P118: The utility of serum markers for diagnosing septic arthritis in the emergency department: do rigid cut-offs improve diagnostic characteristics?

Introduction: Septic arthritis represents one of the most severe diagnoses for a presentation of an acutely swollen joint, with a high level of morbidity and mortality associated with delayed management. There is continued interest in the utility of serum markers of inflammation in diagnosing this dangerous condition, however there is a lack of clear consensus for cut-offs that optimize diagnostic performance for these tests. The objective of this study was to perform a systematic search of the literature to identify optimal cut-offs for commonly ordered serum markers and to assess how these cut-offs perform in a cohort of patients with a diagnosis of septic arthritis. Methods: We performed a systematic literature search aimed at identifying optimal cut-offs for serum makers (white blood cell count (WBC), erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP)) used for diagnosing septic arthritis. We assessed the use of these markers within a retrospective cohort (n=87) of patients diagnosed with septic arthritis (based on positive gram stain, culture, or treatment with a prolonged antibiotic course and/or surgical intervention) that presented to one of four emergency departments in Calgary over a two-year period. We then compared published values to local data. Results: We identified 10 articles that evaluated diagnostic characteristics for serum markers. Although there was variability for cut-offs reported in the literature, classically WBC >11 x 109/L, ESR >30 mm/h, and CRP >100 mg/L were reported to modestly increase the likelihood ratio of diagnosing septic arthritis. In our cohort, a complete blood count was ordered in the emergency department in 97% (n=84) subjects. ESR and CRP were ordered in 66% (n=57) and 85% (n=74) of patients, respectively. When comparing the classic literature based cut-offs to our population group, a WBC <11 x 109/L was found in 38% (n=32), ESR <30 mm/h in 51% (n=38), and CRP <100 mg/L in 30% (n=17). Sensitivity was found to be poor (61% for WBC >11 x 109/L; 70% for ESR >30 mm/h; 48% for CRP >100 mg/L). Conclusion: Data collected from the Calgary Emergency Department supports the published literature suggesting that serum tests are not helpful in the diagnosis of septic arthritis. Future work should evaluate these diagnostic characteristics in relation to patients with non-infectious monoarticular joint pain.