26 - Fresh look at an old score: differentiating between septic arthritis and transient synovitis of the hip in children with an irritable hip

A previously published and validated clinical prediction model (aka Kocher criteria) can be used to stratify the risk for septic hip among children presenting with irritable hip based on clinical and laboratory predictors. A modified version (aka Caired criteria) added CRP to prior clinical and laboratory predictors.


Objective: To assess the performance of this prediction model in stratifying the risk for septic hip.

Design/Methods:

This is a cross sectional retrospective review of healthy patients aged 1-18 years who presented to two tertiary pediatric emergency departments with hip pain and underwent hip arthrocentesis. We had 10 years of data from one facility (2010-2020) and four years of data from the second (2018-2022). We used the original study definition for septic joint: synovial fluid positive for a pathogen and synovial fluid white blood cell count >40k/mm3. We abstracted clinical and laboratory features including the proposed criteria in the Kocher and Caired clinical prediction rules (fever ≥ 38.5ºC, blood WBC ≥ 12,000, ESR ≥ 40 mm/hr, inability to bear weight and  CRP ≥ 2mg/dl). We performed a univariate analysis by Chi-squared test followed by a multivariate logistic regression model.


Results:

One hundred ten patients were identified, among which 45 met the definition of septic joint defined by Kocher et al. The performance of the clinical predictors in a univariate analysis (Table 1). Patients with septic arthritis did not differ from those with other arthritis/synovitis with regards to history of fever,  elevated WBC count, elevated ESR or inability to bear weight on both univariate analysis and multivariate analysis (Table 2).  A CRP ≥ 2mg/dl was the only predictor of septic arthritis with OR 10.6 (95% CI 3.8-29.2) in the univariate analysis and 15.8 (95% C.I. 4.2-59.1) in multivariate analysis.

Table 3 presents our result for the scoring criteria (1 point per positive variable) side by side with the results of the original study. Our results differed significantly with 5% of septic hip patients having none of the criteria present and nearly a third of the septic hip patients had a single positive criteria. 


Conclusion(s):

An elevated C-reactive protein level was the only independent predictor for septic arthritis in this patient population. The diagnostic performance of the Kocher and Caired criteria is reduced when applied to this new patient population. Septic arthritis cannot be excluded even when non of the criteria are present.