Differentiating Between Knee Septic Arthritis and Lyme Arthritis in Children: A Clinical Prediction Algorithm for a Geographically Diverse Population
Ying Li,Maanasa Bommineni,Keith D Baldwin,Ryan M Sanborn,Danielle Cook,Benjamin J Shore,Children’s Orthopaedic Trauma and Infection Consortium for Evidence-Based Studies (CORTICES) Study Group,Allan C Beebe,Todd J Blumberg,Lawson A Copley,Jaime R Denning,Rachel Y Goldstein,Benton E Heyworth,Jaclyn F Hill,Megan E Johnson,Jennifer C Laine,Collin J May,Mark L Miller,Stephanie N Moore-Lotridge,Joshua S Murphy,Brandon A Ramo,Anthony I Riccio,Scott B Rosenfeld,Julia Sanders,Jonathan G Schoenecker,David D Spence,Walter H Truong,Vidyadhar Upasani
DOI: https://doi.org/10.1097/BPO.0000000000002814
2024-09-06
Abstract:Background: Knee septic arthritis (SA) and Lyme arthritis (LA) often have similar presentations but bacterial SA necessitates urgent surgery. Predictive factors for differentiating SA and other infectious/inflammatory conditions have been published. Our purpose was to test these algorithms using a retrospective multicenter musculoskeletal infection database. Methods: Patients ≤18 years old with isolated knee SA or LA were identified. Diagnostic criteria for SA were synovial WBC count >50,000 cells/mm3, imaging with fluid aspiration suggestive of SA, or joint aspirate/tissue sample cultured positive for bacteria. Diagnostic criteria for LA was positive Lyme titer. Demographics, weightbearing status, admission vitals, and laboratory tests were collected. Predictive factors from Baldwin criteria for differentiating knee SA and LA, and Kocher criteria for differentiating hip SA and transient synovitis were tested. Results: One hundred fifty-five patients (119 SA and 36 LA) were analyzed. Patients with SA were younger (2.2 vs. 8.0 y), more nonweightbearing (74% vs. 33%), had a higher pulse (127 vs. 106), and higher WBC (12.4 vs. 10.2) (all P<0.001).Baldwin criteria (pain with joint motion, history of fever, CRP >40 mg/L, age <2 y) were tested. Pain with motion was not collected in our database. Of the remaining factors, the probability of SA was 63% with 0 and 92% with 3 factors (AUC 0.64). Kocher criteria (nonweightbearing, temperature >101.3°F, WBC >12.0, ESR >40) and CRP >20 mg/L were also tested. The probability of SA was 41% with 0 and 96% with all factors (AUC 0.69).Using our cohort data, regression analysis with backward stepwise elimination determined that age <4 years, nonweightbearing, admission WBC >13.0, platelets <325, and ESR >70 were predictive factors for SA. The probability of SA with 0 factors was 16%, 1 factor 52%, 2 factors 86%, 3 factors 97%, and 4 factors 100% (AUC 0.86). Conclusions: Our model identified age <4 years, nonweightbearing, admission WBC >13.0, platelets <325, and ESR >70 as independent predictive factors for knee SA. The more factors present, the higher the likelihood of having SA versus LA. Level of evidence: Diagnostic level III.