Investigators from multiple institutions conducted a case-control study to identify clinical and treatment features associated with antibiotic-refractory Lyme arthritis (ARLA) in children. Study participants were identified by reviewing medical records of patients <18 years old with a diagnosis of Lyme arthritis (LA) seen at 1 of 4 pediatric rheumatology clinics in a Lyme-endemic region between 2000 and 2013. Only children with a clinical diagnosis confirmed by Western blot testing and documented arthritis were eligible. Cases were patients with active LA despite ≥8 weeks of oral antibiotics (amoxicillin, doxycycline, or cefuroxime) or ≥2 weeks of IV antibiotics (ceftriaxone or cefotaxime). Controls were children with documented resolution of LA within 3 months of starting antibiotics; LA was considered resolved when there was no joint pain or stiffness and minimal or no effusion on physical examination. The medical records of study participants were reviewed, and demographic and clinical information were abstracted. Using multivariate logistic regression, independent predictors of ARLA were identified. For the primary analysis, 26 predictor variables selected before the analysis were included. A secondary analysis, including variables selected post hoc, was also conducted.

Data on 49 children meeting criteria for ARLA (cases) and 188 children whose LA resolved within 3 months (controls) were analyzed. In the primary analysis, children with ARLA were significantly more likely to be ≥10 years old than controls (odds ratio [OR], 2.5; 95% CI, 1.1–5.6) and significantly more likely to have arthritis limited to one or both knees (OR, 5.1; 95% CI, 1.4–19.2) and joint symptoms for ≥6 weeks (OR, 9.4; 95% CI, 2.5–34.7) at diagnosis. Cases were also more likely than controls to have clinical worsening after initial antibiotic treatment (OR, 4.2; 95% CI, 1.4–12.6). Conversely, signs of severe inflammation (fever, severe pain, and erythrocyte sedimentation rate ≥40 mm/h) were associated with a significantly reduced risk of developing ARLA (OR, 0.4; 95% CI, 0.2–0.9). In the secondary analysis, additional independent predictors of ARLA were use of amoxicillin at doses less than those recommended by LA treatment guidelines and antibiotic non-adherence (defined as the patient receiving <80% of prescribed doses).

The authors conclude that children with LA who present with prolonged joint symptoms, arthritis limited to the knees, and/or who fail to respond to the initial course of antibiotics are at risk for ARLA.