristics of the four groups. Eighteen (51.4 %) patients had been treated surgically, with the rate of surgical treatment being significantly higher in patients with grade 3 (83.3 %) than in those with grades 0 + 1 (25 %) and grade 2 (45.5 %) (P < 0.05). Multivariate analysis showed that epidural abscess on MRI (odds ratio [OR] 10.8, 95 % confidence interval [CI] 1.68–69.7), grade 3 bone destruction on CT (OR 3.97, 95 % CI 1.21–13.0), and C-reactive protein (CRP) improvement rate after 1 week of treatment (OR 0.95, 95 % CI 0.91–0.99) were risk factors for surgery. ConclusionsEarly surgical treatment should be considered for patients with pyogenic spondylitis who present with an epidural abscess on MRI and bone destruction extending to the posterior wall on CT at the time of diagnosis.