Test characteristics of clinical findings and clinical decision rules for the diagnosis of septic arthritis in children with an acute limp presenting to the emergency department: a prospective observational study.

Abstract

Background: Acute limb pain in young children presenting to the emergency department (ED) can be a sign of serious pathology, with septic arthritis being the most important condition to rule out. Two clinical decision rules (CDRs) have been developed to assist with the diagnosis: Kocher's rule (which allocates points for fever, weight-bearing status, white cell count and erythrocyte sedimentation rate) and Caird's rule (which also includes C-reactive protein). This study aimed to determine (1) the diagnostic accuracy of the two rules for the identification of septic arthritis and (2) other clinical features most strongly associated with septic arthritis.

Methods: Prospective observational study of consecutive children aged <16 years with non-traumatic acute limp presenting to three EDs in Melbourne, Australia between July 2016 and September 2018. Data were prospectively collected on weight-bearing status, duration of symptoms, joint examination findings and signs of systemic disease. Structured chart reviews and telephone follow-up were used to adjudicate the presence/absence of septic arthritis. Area under the receiver operating characteristics curve (AUC) was calculated for each published CDR, and sensitivity, specificity and likelihood ratios were calculated for clinical findings.

Results: Of 583 patients presenting with atraumatic limp, 535 (91.8%) eligible patients had sufficient follow-up data. 14 (2.6%) were diagnosed with septic arthritis. Kocher's rule had an AUC of 0.72 (95% CI 0.42 to 1.00), while Caird's rule had an AUC of 0.78 (95% CI 0.52 to 1.00) for septic arthritis. Univariable analysis demonstrated strong associations between range of joint motion (unadjusted OR 13.9, 95% CI 5.0 to 38.5), signs of systemic disease (OR 20.5, 95% CI 6.2 to 67.7), hip pain (OR 3.8, 95% CI 1.2 to 11.7) and presence of fever (OR 5.1, 95% CI 1.0 to 25.1) with septic arthritis. Markedly reduced range of motion compared with the unaffected side had the highest positive likelihood ratio (12.1, 95% CI: 7.5 to 19.5), while inability to weight bear had a positive likelihood ratio of 3.85 (95% CI 2.49 to 5.95). None of the tested clinical findings had a negative likelihood ratio less than 0.3, or a positive predictive value of more than 25%.

Conclusion: Septic arthritis is a relatively uncommon diagnosis in children presenting to the ED with an acute limp. Markedly reduced range of motion and inability to weight bear appear to be the strongest predictors of septic arthritis; however, their absence is insufficient to rule out the diagnosis.