Vertebral osteomyelitis: Description of a series of 103 cases and identification of predictive variables of the etiological group

Abstract

Background. Vertebral osteomyelitis (VO) is an infrequent disease that requires early antibiotic treatment to prevent serious disabling sequels. The aim of our study was to identify clinical variables capable to guide the initial treatment. Patients and methods. Our retrospective study involved 103 patients seen during a 13 years period, mostly (94%) diagnosed by magnetic resonance imaging. Patients were classified as having brucellar (50 patients), pyogenic (29 patients), tuberculous (20 patients), and other (4 patients) spondylitis. Clinical and laboratory variables entered into a multivariate analysis to identify those independently associated with each etiologic subgroup. Results. Brucellar spondylitis was associated with epidemiological exposure (LR+: 14,9; LR-: 0,02), and with a blood neutrofil percentage ≤ 65% (LR+: 6,6; LR-: 0,12). The presence of an underlying chronic disease (LR+: 29,5; LR-: 0,01), an erythrocyte sedimentation rate ≥ 60 mm/h (LR+: 11,3; LR-: 0,07), and the absence of constitutional symptoms (LR+: 5,2; LR-: 0,18), were independent predictors of a pyogenic spondylitis. Finally, previous exposure or tuberculous disease (LR+: 11,2; LR-: 0,06), and a delay of ≥ 8 weeks in first consulting (LR+: 10,6; LR-: 0,07) were associated with tuberculous spondylitis. Conclusion. The combination of a few clinical and laboratory variables facilitates the differential diagnosis between pyogenic, brucelar and tuberculous vertebral osteomyelitis, allowing the choice of the initial antibiotic treatment.