Reassessing the Kocher Criteria: A Prospective Evaluation of Diagnostic Performance in Pediatric Musculoskeletal Infection

BACKGROUND: The Kocher criteria are widely used to differentiate septic arthritis (SA) from transient synovitis (TS) in children. However, diagnostic performance has since been proven to be inconsistent. Furthermore, the criteria were originally developed to distinguish SA from TS and have not been validated for differentiating osteomyelitis (OM), which frequently presents with overlapping clinical features. The purpose of this prospective study was to evaluate the diagnostic performance of the Kocher criteria and other clinical variables in differentiating SA from both TS and OM. METHODS: A prospective cohort study of 102 pediatric patients (ages 0 to 18 y) presenting with suspected musculoskeletal infection was performed. Patients were categorized into 3 groups: TS (n=42), OM (n=30), and SA (n=30). Clinical variables, laboratory biomarkers (C-reactive protein, erythrocyte sedimentation rate, white blood cell count), and Kocher criteria were analyzed. Clinical, laboratory, and Kocher variables were compared across groups using appropriate univariate analyses and linear regression modeling, with P<0.05 being considered significant. RESULTS: Patients with SA and OM demonstrated significantly elevated inflammatory markers compared with TS, including C-reactive protein and erythrocyte sedimentation rate (both P<0.001). White blood cell count demonstrated weaker discriminatory ability (P=0.041). Tenderness to palpation and erythema were significantly more common in SA and OM than in TS (both P<0.001). Among patients with SA and TS only, the probability of SA increased from 9.5% with 0 criteria to 65% with 2 criteria, then plateaued at 66.7% with 3 or 4 criteria. When OM was included in the analysis, substantial overlap in the Kocher score distribution was observed between SA and OM. Notably, among patients meeting two criteria, SA and OM each accounted for 39.4% of diagnoses. No significant difference in Kocher criteria burden was identified between SA and OM cohorts (P=0.874). Hip-specific subgroup analysis demonstrated improved discriminatory performance, with no cases of SA or OM identified among patients meeting zero criteria. CONCLUSIONS: The Kocher criteria demonstrated moderate utility in distinguishing SA from TS but poor specificity for differentiating SA from OM because of substantial overlap in clinical presentation and inflammatory markers. The historically reported predictive performance of the Kocher criteria was not reproducible in this prospective cohort. The identified diagnostic plateau at ≥2 criteria suggests these scores should serve as an indicator for advanced imaging (magnetic resonance imaging) rather than a definitive surgical algorithm. LEVEL OF EVIDENCE: Level III (prospective cohort study).