Comparison of pyogenic postoperative and native vertebral osteomyelitis

Background context: Postoperative vertebral osteomyelitis (PVO) after spinal surgery is a clinical challenge. However, there is a paucity of evidence regarding the most likely etiologic organisms to guide the choice of empirical antibiotic therapy, and previous reports of treatment outcomes for PVO are scarce. Purpose: To compare the microbiology, clinical characteristics, and outcomes of pyogenic PVO with native vertebral osteomyelitis (NVO). Study design: Retrospective comparative study. Patient sample: Patients with microbiologically proven vertebral osteomyelitis from three university-affiliated hospitals in South Korea between January 2005 and December 2015 with follow-up of at least 12 months after completion of antibiotics or until the patient was transferred. Patients who had a spine operation in the same location within 1 year of diagnosis, and all patients with remnant implants at the time of the vertebral osteomyelitis diagnosis, were defined as having PVO. The remainder of the patients was considered to have NVO. Spinal operations included discectomy, laminectomy, arthrodesis, and instrumentation for stabilization of the spine. Outcome measures: Overall mortality, neurologic outcomes, treatment failure, and relapse of infection. Methods: Demographic data, comorbidities, presenting symptoms, microbiological data, radiographic characteristics, laboratory data (including white blood cell counts, erythrocyte sedimentation rate, and C-reactive protein), surgical treatment, and neurologic outcomes for each patient were reviewed from electronic medical records and analyzed. Mortality rate, treatment failure, and relapse of infection were calculated for the two groups. Factors associated with treatment outcome were evaluated using univariate and multivariate logistic regression analyses. Results: The study evaluated 104 patients with PVO and 441 patients with NVO. In PVO, the most common isolate was Staphylococcus aureus (34%, n=35), followed by coagulase-negative staphylococci (31%, n=32). In NVO, the most common isolates were S. aureus (47%, n=206) and streptococci (21%, n=94). Of the staphylococci, the proportion of methicillin-resistant strains was significantly higher in PVO than that in NVO (75% vs. 39%, p<.001). The proportion of patients with gram-negative bacilli was 14% in PVO and 20% in NVO. Pre-existing or synchronous nonspinal infection was observed more frequently in NVO than in PVO (33% vs. 13%, p<.001). Although the duration of antibiotic use was similar in both groups, surgery for infection control was performed more frequently in PVO. The mortality rate was similar in both groups. However, the treatment failure and relapse rates at 12 months were higher in the PVO group (23% vs. 13%, p=.009; 14% vs. 7%, p=.028, respectively). Methicillin-resistant S.aureus was significantly associated with treatment failure or relapse via logistic regression (odds ratio 3.01, 95% confidence interval [1.71–5.32], p<.001; odds ratio 2.78, 95% confidence interval [1.40–5.49], p=.003). Conclusions: Coverage of methicillin-resistant staphylococci should be considered when prescribing empirical antibiotics for PVO. Although surgery was performed more often in PVO than NVO, the treatment failure and relapse rates at 12 months were higher in PVO.