Objectives To study the sensibility (Se) and specificity (Sp) of clinical and biological signs for the diagnosis of septic arthritis (SA).
Methods This prospective study included all adult patients with a suspicion of SA seen in the emergency department or rheumatology service at the University Hospital of Clermont-Ferrand during a period of 18 months. The sensibility and specificity of the clinical and biological signs for the diagnosis of SA were calculated. A univariate analysis followed by a logistic regression analysis for SA diagnosis was conducted.
Results In total, 105 patients with a suspected SA were included, 38 (36%) presenting with SA (29 with a bacteriologically documented SA). In univariate analysis, chills (Se 39%; Sp 82%, p=0.015) but not fever (Se 53%; Sp 53%, p=0.6), a slow onset duration (Se 54%; Sp 72%, p=0.04), local redness (Se 53%, Sp 72%, p=0.01), as well as an infection entry-site (Se 7%; Sp 54%, p=0.01) were found most often in the SA cases. Microcrystalline antecedents (Se 5%, Sp 72%, 95% CI 0-0.6) and involvement of the knee were more frequent in non-septic arthritis (non-SA) cases. An erythrocyte sedimentation rate (ESR) >50 mm (Se 72%; Sp 60%, p=0.005), a C-reactive protein level >100mg/L (Se 58%; Sp 66%, p=0.019) and, above all, suggestive radiological signs (Se 30%; Sp 95%, p=0.001) were more frequent in the SA cases. The clear (Se 0; Sp 81%, p=0.007) and purulent (Se 57%, Sp 88, p<0.001) aspect of synovial fluid allowed for the differentiation between SA and non-SA, as well as the synovial fluid white blood cell (WBC)>50000/μL (Se 57%; Sp 82%, p<0.001) and percentage of polymorphonuclear >90% (Se 42%; Sp 82%, p=0.02), but not the presence of microcrystals (Se 24%, Sp 58% p=0.09). In multivariate analysis, only chills (OR=4.1, 95% CI: 1.4-12.2), microcrystalline antecedents (OR=0.08, 95% CI: 0.01-0.5), and radiological signs (OR=11, 95% CI: 2.4-50.1) remained significant. The parameters chills, microcrystalline antecedents, infection entry-site, risk factors for SA, the aspect and cellularity of synovial fluid, and radiological signs permitted the elaboration of two logistic regression models for the diagnosis of SA (AUC: 0.85 and 0.87).
Conclusions No clinical or biological (excluding bacteriological) sign, taken alone, is conclusive for the differentiation between SA and non-SA pathology, but the association of several signs, notably chills, microcrystalline antecedents, radiological signs of SA, and the aspect and cellularity of joint liquid may be conclusive.
Disclosure of Interest None Declared