Characteristics of and risk factors for severe neurological deficit in patients with pyogenic vertebral osteomyelitis: A case-control study

Abstract

Severe neurological deficit (SND) is a rare but major complication of pyogenic vertebral osteomyelitis (PVO). We aimed to determine the risk factors and the variables associated with clinical improvement for SND during PVO.This case-control study included patients without PVO-associated SND enrolled in a prospective randomized antibiotic duration study, and patients with PVO-associated SND managed in 8 French referral centers. Risk factors for SND were determined by logistic regression.Ninety-seven patients with PVO-associated SND cases, and 297 controls were included. Risk factors for SND were epidural abscess [adjusted odds ratio, aOR 8.9 (3.8-21)], cervical [aOR 8.2 (2.8-24)], and/or thoracic involvement [aOR 14.8 (5.6-39)], Staphylococcus aureus PVO [aOR 2.5 (1.1-5.3)], and C-reactive protein (CRP) >150 mg/L [aOR 4.1 (1.9-9)]. Among the 81 patients with PVO-associated SND who were evaluated at 3 months, 62% had a favorable outcome, defined as a modified Rankin score ≤ 3. No factor was found significantly associated with good outcome, whereas high Charlson index [adjusted Hazard Ratio (aHR) 0.3 (0.1-0.9)], low American Spinal Injury Association (ASIA) impairment scale at diagnosis [aHR 0.4 (0.2-0.9)], and thoracic spinal cord compression [aHR 0.2 (0.08-0.5)] were associated with poor outcome. Duration of antibiotic treatment was not associated with functional outcome.SND is more common in cervical, thoracic, and S. aureus PVO, in the presence of epidural abscess, and when CRP >150 mg/L. Although neurological deterioration occurs in 30% of patients in early follow-up, the functional outcome is quite favorable in most cases after 3 months. The precise impact of optimal surgery and/or corticosteroids therapy must be specified by further studies.

Figure 1

Flow-chart of 394 patients with pyogenic vertebral osteomyelitis with (cases) or without (controls) severe neurological deficit. Trial Cohort: Patients from randomized trial NCT00764114, Retrospective cohort: Cases included from the French Hospital Discharge Database, Risk factor analysis: patients included for descriptive analysis of evolution and treatment, Outcome analysis: patients included in the analysis of neurological outcome at 3 months. Post-op PVO = postoperative pyogenic vertebral osteomyelitis.

Figure 2

Functional outcome according to parameters significant in univariate analysis, and representation of global model. Good functional evolution was defined by a Rankin score of 3 or better indicating the ability of walking without human help. Each square represents cumulative incidence of good functional evolution over time according to each studied variable, with associated global P value in univariate analysis. In brackets are the effectives of each subgroup. (A) Outcome whether decompressive surgery is performed or not (isolated arthrodesis was not considered as decompressive); (B) Outcome according to the worst AIS at installation of neurological impairment; (C) Outcome according to modified Charlson score with a cut-off at 5; (D) Outcome according to spine level of compression; (E) Graphical representation of final Cox proportional hazards model with confidence interval and 50% cumulative incidence line.

Figure 3

Magnetic resonance imaging (MRI) in pyogenic vertebral osteomyelitis. Fat-Sat: T1 or T2 sequence with saturation of fat signal. Vertebral osteomyelitis show intervertebral disc and adjacent vertebrae hyposignal on T1 sequence, contrast enhancement in T1 sequence with Gadolinium injection. Nucleus of intervertebral disc and adjacent vertebrae appear in hypersignal on T2 sequence. T4-T5 pyogenic vertebral osteomyelitis with disc disappearance, vertebral body destruction, kyphotic angulation, epidural inflammation (^∗), and spinal cord compression by an anterior epidural abscess (arrow).