C-reactive protein of ≥ 20 mg/L and ultrasound finding of an effusion ≥ 7 mm has a high specificity and sensitivity in diagnosing paediatric hip septic arthritis.

Purpose: Differentiating septic arthritis (SA) from transient synovitis (TS) in children remains a diagnostic challenge. Several algorithms have been developed to diagnose SA including Kocher's criteria and its subsequent modifications, but reports show variable efficacy. This study aims to examine the diagnostic utility of a novel method only using C-reactive protein (CRP) and ultrasound (US) findings of effusion in differentiating SA from TS, determine the optimal values for these predictors and validate this method against existing clinical predictors.

Methods: A 5-year retrospective study was performed including all paediatric patients with acute, non-traumatic hip pain with a suspicion of SA. All patients were evaluated using Kocher's criteria, Caird's criteria, and the novel method. Multivariate logistic regression was performed to identify independent clinical predictors of SA. The degree of agreement between the various methods were assessed using Cohen's kappa (k). Receiver operating characteristics (ROC) curves were used to examine the diagnostic accuracy of this novel method as well as to determine optimal cut-offs for US effusion and CRP in diagnosing SA.

Results: Hundred and one patients were recruited. CRP and effusion on US were found to be independent predictors of SA. Both Kocher's and Caird's method showed good specificity (98.9%) but extremely poor sensitivity for SA (0%). When Kocher's four clinical predictors were present, probability of SA was only 59.16%. The k for both Kocher's and Caird's methods, was -0.017 indicating poor agreement. However the k in the novel method was 0.641, indicating good agreement.

Conclusion: Our study showed that the novel method using CRP (≥ 20 mg/L) and US finding of effusion (≥ 7 mm) has a high specificity (97%) and sensitivity (71%) in diagnosing SA.